3D Spatiotemporal Intelligence focuses on observing, representing, and understanding how real-world 3D geospatial environments evolve over time. It uses multi-temporal or continuous observations to characterize the timing, rates, and sequences of changes in 3D geometry, physical and semantic properties, and spatial relationships, and to support predictions of future states. We invite contributions that integrate multi-source remote sensing, geospatial analysis, and artificial intelligence to study the interplay among spatial structures, time-varying states, and geospatial relationships. The theme scope includes the use of LiDAR, photogrammetry, multi-view optical remote sensing, and coordinated observations across satellite, airborne, UAV, mobile, and other platforms. Applications include urban spatial evolution, forest structure and ecosystem monitoring, land-surface deformation, and natural hazard assessment. Contributions may address unified spatiotemporal representations, structural and organizational dynamics, semantic scene understanding, relational reasoning, and predictive modeling. The common scientific focus is to connect 3D spatial structure with environmental dynamics and their geospatial context.

NASA continues to mature the science and technology towards a Surface Topography and Vegetation (STV) observing system to measure global meter-level bare Earth topography and vegetation structure. Envisioned as a disaggregated set of airborne, high altitude, and spaceborne platforms hosting radar, lidar, and stereoimaging instruments, the mission will provide baseline and change measurements. In June 2026 the team flew an airborne campaign called ASCENT (Airborne Surface, Cryosphere, Ecosystem, and Nearshore Topography) hosting UAVSAR radar and SAR-Fusion stereo instrument on NASA’s G-III and LVIS lidar and QUAKES stereoimager on NASA’s G-V. The campaign produced surrogate data sampled from a variety of targets and configurations to develop algorithms to separate vegetation and shallow water from bare Earth topography and fuse the data to mature STV. The mission is well suited to monitor and characterize a variety of hazards and hazard cascades such as earthquakes, floods, volcanoes, landslides, wildfires, and glacial lake outburst floods. This session will focus on STV approaches to studying the geosphere, hydrosphere, biosphere, and cryosphere and associated hazards. This includes technology maturation of radar, lidar, and stereoimaging and techniques to process and fuse data from these three instrument types. 

Rapid advances in artificial intelligence (AI), machine learning, seismic sensing technologies, and high-performance computing are creating new opportunities for seismic data analysis and subsurface imaging. Modern seismic observations are becoming increasingly large, dense, and heterogeneous, ranging from conventional active-source and passive seismic surveys to large-N seismic arrays and distributed acoustic sensing (DAS). These developments call for new AI approaches that can extract information from massive seismic datasets, improve seismic processing and imaging, and integrate complementary observations to better characterize subsurface structures and dynamic Earth processes.
This Community Contributed Theme will bring together researchers developing advanced AI approaches for seismic data analysis, processing, imaging, inversion, and monitoring. Particular emphasis will be placed on emerging AI methodologies that address fundamental challenges in seismic processing and imaging, including noise attenuation, signal reconstruction, seismic event detection, wavefield characterization, velocity-model building, tomography, migration, and full-waveform inversion. DAS will be an important component of the theme because of its rapidly expanding role in dense seismic acquisition, monitoring, and imaging.
The theme will encompass a broad range of modern AI methodologies, including physics-informed and model-driven machine learning, deep neural networks, generative and diffusion models, foundation models, implicit neural representations, self-supervised learning, transfer learning, uncertainty-aware learning, and differentiable physics. These methods offer new opportunities to combine data-driven representations with seismic wave physics and established processing and imaging frameworks.
The theme primarily focuses on seismic data analysis and imaging. InSAR, GNSS, and other remote-sensing observations are included in scope when integrated with seismic observations or seismic imaging workflows. Such complementary measurements can provide additional constraints on surface deformation and Earth processes and can help improve, validate, or interpret seismic monitoring and subsurface imaging. Examples include joint analysis of seismicity and InSAR deformation, integrating GNSS displacement measurements with seismic observations, and AI-enabled fusion of seismic and remote-sensing data to study earthquakes, induced seismicity, geothermal and volcanic systems, fault-zone processes, fluid injection and extraction, and other dynamic subsurface phenomena.

Topics of interest include, but are not limited to:
•	AI/ML for seismic data processing, denoising, reconstruction, interpolation, and enhancement;
•	AI-assisted seismic imaging, tomography, migration, and full-waveform inversion;
•	physics-informed, physics-guided, and hybrid data/model-driven approaches for seismic inverse problems;
•	differentiable seismic modeling and end-to-end learning for seismic processing and imaging;
•	AI for DAS processing, denoising, event detection, monitoring, and subsurface imaging;
•	deep learning and AI for earthquake detection, phase picking, characterization, and seismic monitoring;
•	generative and diffusion models for seismic reconstruction, processing, and inverse problems;
•	implicit neural representations for seismic wavefields, Earth models, and seismic inversion;
•	foundation models, self-supervised learning, transfer learning, and representation learning for seismic data;
•	uncertainty quantification, interpretable AI, robustness, and generalization in seismic applications;
•	AI-enabled real-time analysis of large-scale seismic and DAS datasets;
•	AI-assisted integration of seismic data with InSAR, GNSS, and other remote-sensing observations;
•	joint seismic and remote-sensing analysis for understanding subsurface deformation and dynamic Earth processes;
•	multimodal AI approaches in which remote-sensing observations provide complementary constraints for seismic monitoring, processing, imaging, or interpretation.

The rapidly evolving field of geoscience and remote sensing is at the forefront of innovations driving a global vision for a changing planet. This transformation is fueled by the rapid development of advanced signal processing techniques, enabling deeper insights from the growing volume, variety, modality, and velocity of multimodal geospatial data. Innovative algorithms are increasingly critical for efficiently extracting meaningful and actionable information, supporting both scientific discovery and operational applications.
This session focuses on state-of-the-art signal processing methodologies that enhance geospatial data analysis, interpretation, and utilization. We particularly welcome contributions that address emerging challenges and opportunities in monitoring our changing planet through remote sensing, multimodal and multisensor data fusion, and advanced computational frameworks. Topics of interest include, but are not limited to:
(1) Innovative Algorithms and Advanced Analysis:  novel algorithms for geospatial data analysis, interpretation, advanced sampling, sparse representation, and dictionary learning.
(2) Compressive Sensing and Sparse Reconstruction:  CS methods for geoscience and remote sensing, including radar, subsurface imaging, hyperspectral, SAR, and geophysical signals.
(3) Super-Resolution Techniques: deep learning-based super-resolution, resolution enhancement for optical and radar imagery, and applications in land cover and environmental monitoring.
(4) Emerging and Complex Signal Processing: complex and hyper-complex signal/image processing, and innovative frameworks with promising applications.
(5) Time-Series and Real-Time Processing: techniques for high-velocity, dynamic Earth observation data streams.
(6) Neural Networks and Deep Learning: CNNs, RNNs, GANs, complex-valued and quaternion neural networks, and other DL models for remote sensing analysis.
(7) Multimodal and Multisensor Data/Image Fusion: integration of heterogeneous sensors and modalities for enhanced Earth observation.
(8) Applications in Earth System Domains/Changing Planet: environmental monitoring, disaster management, climate studies, and other applications enabled by advanced signal processing.

Seasonal snow plays a critical role in the Earth's climate system and water cycle. However, retrieving snow properties from satellite observations remains particularly difficult in complex terrain, limiting confidence in derived snow products.
Recent advances in satellite technology, retrieval algorithms, and multi-sensor approaches are improving our ability to estimate key snow variables across diverse mountain and polar environments. New opportunities are emerging from optical, passive microwave, synthetic aperture radar (SAR), and combined satellite observations, enabling improved retrievals of snow cover, snow water equivalent (SWE), and related snow properties. Several upcoming satellite missions will extend these observational capabilities and will greatly benefit from advanced research and development on retrieval methods, helping to ensure that their measurements can be translated into robust and scientifically useful snow products.

This session invites contributions focused on the development, evaluation, validation, and intercomparison of snow retrieval techniques using satellite Earth observation. Topics include:
•	Snow cover retrievals
•	Snow water equivalent (SWE) retrievals
•	Retrieval of snow physical properties and snow albedo
•	Passive microwave snow retrieval algorithms
•	SAR-based snow retrieval techniques
•	Multi-sensor snow retrieval approaches
•	Validation, uncertainty characterisation, and algorithm intercomparison
•	Snow retrievals from current and future Earth observation missions
The session will provide a dedicated forum for researchers advancing the state of the art in satellite snow retrieval methodologies for mountain and polar regions.

The session sits within C.1: Snow Cover but focuses specifically on new satellite methods for retrieving snow properties in mountain and polar regions. It brings together work across sensors, regions, validation, and uncertainty that may otherwise be spread across the wider snow theme. It also supports SP.1: Geoscience and Remote Sensing in the Arctic through its focus on polar snow.

Frozen ground and seasonal freeze–thaw (FT) processes play a fundamental role in regulating terrestrial hydrology, ecosystem functioning, carbon exchanges, and land–atmosphere interactions across cold regions. As climate warming alters the timing, duration, and intensity of soil freezing and thawing, accurate monitoring of these processes is increasingly important for understanding environmental change and improving Earth system predictions. Microwave remote sensing provides unique capabilities for monitoring frozen ground because of its sensitivity to changes in soil dielectric properties and its ability to observe the land surface under a wide range of weather and illumination conditions. However, conventional satellite FT products are predominantly based on binary frozen/thawed classifications, which cannot fully capture gradual transitions, vertical FT dynamics, and the strong spatial heterogeneity of freezing conditions across complex landscapes.
This Community-Contributed Theme aims to bring together researchers working on emerging active and passive microwave remote sensing techniques to advance the monitoring and characterization of frozen ground and FT processes across Arctic, boreal, temperate, mountainous, and agricultural environments. Contributions are encouraged on innovative FT retrieval algorithms, non-binary and probabilistic FT characterization, multi-frequency microwave observations, and the effects of snow, vegetation, soil moisture, and landscape heterogeneity on microwave shoulder season signatures. The theme will also welcome studies exploiting recent and upcoming satellite missions, including NISAR, CIMR, and emerging high-resolution L-band radiometry concepts, as well as approaches integrating microwave observations with in situ measurements, physical models, and machine learning.

This proposed theme complements existing IGARSS General Themes, particularly L.8 Soils and Soil Moisture, by providing a dedicated interdisciplinary forum focused on the physical processes and remote sensing methodologies associated with soil freezing and thawing. While these themes address individual components of the problem, no existing theme specifically integrates active and passive microwave FT monitoring, retrieval development, and applications across multiple ecosystems and spatial scales. The session also complements SP.1 Geoscience and Remote Sensing in the Arctic by extending the discussion beyond Arctic environments to seasonal frozen ground globally. By connecting advances in microwave observations with hydrological, ecological, and climate modelling applications, this theme will foster interdisciplinary collaboration and support the development of the next generation of frozen-ground monitoring products.

Advances in radar systems, signal processing, data assimilation, and radar techniques are enabling unprecedented geophysical observations and discoveries in environmental and planetary science. Novel sensing modalities, including distributed and passive sensing, sensor fusion, and radar tomography, are offering improved spatial coverage, spatial resolution, temporal resolution, estimation accuracy, and 4D imaging capabilities. Reduced cost and increased availability of unmanned platforms and software-defined radios have opened avenues for novel radar applications and sensor systems. Innovations across discipline boundaries are being leveraged in conjunction with radar to address new applications in terrestrial and space sciences. Overall, emerging radar technology is transforming observational capabilities.  

This session aims to bring together pioneering research in radar systems and techniques. We invite researchers to present results on any topic involving the development of radar systems and data processing techniques to understand the terrestrial and space environment. This includes but is not limited to new results from existing radar observations, the development and application of novel observation or data processing techniques, numerical modeling of existing or future observations, and/or the analysis of ground-based, airborne, or orbital radar data for Earth and planetary remote sensing. 
The creation of this session is intended to establish a community for emerging radar technology and bring together researchers of this topic who have been presenting in disparate sessions. Combining knowledge will help push forward the critical and dynamic intersection of developing radar systems and techniques for monitoring the terrestrial and space environment. This session has the potential to advance knowledge across a range of science and engineering fields, benefiting glaciologists, climate scientists, geophysicists, planetary scientists, and all sectors of geoscience that rely on radar remote sensing of the environment for societal well-being.

Artificial intelligence is rapidly transforming Earth Observation (EO), with foundation models, multimodal learning, and advanced retrieval approaches enabling increasingly general representations of the Earth system. Agentic AI represents a further step toward systems that can reason, plan, interact with tools and models, coordinate multiple capabilities, and autonomously execute complex EO and Earth science workflows.

This session will explore emerging approaches for Agentic AI in Earth Observation, spanning the full chain from data acquisition to information and scientific knowledge generation. Topics include AI agents interacting with EO archives, foundation models and geospatial embeddings; multimodal reasoning across EO and non-EO information; agentic orchestration of processing chains and analytical tools; multi-agent systems; semantic interoperability and machine-to-machine communication; autonomous scientific hypothesis generation and testing; and agents supporting dynamic tasking of satellite and ground-based resources. Particular attention will be given to trustworthy autonomy, validation, provenance, uncertainty, benchmarks, and human-agent interaction.

This session would also provide a follow-up to the Agentic AI for Earth Observation workshop organized in Berlin in October 2026, which gathered more than 500 participants, demonstrating the strong and rapidly growing interest of the EO community in this emerging field. The IGARSS session would build on the discussions and research directions initiated in Berlin, providing a forum to consolidate the community, present new scientific results, and advance the development and evaluation of agentic approaches for EO.

Scope:
This session focuses on the emerging convergence of agentic AI, multimodal foundation models, and geospatial embeddings for Arctic and rapidly changing environments. Topics include: autonomous agents that plan and execute multi-step remote sensing workflows (e.g., task decomposition, tool use, retrieval-augmented reasoning); vision-language and multimodal foundation models for heterogeneous Arctic data (optical, SAR, hyperspectral, LiDAR, passive microwave); geospatial embeddings for cross-modal retrieval and semantic reasoning; benchmarks and evaluation protocols for agentic and foundation-model systems in data-scarce polar settings; uncertainty quantification, explainability, and trustworthy autonomy in high-stakes climate monitoring. Applications include sea-ice and glacier monitoring, permafrost degradation, coastal erosion, ecosystem transition, and hazard response in the Arctic and other rapidly changing regions.

Importance:
The Arctic is warming at nearly four times the global average, driving sea-ice retreat, permafrost thaw, and cascading global climate impacts. Monitoring these changes demands not only powerful models but also autonomous systems capable of orchestrating complex, multi-source observation workflows under data scarcity and harsh conditions. Agentic AI, which enables models to plan, reason, and take action via tool use, together with multimodal foundation models and geospatial embeddings, opens a transformative path toward scalable, semantically rich and adaptive Earth observation workflows. Yet deploying such autonomous systems in climate-critical and culturally sensitive regions raises urgent questions of trust, accountability, and evaluation rigor. This session addresses that dual frontier: advancing capability while ensuring responsible, benchmarked autonomy.

Rationale for Complementing General and Special Themes:
This proposed theme sits at the intersection of several IGARSS 2027 priorities. It directly aligns with SP.3: Agentic AI for Earth Observation and SP.5: Foundation Models and Embeddings for Earth Observation and is grounded in the application context of SP.1: Geoscience and Remote Sensing in the Arctic and SP.2: Monitoring Rapidly Changing and Hazardous Environments. It further complements SP.6: Datasets, Benchmarks, and Evaluation Methodologies by emphasizing rigorous evaluation of agentic and foundation-model systems. On the General Theme side, it relates to A.2: Foundation Models in Geoscience, T.18: Multimodal Data Fusion, C: Cryosphere (C.1–C.4), and D.10: Remote Sensing for Climate Change Impacts. While foundation models and multimodal fusion are increasingly recognized, their integration with agentic autonomy and their rigorous benchmarking in polar, data-scarce settings remains underexplored. This session fills that gap.

Space missions rely on remote sensing to characterize planetary environments, identify scientific targets, and support operational decisions. AI and robotics can bring these functions closer together, allowing spacecraft and surface platforms to use observations when planning movements, selecting measurements, and responding to changing conditions. This Community Contributed Theme focuses on the use of AI in planetary remote sensing, robotic exploration, and autonomous mission operations. Operating beyond Earth introduces constraints that affect both scientific observations and robotic activities. Communication delays limit direct supervision, while available power, computing capacity, and transmission bandwidth restrict onboard processing and data return. Unfamiliar terrain and limited training data indeed create further difficulties. This session will examine how AI methods can address these conditions, how their performance can be assessed, and what is required to integrate them into mission systems.

Contributions are welcome on the following topics (but they are not limited to them):
•	Planetary image analysis and mapping: terrain reconstruction, geological interpretation, surface classification, and hazard detection using orbital or surface observations.
•	Robotic navigation and exploration: localization, traversability estimation, route planning, obstacle avoidance, and site selection informed by remote sensing data.
•	Autonomous science and sampling: identification of scientific targets, robotic sample selection, active perception, and adaptation of measurements to observations acquired during a mission.
•	Onboard processing: efficient AI models, hardware-aware implementation, data filtering, compression, and selection of observations for transmission under resource constraints.
•	Sensor fusion and coordinated observations: combining imagery, spectroscopy, radar, lidar, and measurements from orbiters, landers, rovers, and cooperating robotic platforms.
•	Learning across environments: foundation models, transfer learning, adaptation to planetary data, and methods for handling limited labels and differences between training and deployment conditions.
•	Reliability and supervision: uncertainty estimation, failure detection, verification, and operator interaction in autonomous sensing, planning, and control.
•	Testing and evaluation: simulation, digital twins, analogue field experiments, datasets, and benchmarks linking algorithm performance to scientific results and operational requirements.
The proposed theme complements the IGARSS 2027 General Themes in planetary remote sensing, AI and Big Data, Theory and Techniques, and Sensors. It brings these areas together around a specific question: how can information extracted from observations guide the next measurement or robotic action? This focus connects work on individual processing methods with the requirements of complete exploration systems.
The relationship with the Special Themes is strongest for SP.3, Agentic AI for Earth Observation; SP.5, Foundation Models and Embeddings for Earth Observation; and SP.6, Datasets, Benchmarks, and Evaluation Methodologies for Earth Observation. The proposed theme extends these discussions to planetary missions and robotic systems, where decisions affect platform motion, instrument use, and scientific sampling. Its contribution is a focus on the interaction between sensing and physical operations, including the constraints and evaluation needs that arise from that interaction.
The session will bring together researchers in remote sensing, robotics, AI, planetary science, and mission engineering. Methodological studies, mission applications, and experimental results are welcome. Discussion will address practical limitations, lessons from deployment and testing, and methods that can be shared between planetary exploration and terrestrial field robotics.

Forest remote sensing has largely operated at the pixel or stand level, mapping forest cover, canopy height, and biomass. Forest management, carbon accounting, and wildfire and pest risk assessment, however, increasingly require information about individual trees: where each tree stands, its species, size, and health, and how it changes over time. Sub-meter satellite imagery, airborne and UAV LiDAR, spaceborne LiDAR, and foundation models now make tree-level mapping feasible over large areas. This session focuses on AI methods that detect, characterize, and track individual trees in forests, woodlands, plantations, agroforestry systems, and trees outside forests, and that connect tree-level products to forest inventory and management. Topics include individual tree detection and crown delineation from very-high-resolution imagery and LiDAR; tree-level species classification and traits from multispectral, hyperspectral, and LiDAR data; estimation of height, crown size, stem diameter, and biomass, including allometry and upscaling from crowns to stands; early detection of drought stress, insect and disease outbreaks, windthrow, and fire damage in single trees; multi-temporal tracking of growth, harvest, mortality, and regeneration; and linking maps to management through national forest inventories, field plots, forest digital twins, uncertainty quantification, and label-efficient or foundation models that transfer across biomes and sensors.

Forests store a large share of terrestrial carbon and face rising pressure from drought, fire, and pests. Decisions such as selective harvesting, sanitation felling, carbon credit verification, and compliance with deforestation-free supply chain regulations depend on individual trees, whereas national forest inventories sample only a small number of plots and are updated every few years. Pixel-level maps blur mixed stands and sparse woodlands and often miss trees outside forests, which hold substantial carbon. Recent studies have shown that tree-level mapping across countries and continents is feasible, but accuracy varies widely across biomes, species classification remains difficult, and validation practices are inconsistent. In Iceland, where afforestation and the restoration of native birch woodland are under way, young and scattered trees invisible to coarse maps will shape the forests of the future. In line with the theme Global Vision for a Changing Planet, the session brings together forest scientists, LiDAR and image analysis specialists, and AI researchers.

The session complements the existing General and Special Themes. Whereas L.3 (Forest and Vegetation: Application and Modelling) and L.4 (Forest and Vegetation: Biomass and Carbon Cycle) address forests at pixel, stand, and regional scales, contributions here must operate at the individual-tree level or demonstrate how tree-level information improves stand and regional products. Generic techniques from T.11 (Object Detection and Recognition) and T.10 (3D Mapping) are evaluated here against forestry ground truth such as inventory plots and stem maps. The session connects to A.6 (Digital Twins) through tree-based forest digital twins and to S.5 (Spaceborne LiDAR Missions) as a key data source for upscaling. It also complements SP.2 (Monitoring Rapidly Changing and Hazardous Environments with Remote Sensing) by resolving fire, drought, and pest damage at the single-tree level, and draws case studies from SP.1 (Geoscience and Remote Sensing in the Arctic), such as tree line advance in the boreal zone.

Advances in artificial intelligence (AI), remote sensing, and edge computing are enabling satellites to transform observations into actionable information directly in orbit. Processing multispectral, hyperspectral, and synthetic aperture radar data onboard can reduce downlink requirements and shorten the time between observation and response. These capabilities support environmental monitoring, precision agriculture, object detection and tracking, soil assessment, and disaster management, where timely information can substantially improve scientific, operational, and societal outcomes.

Realizing these benefits requires AI methods designed for the constraints of space-based systems. Satellite platforms impose strict limits on computation, memory, energy, and communications. Models must operate reliably despite noisy observations, variable image quality, and shifts across sensors, locations, and acquisition conditions. Representative labeled data from a target instrument may be scarce before launch, creating a gap between ground-based model development and in-orbit performance. Updating deployed models introduces further challenges in adaptation, verification, and operational reliability.

This Community Contributed Session brings together researchers, hardware developers, mission designers, and application specialists to address the complete lifecycle of onboard AI: training-data generation, model development, hardware-aware optimization, deployment, adaptation, and validation. It welcomes approaches that combine computational efficiency with robust performance, including transfer learning, few-shot and zero-shot learning, semi-supervised learning, synthetic data generation, and continual learning. Emphasis is placed on demonstrating how methodological advances translate into reliable operation under realistic mission constraints. The session particularly welcomes work on geospatial foundation models adapted for resource-constrained onboard environments, including model compression, adaptation, and mission-specific specialization.

Beyond accelerating data exploitation, onboard AI is increasingly enabling autonomous mission operations, including dynamic sensing, event-driven tasking, and adaptive observation strategies that can maximize the scientific and societal value of Earth observation missions. The theme complements the IGARSS 2027 General Themes on onboard processing, large-scale machine learning, remote sensing techniques, and sensor technologies by providing an integrated, mission-oriented perspective on AI deployment. It also complements Special Theme SP.4 through hardware-aware computing methods, SP.5 through resource-efficient adaptation of foundation models, and SP.6 through evaluation under representative onboard conditions. Its distinctive contribution is the joint consideration of learning with limited data, constrained execution, in-orbit adaptation, and operational trust across sensing modalities and applications.

Submissions are welcome on (but not limited to):
•	Onboard deep learning for EO,
•	Task-aware satellite data compression, band selection, and feature extraction,
•	Model compression, quantization, pruning, and knowledge distillation,
•	Hardware-aware architecture design and optimization,
•	Onboard training, continual learning, algorithm configuration,
•	Transfer, few-shot, zero-shot, and semi-supervised learning,
•	Robustness to noise, degraded observations, and distribution shifts,
•	Onboard image enhancement and multimodal analysis,
•	Distributed computing and collaborative learning in space,
•	Hardware acceleration and energy-efficient processing,
•	Data-level digital twins and synthetic training datasets,
•	Verification & Validation of AI in space,
•	Explainability, uncertainty estimation, and operational reliability,
•	Validation and benchmarking of accuracy, latency, memory, and energy use,
•	Foundation models for onboard AI,
•	In-orbit demonstrations and evidence of scientific, industrial, or societal impact.

By connecting AI innovation with mission requirements and application needs, this theme aims to advance dependable onboard intelligence and accelerate the delivery of useful EO information.

This CCT proposes to discuss the overview, research achievements, future plans, and international cooperation of the Advanced Land Observing Satellite (ALOS), Japan's flagship Earth observation satellite series mission. 
The ALOS series of the Japan Aerospace Exploration Agency (JAXA) has been continuously operated since 2006, and currently ALOS-2 and ALOS-4 onboard L-band SAR have been in operation since 2014 and 2024, respectively, to perform precise Earth surface observations in various application fields. The next-generation ALOS optical mission will be mounted on the LiDAR instrument to measure terrain and forest height precisely and generate a precise digital surface model (DSM) and digital terrain model (DTM) combined with high-resolution optical satellites. To achieve mission objectives, calibration and validation (Cal/Val) is essential and encourages science and application development. International collaborations are also important to conduct these activities. Based on these results, we will also discuss the prospects for future ALOS series missions, including the continuity of the missions, the importance of international cooperation, and the advantages and disadvantages that should be reflected in future missions. This envisions collaboration with NASA/ISRO’s NISAR and with ROSE-L, the L-band SAR mission planned by ESA. In this session, we will discuss collaboration and cooperation with these international missions. 

This CCT proposes to discuss the overview, research achievements, future plans, and international cooperation of the Advanced Land Observing Satellite (ALOS), Japan's flagship Earth observation satellite series mission. 
The ALOS series of the Japan Aerospace Exploration Agency (JAXA) has been continuously operated since 2006, and currently ALOS-2 and ALOS-4 onboard L-band SAR have been in operation since 2014 and 2024, respectively, to perform precise Earth surface observations in various application fields. The next-generation ALOS optical mission will be mounted on the LiDAR instrument to measure terrain and forest height precisely and generate a precise digital surface model (DSM) and digital terrain model (DTM) combined with high-resolution optical satellites. To achieve mission objectives, calibration and validation (Cal/Val) is essential and encourages science and application development. International collaborations are also important to conduct these activities. Based on these results, we will also discuss the prospects for future ALOS series missions, including the continuity of the missions, the importance of international cooperation, and the advantages and disadvantages that should be reflected in future missions. This envisions collaboration with NASA/ISRO’s NISAR and with ROSE-L, the L-band SAR mission planned by ESA. In this session, we will discuss collaboration and cooperation with these international missions. 

The rapid expansion of spaceborne Synthetic Aperture Radar missions and operational ground-motion services is transforming InSAR from a specialized geodetic technique into a major source of large-scale, long-term Earth-observation time-series data. Continental-scale products such as the European Ground Motion Service (EGMS) now provide millions of deformation measurements across diverse geological, climatic, and anthropogenic environments. This rapid growth in data volume creates a new challenge: the scientific bottleneck is increasingly shifting from the generation of InSAR products to the automatic interpretation, characterization, prediction, and quality enhancement of massive deformation time series.

This Community Contributed Theme, Artificial Intelligence for Large-Scale InSAR Time-Series Analysis, focuses on the emerging integration of artificial intelligence with InSAR deformation analytics. It will bring together advances in machine learning, deep learning, self-supervised learning, state-space models, graph neural networks, foundation models, and physics-guided learning for five closely related tasks: classification of deformation behaviours, clustering and pattern discovery, deformation forecasting, anomaly and change detection, and missing-value imputation and reconstruction. The theme also welcomes work on representation learning, uncertainty quantification, explainable AI, domain adaptation, transfer learning, multimodal fusion, and scalable analysis of continental- and national-scale InSAR archives.

The importance of this topic is driven by both scientific opportunity and operational need. Large InSAR datasets such as EGMS contain rich information about volcanoes, landslides, subsidence, infrastructure instability, tectonic deformation, glaciers, permafrost, and other dynamic Earth-surface processes, but they are also highly heterogeneous, noisy, non-stationary, and frequently incomplete. Conventional manual interpretation and traditional statistical methods are increasingly insufficient for efficiently extracting useful information from such large and complex datasets. AI offers a promising pathway toward automated discovery of deformation patterns, early identification of anomalous behavior, forecasting of future motion, and reconstruction of missing observations, thereby supporting more scalable and intelligent geohazard and infrastructure monitoring.

Iceland provides a particularly timely and scientifically relevant setting for this theme. Rapid volcanic, tectonic, geothermal, glacial, and cryospheric processes generate complex and transient deformation signals that are well suited to testing advanced AI methods. At the same time, snow cover, decorrelation, atmospheric effects, and irregular observations can produce substantial data gaps and unstable time-series quality. Recent deformation activity on the Reykjanes Peninsula further highlights the need for methods capable of distinguishing background variability from anomalous and rapidly evolving ground motion.

The proposed theme complements, rather than duplicates, existing IGARSS General and Special Themes. Differential SAR Interferometry primarily addresses interferometric measurement and processing; Classification and Clustering, Change Detection and Temporal Analysis, and Large-Scale Machine Learning address broader methodological problems across remote sensing. This theme specifically targets the downstream intelligent interpretation of large-scale InSAR deformation time series after InSAR products have been generated. By bridging the InSAR, artificial intelligence, geodesy, and geohazard communities, it provides a focused forum for developing the next generation of scalable, predictive, and interpretable ground-motion analytics.

Over the last forty years, a large number of algorithms and solutions have been developed to fully exploit the potential of Synthetic Aperture Radar (SAR) and extract information from SAR data. Thanks to their coherent nature and their ability to operate in any weather condition, SAR data are now used for many applications, such as 3D reconstruction, deformation monitoring, feature extraction, land cover classification, target recognition, scatterer detection and data fusion, across different domains (urban, natural, marine, cryosphere, etc.). These applications rest on a solid chain of classical processing steps, such as focusing, despeckling, co-registration, phase filtering, unwrapping and parameter estimation. These steps are grounded in physical models and statistical signal processing theory.

In recent years, the SAR community has witnessed a rapid and widespread shift from traditional model-based solutions to Artificial Intelligence (AI) and deep learning (DL) based ones. New sensors with short revisit times produce huge amounts of data, which require fast and accurate processing. The versatility, efficiency and wide applicability of DL methods meet this need. Today, DL-based approaches achieve state-of-the-art performance in a growing number of tasks in amplitude, interferometric, polarimetric and tomographic SAR. More recently, foundation models trained on large SAR archives have also appeared.

The results obtained so far are impressive, and one may now ask whether AI is ready to fully and systematically replace classical methods. However, several boundaries still need to be explored. Reliable ground truth is often missing, and networks are frequently trained on simulated data or on the outputs of classical algorithms. Generalization across sensors, frequencies, resolutions and acquisition geometries is still limited. Preserving phase information and physical consistency is not always guaranteed. Interpretability and uncertainty quantification, which are well established for classical estimators, remain open issues for DL models.

This session addresses the actual potential and, even more, the actual limitations of AI-based methodologies in performing classical, and often not straightforward, signal processing tasks in the SAR domain, with particular attention to SAR amplitude, InSAR, PolSAR and TomoSAR configurations. Contributions comparing AI-based, classical and hybrid (e.g., physics-informed or model-based) approaches on common problems and datasets are particularly welcome.

The session aims to stimulate an open discussion on whether, where and under which conditions AI can replace classical SAR processing, or should rather complement it, seeking new horizons and shared perspectives for the future of SAR data exploitation.

Geospatial Foundation Models (GeoFMs) are increasingly being explored across Earth Observation (EO) applications, offering general-purpose representations that can support downstream tasks with limited labeled data. Their potential to reduce the need for task-specific training data and facilitate EO application development has attracted considerable research interest. However, the effectiveness of GeoFMs for regression tasks remains comparatively underexplored. Existing evaluation efforts have largely relied on classification benchmarks, including EuroSAT, BigEarthNet, UCMerced, and fMoW, leaving open questions about how well these models' learned representations transfer to tasks that estimate continuous variables.
Regression is fundamental to numerous EO applications, including above-ground biomass estimation, forest canopy height prediction, crop yield forecasting, climate modeling, and the retrieval of other biophysical properties. These tasks differ from classification in their prediction objectives and evaluation requirements. Consequently, strong performance on standard classification benchmarks does not necessarily imply that GeoFM representations are equally effective for regression. A more systematic assessment of their practical utility in these settings is therefore needed to better understand their capabilities, limitations, and potential value for downstream EO applications.
This thematic session aims to bring together research investigating the use and evaluation of GeoFMs and their embeddings for regression tasks in EO. Contributions may explore adapting pretrained representations to downstream regression problems, their effectiveness under different data-availability conditions, comparisons with task-specific models or alternative representations, and their practical use in applications such as biomass estimation, canopy height mapping, and agricultural yield prediction. Studies examining the factors that influence regression performance and the limitations of existing representations are also encouraged.
The session complements the special theme SP.5: Foundation Models and Embeddings for Earth Observation by providing a focused perspective on an important but comparatively less explored type of downstream tasks. While the broader theme addresses foundation models and their applications across EO, this session specifically examines their effectiveness for continuous prediction. By bringing together evidence from diverse regression applications, the session aims to foster discussion on the suitability of GeoFMs for biophysical variable estimation and identify opportunities for further research and practical deployment.

With the reentry of the NASA/CYGNSS satellites into the atmosphere, 10 years of high-quality spaceborne GNSS-R data are coming to an end. CYGNSS has become the benchmark dataset for satellite-based GNSS-R, serving as a foundation while other missions were developed and deployed. With over 40 GNSS-R satellites currently orbiting the Earth—ranging from government agency missions to commercial endeavours on platforms from CubeSats to large satellites—this remote sensing technique now offers unprecedented dense and rapid coverage. However, the diversity of instruments, sampling strategies, calibration approaches, and orbital characteristics currently make the joint utilization of these missions challenging. Therefore, developing methodologies for the proper inter-comparison, coordination, and cross-calibration of observables and products to a unified reference is crucial to fully exploiting this rich data ecosystem. The session accepts contributions that study these methodologies as well as the outcome of them: applications that benefit from their resulting fine spatio-temporal resolution.

This session directly addresses General Theme S.3: Spaceborne GNSS-R Missions, while complementing Special Theme SP.6: Datasets, Benchmarks, and Evaluation Methodologies for Earth Observation by discussing methodologies that pave the way for cross-mission fusion. Consequently, the ability to combine multi-mission GNSS-R observations will unlock the fine spatio-temporal resolution required to effectively address Special Theme SP.2: Monitoring Rapidly Changing and Hazardous Environments with Remote Sensing.

Microwave observations provide unique capabilities for investigating the cryosphere across a wide range of spatial scales and depths, from detailed measurements of snow, firn, internal ice structure, and basal conditions to repeated satellite observations of snow and ice over regional to continental scales. However, radioglaciology and satellite remote sensing communities often address closely related physical questions using different observing systems and at different scales. We therefore propose linked CCT themes that aim to bring these communities together and strengthen connections between ground-based and spaceborne observations of snow and ice.

This proposed session focuses on satellite microwave remote sensing of snow and ice. Satellite observations provide repeated measurements over regional to continental scales and enable monitoring of snow, glacier, ice sheets, sea ice, and other cryospheric changes. Particular emphasis will be placed on multi-frequency observations, physical interpretation of microwave signals, retrieval algorithm development, and the integration of satellite observations with ground-based and airborne radar measurements for calibration, validation, and scaling. Contributions exploiting new and emerging satellite missions and combining observations across sensors, frequencies, and platforms are particularly encouraged.

Note that this proposed session is closely linked to another session: Bridging Radioglaciology and Cryosphere Satellite Sensing: Advances in Radioglaciology of Global Ice, which focuses on radioglaciology and subsurface sensing. Together, the two CCT themes complement existing IGARSS General and Special Themes by focusing specifically on connections across observing scales, platforms, and cryospheric domains. Rather than addressing individual sensors or snow-and-ice components in isolation, they will provide a cross-community forum linking detailed physical observations with satellite-scale monitoring, improving our ability to observe and understand a rapidly changing cryosphere.

Ice sheets, glaciers, sea ice and snow play critical roles in Earth’s water cycle, sea-level change, and climate regulation, with radar remote sensing providing essential observations for characterising and monitoring the global cryosphere. Radioglaciology, encompassing both the use of radar to investigate glaciers and ice sheets and the development of radar technologies for these applications, has enabled pivotal scientific discoveries. These include mapping ice thickness across Antarctica and Greenland, constraining the distribution of subglacial and englacial liquid water, identifying debris and internal ice structures, and monitoring the evolution of sea ice. Continued advances in radioglaciology offer opportunities to deepen our understanding of glaciological processes and address outstanding scientific questions.
Recent advances in radar instruments, platforms and analysis methods are opening new ways to observe ice sheets, glaciers, and snow. New software and open-source radar designs are lowering the cost of building new systems. Drones and ground-based platforms have proven essential not only for validating satellite data, but also for understanding subglacial and englacial processes. Autonomous, multi-frequency, polarimetric and phase-sensitive systems are improving both vertical resolution and repeat coverage at a site. New processing, inversion and machine-learning methods are turning these datasets into direct constraints on ice structure, deformation, melt and snow and firn change. 

This session welcomes contributions across radar techniques applied to snow, firn and ice. These include radio-echo sounding of ice thickness, internal structure and basal conditions; airborne, spaceborne and terrestrial SAR and phase-sensitive measurements of surface velocity, strain and deformation; phase-sensitive radio-echo sounders (ApRES) and related methods for vertical strain and basal melt on ice shelves and ice streams; and airborne and ground-based radar for snow and firn density, water content and stratigraphy. It also welcomes work on the instruments, platforms and open-source tools behind these measurements.

The session complements General Theme C.2 (Ice Sheets and Glaciers) by focusing on the radar technologies and methods themselves rather than on ice-sheet science as an outcome. It connects to T.1 (Interferometry), T.2 (Differential SAR Interferometry), T.7 (Sub-surface Sensing), S.16 (Ground Penetrating Radar), C.1 (Snow Cover) and SP.1 (the Arctic).

This session is organised by Rodrigo Gomez Fell (University of Canterbury) and Wolfgang Rack (University of Canterbury), and co-organised by Nicole Bienert (University of Colorado Boulder), together with John Paden (Centre for Remote Sensing and Integrated Systems, University of Kansas). Together, they bring expertise in open-source radar hardware, Antarctic and New Zealand glacier radar surveying, and multi-decadal radar-sounding data processing.

Earth observation is experiencing a growth period for spaceborne imaging spectrometers (hyperspectral imagers). National space agencies and private companies are providing reflectance products from these new sensors at a variety of spectral and spatial resolutions, for example NASA's EMIT (Earth Surface Mineral Dust Source Investigation) at 60 m ground sampling distance (GSD) and DLR's EnMAP (Environmental Mapping and Analysis Program) and Planet’s Tanager-1 and Tanager-2 at 30 m GSD. With proper spectral, radiometric, and spatial calibration and post-launch evaluation, data from multiple sources might be used collectively to support scientific studies and extend applications over larger areas and at greater temporal frequency. This session will provide a forum for commercial, government, and academic organizations and other groups to share their mission-focused efforts and results in advancing imaging spectrometer calibration and validation (cal/val) of products. This session focuses on cal/val of radiance and reflectance products which are the basic geophysical data and the level of data interpreted for surface material composition and chemistry, respectively. This session also provides an opportunity to discuss international collaborations and multi-mission data comparison. Leveraging hyperspectral data with multispectral data is a topic of interest, including presentations on advanced multispectral missions, such as Landsat Next. Airborne surveys and field campaigns often collect fine scale data for calibration and validation of measurements and products; thus, contributions which show applications of airborne and field data to cal/val to drone-based, airborne, and spaceborne imaging spectrometers are sought. Discussion of engagement of the IEEE GRSS Geoscience Spaceborne Imaging Spectroscopy (GSIS) technical committee and working groups is highly encouraged. Rob leads NASA’s EMIT imaging spectrometer mission on the International Space Station and the forthcoming EAGLE-VSWIR global imaging spectrometer mission. 

Change detection has long told us where and when the Earth's surface has changed. Scientists, emergency responders, and policymakers increasingly need answers to deeper questions: what exactly has changed, why it changed, and what is likely to change next. This session focuses on change intelligence, that is, methods that go beyond change maps to interpret, attribute, and anticipate surface change from multi-temporal, multi-sensor Earth observations by coupling imagery with language, knowledge, auxiliary data, and generative modeling. Topics include, but are not limited to:
•	Language-grounded change understanding: change captioning, change question answering, referring change segmentation, and temporal reasoning with multimodal large language models over image pairs and long time series
•	Open-vocabulary, promptable, and cross-sensor (optical, SAR, LiDAR) change perception that generalizes across regions and change types
•	Change attribution: identifying the drivers and processes behind change through causal inference, explainable AI, and fusion with climate, socioeconomic, and in situ data
•	Change anticipation: forecasting future surface states with spatiotemporal generative and world models, including scenario and counterfactual analysis
•	From change maps to change knowledge: decision-ready change reports, change knowledge graphs, and interactive human–AI change analysis
•	Benchmarks and trustworthiness: datasets and metrics for change reasoning, hallucination and temporal consistency of multimodal models, and uncertainty in attribution and forecasting
Importance. Urban expansion, deforestation, glacier retreat, and extreme events are reshaping the planet, and dense time series from public missions and commercial constellations now record these changes in unprecedented detail. Yet conventional change detection outputs, typically binary or class-transition maps, still require substantial expert effort before they can inform science or decisions. Multimodal large language models, open-vocabulary foundation models, and generative world models now make it possible to describe, explain, and anticipate change, but current models struggle with fine-grained temporal reasoning, hallucinate changes that never occurred, and lack rigorous benchmarks. This session speaks directly to the IGARSS 2027 theme, Global Vision for a Changing Planet, and will connect remote sensing with computer vision, natural language processing, causal inference, and land system science.
Complementarity with General and Special Themes. General Theme T.14 (Change Detection and Temporal Analysis) covers change detection and time-series analysis, and work that only improves binary or closed-set change maps belongs there; this session addresses the subsequent questions of what, why, and what next, which require cross-disciplinary methods beyond its usual scope. SP.5 (Foundation Models and Embeddings for Earth Observation) and A.2 (Foundation Models in Geoscience) cover general-purpose models, and SP.6 (Datasets, Benchmarks, and Evaluation Methodologies for Earth Observation) covers generic evaluation, whereas every contribution here centers on reasoning about change across acquisition dates. SP.2 (Monitoring Rapidly Changing and Hazardous Environments with Remote Sensing) is organized around specific environments and hazards, whereas this session is method-centric and spans urban, forest, agricultural, cryospheric, and disaster applications. It also differs from SP.3 (Agentic AI for Earth Observation) by focusing on the model-level capability to interpret and anticipate change rather than on autonomous agent workflows.

Ground-level sensors provide unique contextual, geometric, and semantic information that complements satellite and aerial Earth observation. Establishing reliable connections across these disparate sensing perspectives, despite differences in viewpoint, scale, resolution, sensing modality, temporal sampling, and content visibility, can enable a broad range of emerging applications. These include multi-source mapping and scene understanding, cross-view geolocation, 3D scene reconstruction across varying altitudes, data fusion for digital twins, and cross-source change detection and monitoring of environmental, urban, and infrastructure dynamics.

Recent advances in deep learning, multimodal foundation models, geometric reasoning, and representation learning have created new opportunities to integrate Earth observation data with observations collected at or near ground level. These ground-level observations may include conventional imagery and video as well as LiDAR, mobile mapping systems, vehicle-mounted sensors, Internet-of-Things (IoT) sensors, and other geospatially referenced measurements. Despite this progress, robust fusion remains challenging because observations may differ dramatically in geometry, spatial and temporal resolution, radiometry, modality, coverage, and uncertainty. The next frontier is moving beyond pairwise image matching toward flexible methods that can associate, align, and jointly reason over heterogeneous observations collected across platforms, viewpoints, modalities, and time.

This theme session invites contributions that advance algorithms, datasets, evaluation methods, and applications for connecting Earth observation with ground-level sensors. Topics of interest include, but are not limited to:
-	Multi-altitude, multi-view, and multi-sensor data fusion
-	Fusion of satellite and aerial imagery with ground imagery, video, LiDAR, mobile mapping, and other sensor observations
-	Cross-view localization, registration, correspondence, and geometric alignment
-	Cross-view 3D reconstruction, scene modeling, and digital-twin generation
-	Multi-source change detection, temporal fusion, and activity understanding
-	View synthesis and neural rendering across sensing perspectives
-	Foundation models, multimodal embeddings, and vision-language models for cross-source association and reasoning
-	Methods for handling differences in resolution, geometry, modality, uncertainty, and temporal sampling
-	Applications in environmental monitoring, infrastructure assessment, mapping, navigation, disaster response, and situational awareness

To promote innovation, reproducibility, and quantitative comparison, the session is also expected to include competition activities exploring challenging problems at the intersection of Earth observation and ground-level sensing. Competition tasks and datasets will be announced separately as details are finalized. Competition results and winning teams will be announced during the session.

Earth observation with imaging spectrometers (hyperspectral imaging) and advanced multispectral sensors offers unprecedented insight into the composition and distribution of critical mineral resources. These techniques provide detailed spectral information that can reveal hidden or previously overlooked deposits, significantly improving the accuracy of resource assessments for elements essential to clean energy technologies and modern economies. The accelerating global demand for raw materials - combined with the designation of numerous elements as critical - has created heightened pressure to evaluate known critical mineral resources and efficiently screen prospective areas. This session focuses on the transformative impact of spectral techniques and Earth observation data in revolutionizing geologic mapping, ore-body knowledge, and exploration discoveries for critical minerals. Through global case studies, the session highlights how advanced remote sensing technologies at drone, airborne, and spaceborne levels are redefining our understanding of critical mineral deposits. These remote observations are complemented by proximal sensing systems, including field spectroradiometers, drill-core sensors, and laboratory spectrometers. A key component of mapping geology and specific minerals from multispectral and hyperspectral remote sensing imagery is ground validation of derived products. Publicly available ground validation datasets—such as Cuprite, Nevada, U.S.A., and Rocklea Dome, Western Australia, Australia—provide opportunities for the international research community to compare different remote sensing image-processing and calibration methods. However, publicly available ground validation datasets that go beyond simple image classification remain scarce. In addition to soliciting global case studies on mapping critical minerals and associated geology, this session invites presentations on workflows and methods for validating geoscience products derived from optical satellite imagery, including ongoing and planned field campaigns and conceptual approaches.

Topics of Interest Include:
•	Hyperspectral and multispectral mapping of critical mineral deposits and host rocks
•	Spectral characterization of lithium, rare earth elements, cobalt, nickel, copper, and other critical minerals
•	Drone-, airborne-, and spaceborne-based exploration case studies for critical minerals
•	Integration of remote sensing with proximal sensing (field spectroradiometers, drill-core sensors, lab spectrometers)
•	Ground validation strategies, public datasets, and calibration/validation workflows
•	Methods for validating geoscience products beyond simple image classification
•	Ongoing and planned field campaigns for critical mineral mapping
•	Conceptual approaches for improving accuracy and reproducibility in critical mineral remote sensing

The increasing volume, dimensionality, and acquisition rates of remote sensing data pose significant challenges for onboard storage, downlink capacity, processing resources, and the timely delivery of Earth observation products. These challenges are particularly pronounced for missions delivering high-resolution (temporal and spatial) products, as well as for platforms operating under stringent power, memory, and communication constraints.

This session will address advances in data compression and onboard processing techniques designed to reduce the volume of remote sensing data while preserving the information required for scientific analysis and operational applications. Topics of interest include lossless, near-lossless, and lossy compression, adaptive and content-aware methods, learning-based compression, dimensionality and feature reduction, intelligent data selection, and real-time onboard preprocessing and edge computing. Contributions covering optical, multispectral and hyperspectral imaging, synthetic aperture radar (SAR), LiDAR, thermal sensing, and other remote sensing modalities are encouraged. Particular emphasis will be placed on integrated and application-driven approaches that combine compression with onboard analysis, enabling efficient data transmission, timely information delivery, and more autonomous operation of Earth observation systems.

The importance of this topic stems from the growing mismatch between the volume of data generated by modern sensors and the resources available to store, transmit, and process it. Effective data reduction can alleviate communication and storage constraints, reduce energy consumption, and enable faster responses to dynamic events and time-critical applications. However, achieving these benefits requires balancing compression efficiency against information preservation, computational complexity, latency, and power consumption. Evaluating these trade-offs across different sensors and applications is therefore essential for developing robust and operationally relevant solutions.

The proposed session most directly complements General Theme A.4: Real-time Analytics, On-board Processing and Edge Computing, by focusing specifically on compression and data reduction as enabling technologies for efficient onboard operations. It also complements T.8: Feature Extraction and Reduction through methods that reduce data volume and dimensionality while retaining task-relevant information, and connects with sensor-oriented themes addressing hyperspectral missions, multi-platform Earth observation, optical sensors, and SAR instruments. Furthermore, it supports relevant Special Themes, including SP.2 on monitoring hazardous and rapidly changing environments, SP.3 on agentic AI for Earth observation, SP.5 on foundation models and embeddings, and SP.6 on datasets, benchmarks, and evaluation methodologies. Rather than duplicating these themes, the session provides a cross-cutting forum connecting compression algorithms, onboard computing architectures, sensor characteristics, and downstream applications. Its distinctive contribution is to emphasize data reduction as a means of improving resource efficiency while preserving the scientific and operational value of remote sensing information.

Scope: This theme brings together data-centric AI and deployment-oriented evaluation for foundation models in Earth observation (EO). On the data side, we invite work on the creation, curation, and governance of large, open, multi-sensor, and multi-temporal EO datasets, including datasets for multimodal pretraining, label quality assessment, harmonization of class definitions, and documentation through dataset and model cards. We also welcome contributions that apply or extend standards and best practices from the GEO, CEOS, OGC, ISO, and STAC communities to improve interoperability, metadata consistency, and transparent model reporting. On the evaluation side, we seek benchmark designs that span diverse geographies, sensors, and observation conditions, together with shared metrics for robustness, generalization, uncertainty, and fairness. We are especially interested in studies that test transferability directly, including controlled deployment experiments, systematic failure analysis, confidence calibration, performance without local labeled data, the gains and limits of domain adaptation, sensitivity to observation windows, and the trade-off between timeliness and accuracy. Contributions comparing released foundation models and geospatial embedding products under common protocols are encouraged, as are methods for detecting and correcting silent failures across the full monitoring workflow. Work that separates true transfer failure from artifacts of regionally inconsistent reference data is particularly welcome. Submissions may describe new open benchmarks, scalable evaluation pipelines, reproducible model comparisons, community platforms for sharing datasets and models, or operational case studies with validated corrections.

Importance: Foundation models are changing how EO data are analyzed, and released models now enable extending flood, crop, vegetation, and land-cover monitoring to new regions with little additional training. Progress depends less on model scale than on the data used to train, validate, and document these models, and on whether reported performance holds once a model is deployed. Current practice falls short on both counts. EO datasets differ in quality, licensing, metadata, and class definitions. Benchmarks rarely follow shared protocols, so results are hard to compare across sensors, resolutions, and tasks. Benchmark accuracy often says little about behavior under operational conditions, where observation timing, local phenology, missing spectral bands, cloud contamination, and mismatched label schemes alter the resulting maps. Failures at any stage of the workflow, from data acquisition and preprocessing to inference and postprocessing, can silently degrade the output map while the pipeline runs to completion, and the model's predicted confidence may remain high even as accuracy collapses. By connecting dataset quality, standardized benchmarking, and evidence from real deployments, the theme aims to provide researchers, agencies, and industry with a common basis for judging when a geospatial AI system can be trusted beyond the benchmark it was trained on.

Rationale for complementing General and Special Themes: The theme complements SP.5 and SP.6, which address the development of foundation models and the construction of datasets and benchmarks, by focusing on whether performance established on benchmarks survives operational deployment. It builds on the Community Contributed Themes on data-centric AI and the Special Theme on responsible and ethical AI for Earth observation previously supported by the IEEE GRSS Earth Science Informatics Technical Committee.

Scope
This theme covers data-centric approaches to machine learning for Earth observation (EO). These approaches treat the systematic focus on training and evaluation data, rather than ever-larger model architectures, as the main lever for reliable and improved performance. We invite contributions on:
    • label quality, annotation uncertainty & data noise
    • weakly, semi-, and self-supervised learning (exploiting sparse or imperfect labels)
    • data selection, curation, pruning, and valuation 
    • dataset-driven model design and evaluation protocols, such as data slicing, stratification, and analysis of spatial autocorrelation and distribution shifts 
Hybrid methods that combine data- and model-centric perspectives are equally welcome, as are studies on the reproducibility and documentation of data pipelines.

Importance
EO has abundant data but few labels. Reference data are expensive and unevenly distributed across regions and seasons. They often come from outdated or inconsistent sources and suffer from issues such as geolocation errors, class ambiguity, and temporal mismatch with the imagery. Because of spatial autocorrelation and geographic bias, standard random splits can substantially overestimate how well a model generalizes. Missing or flawed label information affects model training and model evaluation, where predictions are assessed against biased ground truth.
The choice of which data is used for pretraining, fine-tuning, and evaluation has become decisive for accuracy, fairness, and trust in operational products. Data-centric learning offers principled ways to:
    • get more value from limited annotation budgets 
    • expose and correct systematic data deficiencies 
    • make model behavior transparent across regions, sensors, and acquisition conditions 
The theme builds on the position paper "Data-Centric Machine Learning for Earth Observation" (Roscher et al., 2024).

Complementarity with General and Special Themes
Data-centric learning is a cross-cutting methodological perspective that goes beyond the existing themes. It thus enriches the community discussion complementing the following general and special themes:
    • A themes focusing on novel AI, machine learning, and big data methodologies for analyzing remote sensing data. 
    • T themes that cover different tasks such as classification, segmentation, change detection, all of which are targeted and enriched by data-centric learning
    • SP.5 which centers on model representations. This theme relates in asking how data properties determine the quality of those representations and of downstream adaptation. 
    • SP.6 treating datasets and benchmarks as community resources and asking how models are compared. Our theme adds to this by focusing on methods that improve, select, generate, and diagnose data within the learning process. 
Today this work is scattered across task- and application-specific sessions. Collecting it in one theme connects machine learning researchers, producers of reference data, and application domains such as agriculture, land cover mapping, and hazard monitoring. The result is an interdisciplinary forum for an emerging topic of broad community interest.

Freshwater systems are exposed to interrelating environmental compressions arising from climate change, extreme hydroclimatic events, land-use changes, urban development and anthropogenic pollution. These pressures can change water quality, impact pollutant transport and human exposure and had an impact on the resilience of urban and river-basin water systems. To address these challenges, methods integrate Earth observation with environmental processes, human-health risk assessment and climate-resilient water management. Satellite based remote sensing, in-situ monitoring, geographic information systems (GIS), hydro-informatics and artificial intelligence (AI) offer opportunities to investigate water-related risks across spatial and temporal scales. Integrating remotely sensed observations with ground-based measurements, hydroclimatic records and environmental models can improve the characterization of water-quality dynamics, contaminant distribution and the vulnerability of water systems and dependent populations. However, the relationship between observed environmental changes, water-quality degradation, human exposure and climate-related water-system impacts remain insufficiently integrated.
The proposed theme will promote an interdisciplinary understanding of compound water risks by integrating Earth observations with water-quality dynamics, environmental exposure and human-health risk and water-system resilience under changing climatic and anthropogenic conditions. It invites papers on the impacts of floods and droughts on contaminant movement, changes in land use and the impacts of hydroclimatic variability on freshwater quality and the combinations of stressors that threaten vulnerable populations and water facilities.
Topics of interest include Earth observation of freshwater environments; integration of satellite and in-situ measurements; geospatial contaminant assessment; environmental exposure and health-risk modelling; hydroclimatic extremes and compound water hazards; integrated hydrological, hydraulic, and water-quality modelling; AI-enabled environmental assessment; uncertainty quantification; and geospatial approaches to climate adaptation and water-system resilience.
The distinctive contribution of this theme is its emphasis on connecting environmental observations with exposure pathways and water-system vulnerability, rather than focusing exclusively on individual AI techniques, water-quality retrieval methods or specific environmental hazards. The theme aims to foster integrated, scientifically sound and water security-oriented approaches by integrating Earth Observation, Environmental Engineering, Hydro-informatics, Human-Health risk assessment and Climate-resilient Water Management. 

Latin America faces a complex combination of environmental, geological, and societal challenges driven by its diverse geography, active tectonic setting, climate variability, and rapid urbanization. The region encompasses globally significant ecosystems, including the Amazon basin and the Andean cordillera, while simultaneously experiencing recurrent natural hazards such as earthquakes, tsunamis, floods, landslides, droughts, and wildfires. These challenges are further intensified by climate change, land-use transformations, and increasing population and infrastructure exposure. Recent seismic events across South America, together with recurrent extreme hydrometeorological events associated with the El Niño–Southern Oscillation (ENSO), underscore the importance of reliable, timely, and accessible geospatial information for understanding environmental processes, monitoring territorial changes, and supporting disaster risk reduction. 

This Community Contributed Theme aims to bring together scientific advances and operational applications of Earth Observation technologies addressing Latin America's distinctive challenges. It welcomes contributions employing spaceborne SAR, optical imagery, LiDAR, unmanned aerial systems (UAS), and emerging artificial intelligence methodologies for environmental monitoring, urban and infrastructure dynamics, natural hazard assessment, and disaster resilience. Attention is given to methodological challenges associated with complex mountainous terrain, persistent cloud cover, rapidly changing urban landscapes, limited ground observations, and the transfer of remote sensing technologies into operational decision-making.

The proposed session complements the IGARSS 2027 technical program by highlighting Latin America as a region for developing, evaluating, and operationalizing Earth Observation methodologies across diverse environmental, geological, and socioeconomic conditions. Specifically, it contributes to D.10 (Remote Sensing for Climate Change Impacts) through applications addressing ecosystem transformations, agricultural vulnerability, water resources, and urban climate dynamics. It complements SP.2 (Monitoring Rapidly Changing and Hazardous Environments with Remote Sensing) by addressing the observation and characterization of dynamic environmental processes and natural hazards, including floods, earthquakes, landslides, droughts, and wildfires. Furthermore, it contributes to D.7 (Risk and Disaster Management) by emphasizing the application of remote sensing information to hazard assessment, exposure analysis, disaster impact evaluation, and decision-making for risk reduction and emergency response. By connecting these themes through Latin American case studies, the session seeks to advance transferable methodologies and operational applications that respond to the region's distinctive environmental and societal challenges.

Additionally, the proposed session seeks to foster collaboration among research groups engaged in knowledge generation, enabling them to present their latest findings, methodologies, and innovations, emphasizing connecting local researchers with the broad scientific community. This collaborative environment will facilitate the exchange of ideas and expertise crucial for effectively tackling Latin America's unique challenges.

This Community Contributed Theme is proposed in connection with the Committee on Earth Observation Satellites (CEOS) Working Group on Disasters (WGD) and its new three-year Landslide Pilot, “Earth Observation Action for Landslides (EOAction 4 Landslides),” led by the Italian Space Agency (ASI) and launched in 2026. The theme will bring together research and operational communities working on the use of Earth Observation (EO) for landslide risk management, with emphasis on methods that can move beyond isolated case studies toward transferable, validated, and user-oriented workflows.

The proposed CCT addresses five complementary dimensions identified by the CEOS Pilot: (1) multi-sensor EO approaches for landslide detection and monitoring, exploiting the complementarities of SAR, optical and other multi-mission geospatial observations; (2) Artificial Intelligence for landslides, including machine learning, deep learning, computer vision and related methods for detection, mapping and multi-temporal analysis; (3) EO-based landslide vulnerability assessment, extending the role of EO from hazard characterization toward the assessment of exposed buildings, infrastructure and other elements at risk; (4) landslides in humanitarian contexts, where rapid and reliable geospatial information can support accessibility assessment, emergency response and humanitarian operations; and (5) landslides and climate change, with particular attention to long EO and climate data time series and their use for investigating changes in triggering conditions, frequency and severity.

The session will encourage contributions that compare methods across geographic regions, exploit multi-mission archives and new acquisitions, quantify limitations and uncertainties, and demonstrate relevance to different phases of the disaster risk management cycle. Particular attention will be given to demonstrations in landslide-prone hotspots, use of CEOS-agency satellite assets (e.g. Copernicus, Landsat, NISAR, COSMO-SkyMed, SAOCOM, Pleiades) and virtual-constellation concepts, and the transfer of scientific and algorithmic advances into pre-operational or operational products for civil protection authorities, public administrations, humanitarian organizations and other end users.

The proposed theme complements rather than duplicates the IGARSS General and Special Themes by creating a focused cross-cutting forum at the intersection of AI, multi-sensor EO, vulnerability, humanitarian response and climate-driven landslide risk. It is closely aligned with D.7 Risk and Disaster Management and SP.2 Monitoring Rapidly Changing and Hazardous Environments with Remote Sensing, while also connecting to AI and Big Data and Land Applications. 

Arid and semi-arid regions face increasing environmental pressures associated with water scarcity, extreme heat, rapid urbanization, land degradation, coastal development, and climate variability. These environments are particularly challenging to monitor because of sparse in situ observations, high spatial heterogeneity, strong seasonal variability, and the close interaction between terrestrial, coastal, atmospheric, and human systems. Earth observation provides a critical capability for addressing these challenges by enabling consistent, spatially distributed, and repeated measurements over large and often data-limited regions.
This Community-Contributed Theme will focus on advances in Earth observation for monitoring, understanding, and predicting water and environmental processes in arid and semi-arid regions. The scope includes coastal and inland water quality, marine and coastal ecosystems, water-resource monitoring, soil and vegetation conditions, agricultural systems, land degradation, urban environments, greenhouse-gas observations, extreme heat, and climate-driven environmental change. Contributions using satellite, airborne, UAV, and ground-based remote sensing are encouraged, particularly studies that integrate multi-sensor observations with in situ measurements, environmental modeling, artificial intelligence, and data-driven approaches.
The proposed theme is important because many remote-sensing methodologies developed for temperate or humid environments cannot be directly transferred to arid regions without accounting for their distinctive atmospheric, surface, hydrological, and coastal characteristics. Bringing these studies together will promote methodological advances, cross-regional comparisons, improved validation strategies, and stronger integration between Earth observation and environmental-process understanding.
The theme complements the existing IGARSS General and Special Themes by providing a cross-cutting environmental perspective that connects topics traditionally addressed separately under oceans, inland waters, agriculture, soils, atmosphere, climate change, and artificial intelligence. Rather than duplicating these themes, it provides a unifying framework centered on the unique observational challenges and environmental priorities of arid and semi-arid regions. This interdisciplinary focus is expected to attract contributions from researchers working across the Middle East and North Africa, as well as other water-limited regions worldwide, while strengthening the broader relevance of Earth observation to environmental resilience and sustainable resource management.

The NASA Soil Moisture Active Passive (SMAP) satellite mission is in its eleventh year of measurements acquisition. SMAP is the first NASA Decadal Survey Mission launched in 2015 with a conical scanning polarimetric radiometer based on a 6-m deployable mesh antenna.  On board radio frequency interference (RFI) algorithms, utilizing temporal, frequency and polarization characteristics of the radiometer data, were implemented to mitigate the impact of RFI. The global SMAP L-band microwave radiometry has been used to derive surface soil moisture, vegetation optical depth (indicator of integrated above-ground vegetation water content), and landscape freeze/thaw status estimates over land regions. Over the oceans the data are used to estimate sea surface salinity and surface wind speed.  Over the cryosphere the data are used to estimate snow and ice properties as well as firns. The soil moisture estimates are the bases for understanding water stress in vegetation and hence used for applications in carbon cycle studies.  They are also used for monitoring drought, water resource management, and forecast of agriculture productivity.  The SMAP radiometer data and the NASA-ISRO Synthetic Aperture Radar (NISAR) data have been used to produce a high-resolution Active-Passive (AP) soil moisture product with a resolution of 100 m, which are essential for applications to many agriculture fields with small plots globally. To calibrate and validate (Cal/Val) the new high-resolution datasets, SMAP project has been collaborating with many partners internationally for in situ observations. This session includes presentations on the status of the SMAP observatory, enhancements in the science data processing, new application of the decade-long SMAP data and Cal/Val of the new SMAP-NISAR AP product. 

New and emerging missions, including ESA’s Biomass and the NASA–ISRO NISAR mission, expand the range of available frequencies, polarimetric capabilities, acquisition modes, and observation strategies. The growing diversity of SAR observations, from multi-baseline and multi-frequency acquisitions to multi-polarimetric and multi-angle measurements, acquired by one or more systems, offers new ways to investigate environmental processes and to exploit the complementary capabilities of different sensors. This Community-Contributed Theme invites contributions on multi-sensor SAR for environmental monitoring, drawing on established and emerging missions alike. Topics may include forest structure and biomass, carbon monitoring, land-cover and land-use change, agriculture, soil moisture, wetlands, snow and ice, coastal environments, and natural hazards. We also welcome studies of tropospheric properties and effects, such as water-vapour mapping and atmospheric phase delays, as well as urban applications, including ground deformation, infrastructure monitoring, and urban change. Contributions may develop new applications, exploit underused observables, or introduce methods for combining data across frequencies, polarizations, baselines, incidence angles, spatial resolutions, and revisit intervals. Submissions may examine sensor synergies, retrieval strategies, validation frameworks, and transferability of methods across regions, ecosystems, and different monitoring scales.
The growing diversity of SAR missions and acquisition strategies creates opportunities for more complete, timely, and reliable environmental indicators, while raising challenges in interoperability, consistency, and interpretation. Addressing these challenges requires collaboration across sensor, signal-processing, and application communities. The proposed theme complements IGARSS General and Special Themes by offering a cross-cutting perspective on the new capabilities, applications, and methodologies enabled by multi-mission SAR observations. It is not restricted to a particular mission or environmental domain and aims to encourage transferable approaches to environmental monitoring.

The ESA Scout HydroGNSS satellites were launched in November 2026 with the intention to use GNSS Reflectometry technique to sense hydrological land variables, including soil moisture, inundation, freeze-thaw and biomass, with secondary products of ocean wind speed and sea ice extent. The payload has advanced since TDS-1 and CYGNSS, and HydroGNSS introduces Galileo, dual frequency, dual polarisation and complex channel measurements. This session covers the performance and validation of baseline products, and advances and applications enabled by the innovative measurements.

This is important as it is the first mission launched of the Scout Programme of the ESA Earth Observation Directorate, applying small satellites for Earth Observation science. HydroGNSS introduces several significant GNSS-R innovations, such as dual polarisation, dual frequency and complex channel which could have a large impact on the technique GNSS reflectometry. The data has been released on a free and open basis from June 2026 onwards and many users of HydroGNSS data are expected to be presenting for the first time.

Traditionally, many Earth Science studies have focused on the use of low-Earth orbiting (LEO) platforms due to generally more spectral information and higher spatial resolution. However, as technology improves, the spatial and spectral characteristics that used to be unique to LEO are now available in geostationary (GEO) orbit, which brings in the often valuable temporal component. Earth processes that change on the order of minutes or hours can be sufficiently sampled from GEO to provide brand new insights not previously possible. Add to that the thriving tool of Machine Learning (ML), which is often able to pick up on signals in spatial, spectral, and temporal simultaneously, and research using GEO sensors has near limitless applications.

In this session, we are looking for papers that bring in data from sensors flying on either current or future planned geostationary satellite platforms to tackle a modern Earth science problem. Examples may include current or future geostationary imagers, current or future geostationary hyperspectral infrared sounders, or other geostationary spectrometers aimed at applications such as air quality or ocean color. EUMETSAT launched the Meteosat Third Generation Infrared Sounder (IRS) in 2025, designed for weather applications; tackling the frequently updating hyperspectral data from this platform is a good example. Papers may also include data from other platforms, including LEO satellite, surface-based, etc. The use of Machine Learning is optional but encouraged, and papers that take advantage of frequent revisits are especially encouraged. This session will cover multiple conference themes, including Multi-Platform Earth Observation (S.6), Spacebourne Hyperspectral Missions (S.4), Large-scale Machine Learning (A.3), and Atmospheric Sounding (M.3), among others.

Scope. This session explores how agentic remote sensing can move from automating analyses to making verifiable discoveries about the Earth system. It spans the entire research cycle, from autonomous generation of scientific products and knowledge synthesis to hypothesis formulation, validation, and mechanism discovery. Particular emphasis is placed on recursive self-improvement (RSI), in which scientific agents iteratively refine their workflows, analytical methods, and hypotheses through evidence-based evaluation and feedback, within bounded and auditable improvement loops. By integrating multi-source Earth observations, physical models, and domain knowledge, such agents can accelerate investigations across the atmosphere, hydrosphere, cryosphere, biosphere, and geosphere, and are particularly suited to uncovering complex cross-sphere interactions and feedback mechanisms. Topics include, but are not limited to:
•	End-to-end agents that autonomously generate, validate, and document scientific products from multi-source Earth observations
•	Knowledge synthesis across literature, data archives, and physical models to support hypothesis formulation
•	Hypothesis generation, testing, and falsification, including causal and mechanism discovery of cross-sphere interactions and feedbacks
•	Recursive self-improvement of workflows, tools, code, and models driven by evidence-based evaluation and experimental feedback
•	Physics-aware verification, uncertainty quantification, and safeguards against spurious or irreproducible findings
•	Benchmarks, provenance tracking, reproducibility, and human oversight for agent-driven Earth science
Importance. Earth observation archives now grow faster than scientists can formulate and test hypotheses against them, and many of the most consequential open questions, such as how a perturbation in one sphere propagates to others, involve high-dimensional, multi-sensor, and multi-scale evidence. Agents built on large language and foundation models can already plan multi-step analyses, call remote sensing tools and physical models, and write and execute code. What is still missing is a rigorous path from automating analyses to producing discoveries that are novel, reproducible, and physically consistent. RSI offers such a path, but it raises new questions of evaluation, reliability, and scientific integrity that the community has yet to address collectively. Held in Iceland, where volcanic, glacial, oceanic, and atmospheric processes are tightly coupled, IGARSS 2027 is a fitting venue to bring the AI, remote sensing, and Earth system science communities together around these questions.
Complementarity with General and Special Themes. While Special Theme SP.3 focuses broadly on agentic AI for Earth observation, this session emphasizes self-improving, end-to-end scientific discovery, moving beyond the automation of individual remote sensing tasks toward autonomous, iterative, and scientifically verifiable advances in Earth system understanding, judged by whether agent-generated hypotheses survive falsification, physical-consistency checks, and independent reproduction. Agent studies without a discovery or verification component remain best suited to SP.3. It also connects SP.3 with SP.5 (Foundation Models and Embeddings for Earth Observation), whose models agents can use as tools, and with G.1 (Earth System Modeling) and the Applications themes (C, L, M, O), which supply the physical models and scientific questions against which agent-generated discoveries must be verified. SP.1 (Geoscience and Remote Sensing in the Arctic) offers natural test cases, such as snow–vegetation–permafrost feedbacks.

Rapid advances in Earth observations are transforming our ability to monitor greenhouse gases, atmospheric pollutants and air quality. Newer satellite missions, airborne and UAV platforms, ground-based sensors and emerging commercial sensing systems provide large amounts of spatial, spectral and temporal information on atmospheric composition. To transform this data into meaningful information, advances spanning multiple area/subareas including remote sensing, atmospheric physics, inverse modelling, multi-modal data fusion, and AL/ML models are required. Realizing their full potential would require bringing together these distinct areas into coherent frameworks and methodologies.

The main objective of our session is integration of satellite, air-borne, mobile or ground-based sensing with atmospheric transport models, meteorological information as well as data-backed past trends. We aim to bring together researchers working across the complete “pixels-to-plumes-to-concentration” information chain, encompassing areas mentioned above to achieve this meaningful data transformation.  Through our session, we also intend to pay particular emphasis on approaches that bridge traditionally separate components of atmospheric monitoring (sensor measurements, traditional retrieval algorithms, transport and inversion models, source identification for emissions, actionable environment information and many more). Connecting these traditional modalities and techniques with modern frameworks like AL/ML based models, in-site sensors, and newer multispectral and hyperspectral satellites can help identify common methodological challenges, promote interdisciplinary approaches and encourage the creation and usage of multi-modal datasets to ensure optimal detection, quantification and source-identification of harmful emissions. Refinement of existing air-quality and exposure-related indicators and development of newer ones is also encouraged.

Call for Papers: This session will receive submissions, talks, and posters on a wide range of topics, perspectives, and ideas including but not limited to:
•	Remote sensing, detection and quantification of greenhouse gases, including CH4, CO2 and N20, and air pollutants including SOx, NOx, CO, O3 as well as aerosols, particulate matter including PM2.5 and PM10, smoke, dust, etc.
•	Flux estimation by forward and inverse methods, atmospheric transport modeling, derived air quality indicators and exposure-related metrics.
•	Automated plume detection and segmentation, enhanced background estimation for HSI/related data products for anomaly detection for a given scene, analyzing plume morphology and dynamics.
•	AI/ML models for atmospheric retrieval, plume detection, segmentation, physics informed machine learning, multi-modal and multi-platform learning including multi-satellite, MSI/HSI, thermal and UAV data products and ground-based spectrometers.
•	Sensor systems and emerging missions for atmospheric composition and emissions, small payloads and novel atmospheric sensing concepts.
•	Case studies including industrial and power sector emissions, emissions due to wildfires, volcanoes, transportation, mining, determining emission hotspots, etc.
•	Decision-support systems towards urban monitoring, climate change monitoring, open atmospheric data products and real-time verification systems.

Every year, millions of people around the world are affected by natural and man-made disasters. Floods, heat waves, droughts, wildfires, tropical cyclones and tornadoes are causing increasing damage. In addition, civil wars and regional conflicts in different parts of the world lead to growing numbers of refugees and major changes in population dynamics. Emergency responders and aid organizations need timely, comprehensive and accurate information about the extent of hazards, exposed assets and damage in order to respond quickly and effectively.

For decades, emergency mapping has used remotely sensed data to support rescue operations. However, providing information in a rapid, scalable and reliable manner remains a major challenge, as it often relies on manual interpretation. Recent advances in machine learning have opened up new possibilities for automating the analysis of remote sensing data to cope with the increasing volume and complexity of data, as well as the inherent spatio-temporal dynamics of disaster situations.
In this session, we aim to provide a platform for research groups to present their latest research activities aimed at addressing the problem of automatic, rapid, large-scale and accurate information retrieval from remotely sensed data to support disaster response. More specifically, the focus lies on:

- Deep learning-based approaches for extracting relevant information on hazard extent (e.g. flood mapping), exposed assets (e.g. road condition assessment) and impacts (e.g. building damage assessment) acquired from satellite, airborne or drone sensors. 
- On-board deep learning-based approaches for infrastructure or damage detection from airborne platforms.
- Challenges and limitations of current AI methods in the context of disaster response. 
- New public benchmark datasets for disaster response, for the training, evaluation, and comparison of AI models.

This session explores state-of-the-art research, while the complementary application-oriented session, “From Research to Action, Part II: Earth Observation for Operational Crisis and Disaster Management and Response” focuses on translating research advances into practical solutions for operational crisis and disaster management.

Natural hazards and armed conflicts cause widespread damage, disrupt livelihoods, and trigger displacement at local, regional and global scales. A key challenge for effective disaster response and recovery measures is to obtain a comprehensive and up-to-date situational picture within a very short time. Emergency responders and humanitarian organizations need to quickly identify affected areas, damaged or vulnerable buildings and infrastructure, and emerging needs. Earth observation and other geospatial data have long supported this task by enabling large-scale mapping and the monitoring of changes over time.
For many years, methods, processes, and information products based on Earth Observation and geospatial data have been developed and tested in close collaboration with emergency responders and humanitarian organizations. Recently, AI-based methods for the automatic, rapid, large-scale, and accurate extraction of information from remotely sensed data have advanced enormously. An overview of the latest innovations in this field will be presented in the session “From Research to Action Part I: Machine Learning and Remote Sensing Data for Rapid Disaster Response.”
However, translating these new methods and the information derived from diverse data sources into operational practice remains a major challenge. The relevant information has to be available for different stakeholder groups in short time, at an appropriate spatial scale and in a suitable format. Joint development and testing with operational users are therefore essential to bridge the gap between research and practical application. These efforts can contribute to improved preparedness, situational awareness, decision-making, and coordination of emergency and humanitarian response.
In this session, we invite research groups to present their latest research activities and developments in the field of practical demonstrations, operational applications, services, and products across the disaster management cycle. Contributions may address new analytical approaches with a focus on methods for supporting the coordination of rapid mapping, impact assessment, infrastructure and accessibility monitoring or the assessment of exposure and vulnerability. In addition, participants are invited to present application-oriented products and solutions, such as multimodal situational pictures, monitoring tools, and dashboards. We specifically invite contributions that demonstrate how Earth observation can be transferred into practical and operational use. Experiences and challenges related to working with users, testing and demonstrating solutions, and sustainably integrating research results into operational processes are particularly welcome.

Earth-observing systems now provide unprecedented information on precipitation, soil moisture, snow, evapotranspiration, surface water, groundwater, vegetation, topography, and hydrologic extremes. At the same time, advances in multimodal data fusion, large-scale computing, artificial intelligence, foundation models, and next-generation hydrologic modeling are changing how these observations can be combined and interpreted.
The central challenge is no longer simply to acquire more sensors, more observations, or more data. It is to transform heterogeneous observations into hydrologic intelligence: physically meaningful, uncertainty-aware, transferable, and actionable understanding of water-cycle states, fluxes, processes, extremes, and interactions across scales.
This Community-Contributed Theme will bring together remote sensing scientists, hydrologists, AI researchers, Earth-system modelers, and data scientists to explore the transition from sensors to hydrologic intelligence. The theme emphasizes approaches that move beyond individual retrievals or isolated variables toward integrated inference, understanding, modeling, and prediction of the water cycle. We invite contributions on multisensor and multimodal observations of water-cycle states and fluxes; integration of satellite, radar, in situ, reanalysis, and model information; remote-sensing-driven hydrologic modeling; AI–physics integration and foundation models for hydrologic inference and prediction; uncertainty, observability, scale mismatch, and cross-variable consistency; and Earth-observation-enabled prediction of floods, droughts, and other hydrologic extremes.
Remote sensing has greatly expanded our ability to observe the global water cycle, yet important gaps remain between observing individual variables and understanding hydrologic systems as a whole. Uncertainty propagation, scale mismatch, cross-variable inconsistency, and limited observability of key processes remain major barriers. Recent advances in complementary satellite missions, long-term Earth-observation records, multimodal AI, and hybrid modeling make this an especially timely opportunity to address these challenges.
This CCT complements existing IGARSS themes such as T.18 Multimodal Data Fusion, SP.3 Agentic AI for Earth Observation, SP.5 Foundation Models and Embeddings for Earth Observation, G.1 Earth System Modeling, M.1 Precipitation and Clouds, and L.10 Inland Waters. While these themes primarily focus on specific methods, technologies, Earth-system components, or application domains, the proposed CCT is organized around a broader hydrologic question: how Earth observations can be translated into coherent understanding and predictive capability across the water cycle. It therefore provides a cross-cutting forum linking remote sensing, hydrology, AI, and Earth-system science without duplicating any single existing General or Special Theme.

This session explores the paradigm shift brought by generative artificial intelligence (GenAI) within the Earth Observation (EO) domain. Moving beyond traditional discriminative machine learning, the session focuses on generative architectures (e.g. diffusion models, flow matching, and multi-modal foundation models) which can be leveraged for a wide range of remote sensing tasks, including data synthesis, super-resolution, cloud removal, and rare event simulation. Furthermore, the session will explore both theoretical model development and operational deployment, showcasing real-world applications in disaster response, agriculture, climate monitoring, and urban planning.

Earth Observation data is increasingly high-resolution, multi-modal, and voluminous, yet persistent challenges remain regarding data scarcity and availability, obscured observations (such as cloud cover), and the simulation of unobserved future climate or crisis scenarios. Generative AI offers powerful capabilities to address these bottlenecks by synthesizing physically consistent geospatial data, enhancing data fidelity, and enabling zero-shot generalization across diverse geographic regions. By examining both methodological advances and practical constraints such as uncertainty quantification, computational overhead, and hallucination mitigation, this session showcases new developments shaping the field.

This session bridges several key tracks of the IGARSS 2027 technical program. Methodologically, it offers a complementing look to SP.5 (Foundation Models and Embeddings for Earth Observation), SP.3 (Agentic AI for Earth Observation), and A.2 (Foundation Models in Geoscience) by exploring how generative models expand beyond fixed feature representations to data synthesis. Furthermore, it can be linked with SP.6 (Datasets, Benchmarks, and Evaluation Methodologies) by addressing crucial evaluation metrics for synthetic geospatial data. Going further than foundation models, this session provides a venue to showcase cutting edge works on Generative AI across diverse applications.

The field of Geometric Calibration/Validation of Optical EO Satellite Products is still somehow fragmented and mission-specific in the community of practice. Independent Validation of global performances are often lacking good references worldwide at the accuracy level requested by a given mission. Several activities are run and led by International Agencies such as ESA, NASA, USGS, Geoscience Australia,.. with activities like the AGRI dataset, GenCP, the NAIP dataset, Sentinel-2 GRI, collection of aerial/in-situ GCPs; at CEOS-level, several activities have also started, such as the GCPIX group and the GCP Viewer. While these activities are helpful to the community, the fragmentation and lack of standard, worldwide, approaches still remains. The Theme of this CCT is therefore very important to address such important topic towards a potential federation of resources worldwide to eventually enable both validation and calibration at different resolution levels.

Climate change is intensifying pressures on water resources through shifting rainfall patterns, prolonged droughts, flooding, and deteriorating water quality, with implications for human health and ecosystem resilience. Addressing these interconnected challenges requires approaches that identify where risks arise, who is most exposed, and which interventions are appropriate for local environmental and social conditions. This proposed study explores the integration of geospatial techniques and nature-based solutions to support climate-resilient water security and reduce water-related health risks.

The scope encompasses the use of satellite remote sensing, geographic information systems, and spatial analysis to assess changes in land use, surface water availability, vegetation cover, catchment conditions, and exposure to climate-related hazards. These observations will be integrated with available water-quality measurements, demographic information, and health indicators to identify vulnerable communities and priority intervention areas. A spatial assessment framework will examine the suitability of measures such as wetland restoration, riparian buffers, watershed revegetation, and urban green infrastructure. Intervention priorities will consider ecological suitability, potential benefits for water availability and quality, community vulnerability, and implementation feasibility. Field observations and stakeholder knowledge will support validation and interpretation, recognising that satellite-derived indicators cannot independently establish drinking-water safety or health outcomes.

The importance of this approach lies in connecting environmental monitoring with practical, place-specific action. Nature-based solutions can contribute to runoff regulation, pollutant retention, groundwater recharge, and ecosystem restoration, while their effectiveness depends on local hydrology, maintenance, and governance. Geospatial assessment provides a basis for targeting these interventions, evaluating trade-offs, and monitoring outcomes over time. The study complements General themes concerning climate adaptation, environmental sustainability, public health, and sustainable development by examining their shared dependence on resilient water systems. It also contributes to Special themes focused on remote sensing, geospatial modelling, ecosystem restoration, and nature-based interventions through an integrated application framework. The rationale for bringing these themes together is to bridge broad sustainability objectives with spatially explicit evidence and locally relevant implementation. The anticipated outcome is a decision-support framework that helps planners and communities prioritise interventions with potential benefits for water security, ecosystem health, and human well-being under a changing climate.

Artificial intelligence (AI) applied to satellite remote sensing offers substantial opportunities for real-time, large-scale environmental monitoring. Multispectral and hyperspectral imagery, complemented by other sensing modalities, can support greenhouse gas emission detection, inland water quality assessment, soil characterization, pollution tracking, and monitoring of floods, wildfires, and volcanic ash, among others. These applications address urgent societal needs, including – but not limited to – disaster preparedness, ecosystem protection, climate change mitigation, and public health. AI onboard satellites and on the ground can accelerate the transformation of observations into actions.

However, reliable deployment requires overcoming challenges shared across environmental applications. Ground-truth observations are often scarce, unevenly distributed, or insufficiently representative of changing environmental conditions. Variations across sensors, locations, and observation periods further affect model generalization. Additionally, rapidly evolving hazards demand timely inference and robust validation under conditions that may differ from those represented in training data.

This Community Contributed Session brings together researchers and practitioners developing, validating, and deploying AI for environmental monitoring, with particular emphasis on multispectral and hyperspectral satellite data. Its focus is the transition from algorithm development to reliable environmental information under limited supervision and variable observation conditions. It encourages exchanges across application communities to identify transferable methods, shared data challenges, and practical evaluation strategies.

The proposed theme complements the IGARSS 2027 General Themes by connecting AI methodology with land and atmospheric applications, remote sensing techniques, and societal decision-making. It complements Special Themes SP.2, SP.5, and SP.6 via application-driven research on environmental hazards, adaptation of foundation models, and representative datasets and evaluation. Its distinctive contribution is an integrated environmental monitoring perspective linking data generation, learning strategies, validation, and deployment across multiple environmental problems. This creates a common forum for methodological advances and operational requirements that span individual technical and application themes.

Submissions are welcome on (but not limited to):

•	Classical and AI-based methods for environmental monitoring,
•	Synthetic training data and data-level digital twins,
•	Few-shot, zero-shot, and weakly supervised learning,
•	Learning from large-scale unlabeled Earth observation archives,
•	Foundation model adaptation for environmental applications,
•	Multimodal satellite data fusion and temporal analysis,
•	Greenhouse gas monitoring, including methane plume detection and segmentation,
•	Flood, wildfire, burned area, and volcanic ash monitoring,
•	Bare soil detection, soil characterization, and inland water quality assessment,
•	Air quality monitoring and detection of industrial pollution,
•	Validation, representativeness, and generalization across environmental conditions,
•	Addressing domain/sensor gap in environmental monitoring,
•	Explainability and trustworthiness of AI in environmental monitoring,
•	Onboard and ground-based deployment, with demonstrated scientific, industrial, or societal impact.

By connecting AI development with environmental monitoring needs, the theme will advance reliable approaches that support timely and informed action.

Generative artificial intelligence is creating new opportunities to transform Earth observation and geospatial data into timely, accessible, and actionable disaster intelligence. Large language models (LLMs), multimodal generative models, retrieval-augmented generation (RAG), and agentic workflows can integrate satellite and UAV imagery, geospatial databases, sensor observations, scientific knowledge, and textual information to support hazard monitoring, impact assessment, situational awareness, emergency response, and recovery planning. However, their use in high-consequence disaster contexts raises critical challenges related to factual grounding, hallucination, uncertainty, provenance, explainability, and reliability. A key challenge is ensuring that AI-generated claims and interpretations can be verified against authoritative geospatial information and underlying observational evidence.

This Community Contributed Theme focuses on grounded and trustworthy generative AI, where generated interpretations are anchored in verifiable Earth observation evidence, authoritative geospatial information, and domain knowledge. It invites contributions on evidence-grounded RAG for Earth observation, multimodal geospatial reasoning, knowledge-enhanced LLMs, human-supervised agentic workflows, hallucination detection and mitigation, uncertainty communication, evidence provenance and traceability, explainable generative AI, and methods for evaluating the factual and scientific reliability of AI-generated geospatial information. Applications may address bushfires, floods, earthquakes, landslides, volcanic events, severe storms, and other rapidly evolving hazards.

The theme brings together remote sensing, geospatial science, artificial intelligence, disaster-risk management, and human-centred decision support around a fundamental emerging challenge: ensuring that generative AI outputs are scientifically defensible, traceable, and appropriately reliable for disaster decision-making.

This CCT complements existing IGARSS 2027 themes by addressing a distinct challenge at their intersection. SP.2 addresses monitoring rapidly changing and hazardous environments, SP.3 focuses on Agentic AI for Earth Observation, SP.5 on foundation models and embeddings, and SP.6 on datasets, benchmarks, and evaluation methodologies. Rather than focusing on a particular AI architecture, autonomous agent, or hazard-monitoring technique, this CCT centres on the grounding, verification, and traceability of AI-generated disaster intelligence derived from heterogeneous Earth observation and geospatial evidence.

Earth Observation (EO) is characterized by huge data volumes, increasing complex sensing modalities, and the rapid adoption of data-driven methods. Modern satellite missions, constellations, and multi-source observation systems continuously generate massive datasets whose effective exploitation requires not only advanced algorithms, but also new computing strategies capable of meeting growing demands in scalability, latency, energy efficiency, and operational deployment.
Artificial Intelligence (AI), including Deep Learning (DL) and foundation models, is becoming central to many EO applications. However, the increasing size and complexity of these models impose substantial computational requirements and motivate the use of High-Performance Computing (HPC), distributed and heterogeneous systems, and specialized hardware accelerators. At the same time, emerging deployment scenarios such as edge computing and in-orbit processing are shifting part of the computation closer to the sensors, introducing new challenges related to limited resources, energy consumption, model compression, and hardware-aware optimization.
This Community Contributed Session on “Computing Technologies for AI in Earth Observation” aims to bring together recent advances at the intersection of EO and computing. We invite contributions on novel methods, architectures, algorithms, and software tools that improve the efficiency, scalability, and deployment of EO processing workflows. Topics of interest include, but are not limited to, parallel and distributed computing, HPC, GPUs, FPGAs and other accelerators, edge and in-orbit AI, energy-efficient and hardware-aware computing, model optimization and compression, neuromorphic and unconventional computing paradigms, and emerging computing technologies supporting next-generation EO applications.

Related General or Special Theme: 
This proposal aligns closely with the IGARSS 2027 Special Theme: SP.4 Computing Paradigms and Quantum Technologies for Earth Observation and with the General Themes: AI and Big Data (A) by addressing the development and deployment of advanced AI, deep learning, and high-performance computing methods, as well as Applications (C, L, M, O, P) through its focus on enabling scalable and efficient EO data processing for a wide range of Earth science applications.

This session focuses on research and developments in an important topic in active optical remote sensing: Space Lidar. Lidar's unique capability to observe a diverse variety of geophysical phenomena from orbit around the Earth and planets has stimulated new areas of remote sensing research that now attracts the attention of scientists and engineers worldwide. With a number of instruments already operational or pending launches within the coming years, many of their original technological issues have been resolved, still the long-term reliability of key active components and the survival on harsh space environment requires additional efforts and investments. Some of the key topics for this session are:

• Space Agencies (NASA, ESA, JAXA, CNES etc.) on-orbit missions, future missions, technical challenges, observations, and science products
• Continuation of work in the domain of long-lived/ high power UV/visible/infrared lasers and optics, especially contamination and optical damage
• Research to improve the reliability of lasers/diode lasers and high-power optics operated in vacuum 
• Space-qualification of tunable lasers and optics to support trace gas lidars operating in the 1-5 µm region 
• Space-qualification of higher efficiency lasers such fiber lasers and amplifiers. Radiation hardening is an area of particular concern 
• General power scaling of space-qualified lasers with a focus on improved efficiency and thermal tolerances 
• Improved high gain, low dark noise and low NEP space qualified/qualifiable array/detector at all wavelengths, and in particular in IR bandwidth 

This session will focus on science and applications addressed by space-based lidar, as well as on techniques and supporting technology. Our goal is that this session is to provide a stimulating forum where members of the international lidar and related technology communities can present and discuss results, trends, and future directions.

This session focuses on research and developments in an important topic in active optical remote sensing: Space Lidar. Lidar's unique capability to observe a diverse variety of geophysical phenomena from orbit around the Earth and planets has stimulated new areas of remote sensing research that now attracts the attention of scientists and engineers worldwide. With a number of instruments already operational or pending launches within the coming years, many of their original technological issues have been resolved, still the long-term reliability of key active components and the survival on harsh space environment requires additional efforts and investments. Some of the key topics for this session are:

• Space Agencies (NASA, ESA, JAXA, CNES etc.) on-orbit missions, future missions, technical challenges, observations, and science products
• Continuation of work in the domain of long-lived/ high power UV/visible/infrared lasers and optics, especially contamination and optical damage
• Research to improve the reliability of lasers/diode lasers and high-power optics operated in vacuum 
• Space-qualification of tunable lasers and optics to support trace gas lidars operating in the 1-5 µm region 
• Space-qualification of higher efficiency lasers such fiber lasers and amplifiers. Radiation hardening is an area of particular concern 
• General power scaling of space-qualified lasers with a focus on improved efficiency and thermal tolerances 
• Improved high gain, low dark noise and low NEP space qualified/qualifiable array/detector at all wavelengths, and in particular in IR bandwidth 

This session focuses on research and developments in an important topic related to Quantum sensors and sensing technology for earth observation, a unique capability to observe a diverse variety of geophysical phenomena from orbit around the Earth. This emerging capability has stimulated new areas of remote sensing research that now attracts the attention of scientists and engineers worldwide.

Quantum Sensing is the most exciting quantum technology, and it has the most potential to change our lives, in terms of societal benefits, in the next decade and beyond. Quantum sensing uses quantum properties to achieve unprecedented measurement sensitivity and performance, including quantum-enhanced methodologies that outperform their classical counterparts. Quantum sensors are highly relevant to overlapping areas such as precision navigation and timing; electromagnetic field sensing; attitude control; communications; and gravimetry. Typical quantum sensors exploit techniques such as atomic systems, matter waves, quantum entanglement, quantum superposition of states, quantum illumination methods, and manipulation of photons and atoms, in general. Significant gains include technologies important for a range of space-based remote sensing, in situ measurements, metrology, interferometry, quantum communication, ranging, imaging, radar and lidar receivers, and gravity measurements.

This community contributed session will focus on quantum sensing techniques, technologies and geoscience and remote sensing applications related to earth science, and will extend to astrophysics, and planetary Science, as well.

This session focuses on research and developments in an important topic related to Quantum sensors and sensing technology for earth observation, a unique capability to observe a diverse variety of geophysical phenomena from orbit around the Earth. This emerging capability has stimulated new areas of remote sensing research that now attracts the attention of scientists and engineers worldwide.

Quantum Sensing is the most exciting quantum technology, and it has the most potential to change our lives, in terms of societal benefits, in the next decade and beyond. Quantum sensing uses quantum properties to achieve unprecedented measurement sensitivity and performance, including quantum-enhanced methodologies that outperform their classical counterparts. Quantum sensors are highly relevant to overlapping areas such as precision navigation and timing; electromagnetic field sensing; attitude control; communications; and gravimetry. Typical quantum sensors exploit techniques such as atomic systems, matter waves, quantum entanglement, quantum superposition of states, quantum illumination methods, and manipulation of photons and atoms, in general. Significant gains include technologies important for a range of space-based remote sensing, in situ measurements, metrology, interferometry, quantum communication, ranging, imaging, radar and lidar receivers, and gravity measurements.

The IEEE GRSS Data Fusion Contest (DFC), organized annually by the Image Analysis and Data Fusion Technical Committee (IADF TC), promotes advances in remote sensing data fusion and information extraction through an open competition based on freely available, multi-source datasets. The contest provides a common experimental framework for evaluating novel algorithms and encourages the development of effective approaches to challenging remote sensing problems.
This special session showcases the outstanding research contributions emerging from the 2027 IEEE GRSS Data Fusion Contest. It will feature presentations of the highest-ranked submissions, highlighting novel methodologies, experimental findings, and advances in remote sensing image analysis and data fusion. In addition, the contest organizers will provide an overview of the competition, summarizing its outcomes and highlighting key findings and lessons learned. 
By bringing together the leading contributions to a common challenge, the session complements the general and special themes of IGARSS 2027 with a focused perspective on comparative evaluation, methodological innovation, and reproducible research in remote sensing. The shared datasets and evaluation framework provide a basis for assessing different approaches under consistent conditions, facilitating meaningful comparisons and fostering the exchange of ideas within the remote sensing community. 

Earth observation (EO) plays a fundamental role in understanding and monitoring our planet, supporting applications ranging from mapping and environmental monitoring to urban development, disaster risk management, and climate change assessment. Advances in EO technologies have significantly increased the diversity, volume, and resolution of available data, including optical multispectral and hyperspectral imagery, synthetic aperture radar (SAR), LiDAR point clouds, and multi-temporal observations. These data are increasingly complemented by other information modalities, including natural language and textual descriptions, crowdsourced imagery, oblique photographs, in situ measurements, and georeferenced information from social networks, creating unprecedented opportunities for comprehensive and semantically enriched Earth system analysis.
The integration and exploitation of these heterogeneous data sources, however, present substantial methodological and computational challenges. Differences in spatial, spectral, and temporal resolutions, data quality, acquisition geometries, and sensing principles require innovative approaches to image analysis, multimodal learning, and information fusion. Recent advances in artificial intelligence (AI), particularly deep learning, self-supervised learning, foundation models, and vision-language models (VLMs), are transforming the way EO data are processed, interpreted, and integrated. By connecting visual and geospatial information with natural language, these approaches enable new capabilities in semantic understanding, cross-modal retrieval, visual question answering, and language-guided EO analysis. At the same time, important open challenges remain concerning model generalization and transferability, data efficiency, uncertainty quantification, explainability, robustness, and the trustworthy deployment of AI-based solutions in real-world applications.
The Image Analysis and Data Fusion Technical Committee (IADF-TC) of the IEEE Geoscience and Remote Sensing Society promotes scientific exchange, collaboration, education, and best practices in these rapidly evolving research areas. Building on more than a decade of successful community-contributed sessions at IGARSS, the proposed session provides a dedicated forum for researchers and practitioners from academia, industry, and government to present recent advances, exchange experiences, and discuss emerging research directions.
The session welcomes contributions spanning the full EO data processing and analysis pipeline, including data preprocessing and harmonization (pansharpening, co-registration, geocoding), advanced image analysis and representation learning, multimodal and multi-temporal data fusion, vision-language models, multimodal foundation models, and generative AI, as well as model evaluation, uncertainty quantification, and explainable and trustworthy AI. Particular attention is given to methodological innovations, benchmark datasets, reproducibility, and the transition from research developments to operational applications.
Complementing the broader IGARSS General and Special Themes, this session offers a cross-cutting perspective that connects fundamental methodological advances with diverse EO applications. Its established interdisciplinary community fosters dialogue across research domains, encourages contributions from early-career scientists, and helps identify emerging challenges and opportunities shaping the future of image analysis and data fusion in the AI era.

Scope: Traditional deep learning paradigms in Earth Observation (EO) rely predominantly on fixed-category taxonomies and rigid target classes. This limits their adaptability to complex, open-world dynamics. The addition of Natural Language presents a transformative shift by leveraging multi-modal vision-language alignment, cross-modal embeddings, and prompt-based tuning to bridge pixel-level imagery with textual domain knowledge. This session explores state-of-the-art methodologies integrating natural language guidance into satellite, aerial, and hyperspectral image analysis. The scope encompasses the following but are not limited to:
•	Vision-Language Models (VLMs) customized for geospatial including climatic, agricultural, environmental, meteorological data analysis
•	Zero-shot and open-vocabulary land-cover classification, change, and target detection algorithms 
•	Text-guided semantic segmentation and multi-modal information retrieval approaches 
•	Parameter-efficient prompt adaptation and test-time adaptation strategies
•	Joint text-image embedding spaces designed for heterogeneous EO data modalities

Importance: Earth Observation imagery is intrinsically multi-scale, multi-spectral, and geographically diverse. Annotating dense geospatial datasets with custom class hierarchies is prohibitively expensive, time-intensive, and inherently inflexible. It is seen that natural language supervision bypasses the bottleneck of closed-set annotations of geospatial data by embedding rich semantic descriptors directly into representation learning pipelines. It is proved to be efficient for 2-dimensional natural images. This session attempts to explore the conditioning of visual representations on flexible textual prompts that enable models to generalize across unseen geographic regions, novel land-use categories, and evolving atmospheric or lighting conditions without heavy fine-tuning. Unlocking open-vocabulary reasoning across satellite imagery is paramount for rapid disaster response, environmental change monitoring, and scaling global Earth monitoring pipelines.

Complementarity to General and Special Themes: This session acts as a focused, multi-modal extension (vision, language, and others) to the IGARSS conference themes, directly integrating and bridging SP.5, A.2, and A.3:
•	Complementing SP.5 (Foundation Models and Embeddings for Earth Observation): While SP.5 addresses foundational representation architectures broadly, this session concentrates on the critical interaction between visual representations and text-based embeddings. It provides a targeted exploration into how linguistic conditioning and multimodal embedding spaces refine geospatial foundation models to varying tasks.
•	Complementing A.2 (Foundation Models in Geoscience): A.2 examines large-scale pre-training across Earth science domains. Our session directly complements this theme by analyzing the mechanism of natural language guidance within these foundation models, evaluating how prompt learning, test-time adaptation, and vision-language alignment enhance model generalizability across multi-sensor EO data.
•	Complementing A.3 (Large-scale Machine Learning): Large-scale machine learning emphasizes scalable computational frameworks and data efficiency. By leveraging zero-shot transfer and parameter-efficient prompt tuning, language-supervised methods dramatically reduce the need for massive labeled downstream datasets, directly addressing A.3's objective of scalable, data-efficient AI deployment.

Water resources are increasingly affected by environmental contamination, changing hydroclimatic conditions, land-use transitions, and growing demands for sustainable water management. Understanding these challenges requires integrated approaches that connect water-resources assessment and environmental modelling with advanced water-quality analysis and human-health risk characterization. Earth observation, GIS, in-situ monitoring, and artificial intelligence (AI) provide opportunities to integrate environmental information and develop spatially explicit assessments of water-system conditions and associated risks.
The proposed session aims to bring together research on water resources engineering, hydroinformatics, environmental modelling, AI-driven water-quality assessment, geospatial analysis, and human-health risk assessment. It will provide a platform for investigating how water-quality conditions, hydrological processes, environmental contamination, and human exposure can be assessed within an integrated geospatial framework. Contributions combining satellite observations, ground-based water-quality measurements, hydroclimatic data, land-use and land-cover information, and water-resources models are particularly encouraged.
The scope includes surface-water and groundwater quality assessment, water-quality indices, AI-based environmental modelling, hydrological and water-quality interactions, contaminant source identification, geospatial pollution mapping, environmental exposure assessment, and human-health risk characterization. Studies examining the influence of climate variability and land-use change on water quality, the identification of vulnerable locations and populations, and the integration of environmental and health-risk indicators into water-resource planning are also welcome. Machine learning, explainable AI, spatial statistics, uncertainty analysis, and multisource data integration may be used to improve the reliability and interpretability of these assessments.
The distinctive contribution of this theme is the integration of water-resources modelling with AI-enabled water-quality assessment and geospatial human-health risk evaluation. Rather than addressing water management, water quality, and health risks as separate topics, the session will encourage studies that connect water-system processes with contamination patterns and potential human-health implications. It will complement existing IGARSS themes on inland waters, water quality, and machine learning by emphasizing interdisciplinary modelling frameworks that translate Earth observation and environmental data into actionable information for water security, pollution management, and public-health protection.

KOMPSAT-8 is a next-generation Synthetic Aperture Radar (SAR) Earth-observation mission under development in the Republic of Korea, targeting 0.25 m-class ultra-high-resolution imaging and a distributed architecture composed of one main SAR satellite and three companion satellites. Beyond advanced monostatic imaging, the mission is being developed to enable bistatic and multistatic observations, formation-flying operations, interferometric measurements, three-dimensional reconstruction, and new applications based on spatially separated observations.
This proposal is Part I of a coordinated two-part Community-Contributed Theme (CCT) on “KOMPSAT-8 and Next-Generation Distributed Multistatic SAR.” Part I focuses on the mission, spacecraft, and payload technologies required to realize distributed spaceborne SAR. The proposed contributions cover KOMPSAT-8 system architecture and development status, formation-flying flight dynamics, high-resolution wide-swath (HRWS) SAR payload design, wideband active electronically scanned array antenna design, and high-power electrical-system design. Together, these topics address the tightly coupled engineering challenges that determine observation geometry, imaging capability, synchronization feasibility, payload performance, and operational robustness.
The companion Part II proposal, “KOMPSAT-8 and Next-Generation Distributed Multistatic SAR II: Processing, Interferometry, and Applications,” addresses end-to-end multistatic simulation, calibration, signal processing, interferometric 3-D reconstruction, and environmental applications. The two parts are intentionally complementary: Part I establishes how a distributed SAR mission is designed and implemented, while Part II addresses how its observations are simulated, calibrated, processed, and transformed into geophysical information.
This theme is timely as spaceborne SAR evolves from conventional single-platform architectures toward coordinated and distributed sensing. It complements existing IGARSS themes on Spaceborne SAR Missions, SAR Instruments, Multi-Channel DBF Imaging Techniques, Bistatic and Multistatic SAR, and Advanced Future Instrument Concepts by integrating system engineering, formation flying, payload, antenna, and spacecraft power technologies within a coherent distributed-SAR framework. KOMPSAT-8 serves as the central reference architecture, while related contributions on other distributed and formation-flying SAR systems are welcome.

KOMPSAT-8 is a next-generation Synthetic Aperture Radar (SAR) Earth-observation mission under development in the Republic of Korea, targeting 0.25 m-class ultra-high-resolution imaging and a distributed architecture composed of one main SAR satellite and three companion satellites. The configuration is intended to support monostatic, bistatic, and multistatic observations and to provide multiple spatial baselines for advanced imaging, interferometry, three-dimensional reconstruction, and application development.
This proposal is Part II of a coordinated two-part Community-Contributed Theme (CCT) on “KOMPSAT-8 and Next-Generation Distributed Multistatic SAR.” Part II focuses on the processing chain that converts distributed SAR observations into calibrated imagery and higher-level information. The proposed contributions cover end-to-end multistatic SAR simulation and performance assessment, calibration strategies for distributed observations, HRWS SAR signal processing and image formation, interferometric geometry and 3-D reconstruction, and forest height and biomass retrieval using multistatic observations. These topics address key challenges created by spatially separated platforms, including geometry modeling, timing and phase consistency, channel calibration, multichannel image formation, baseline-dependent interferometric sensitivity, and exploitation of multiple simultaneous observations.
The companion Part I proposal, “KOMPSAT-8 and Next-Generation Distributed Multistatic SAR I: System Architecture and Enabling Technologies,” addresses system architecture, formation-flying flight dynamics, HRWS payload design, SAR antenna technology, and high-power electrical systems. The two parts form a coherent mission-to-application sequence: Part I establishes the distributed sensing system and its enabling technologies, while Part II follows the data from acquisition and simulation through calibration and processing to interferometric and geophysical products.
This theme complements existing IGARSS topics on Bistatic and Multistatic SAR, Interferometry, Spaceborne SAR Missions, and Multi-Channel DBF Imaging Techniques by connecting methods that are often discussed separately within a single end-to-end distributed-SAR processing framework. KOMPSAT-8 provides a concrete reference case, while submissions addressing other distributed, bistatic, multistatic, and formation-flying SAR systems are also encouraged to broaden international participation and methodological exchange.

Since forest structure is of great value to terrestrial ecology, habitat biodiversity, and global carbon storage assessments, it is desired to monitor and quantify the state of, and change in forest vertical structural profile, aboveground biomass and height along with other forest biophysical characteristics (e.g., LAI). It is important to generate large-scale (e.g., regional and global) moderate-resolution (e.g., few hectares down to sub-hectare) such products because these are useful not only on carbon storage accounting and carbon cycle dynamics modeling but also on supporting efforts aimed at quantifying biodiversity, particularly given the rapid declines and losses of many plant and animal species world-wide.

To address this scientific goal, the remote sensing community has been working towards combining multi-sensor measurements, such as Synthetic Aperture Radar (SAR) and lidar as well as optical sensors. For example, these include JAXA's ALOS/ALOS-2/ALOS-4 (single L-band SAR) and MOLI (lidar) missions, NASA'S NISAR (single L-band SAR) and GEDI (lidar), DLR's TanDEM-X (twin X-band SAR), ESA's BIOMASS (single P-band SAR) as well as China's Lutan-1 (bistatic L-band SAR) and TECIS (lidar). Commercial SAR satellites have also been emerging in recent years, such as the X-band SAR constellation from Capella Space, and the cartwheel multi-static X-band SAR constellation PIESAT-1 (or called Hongtu-1) from PIESAT Information Technology. Other spaceborne passive optical sensors including NASA’s Landsat and ESA’s Sentinel-2 have also been successfully combined with lidar missions for large-scale forest parameter mapping.

It is known that interferometric radar (and passive optical sensor) have complete spatial coverage and good spatial resolution with moderate accuracy in measuring vertical structure, where microwave radar has deeper penetration through the canopy than passive optical sensors. In contrast, spaceborne lidar has a sparse coverage with much better vertical confidence. Therefore, this proposal will demonstrate a few novel scientific algorithms to combine the complete spatial coverage of radar/optical and precise vertical measurements of lidar so that large-scale (potentially global-scale) maps of forest vertical structure, biophysical parameters (e.g. aboveground biomass/height) and their time-series changes can be generated, through the fusion of multiple spaceborne SAR, including (but not limited to) Interferometric SAR (InSAR), Polarimetric InSAR (PolInSAR) and Tomographic SAR (TomoSAR) data, as well as lidar and/or passive optical data.

Scope: Supervised deep learning models in Earth Observation (EO) rely heavily on vast amounts of high-quality, pixel-precise annotations. However, acquiring labels for satellite, aerial, and hyperspectral imagery is prohibitively expensive, labor-intensive, and requires specialized domain expertise; sometimes infeasible too. This session focuses on cutting-edge methodologies designed to reduce the annotation bottleneck across remote sensing tasks. The scope encompasses but are not limited to:
•	Weakly supervised, semi-supervised, and self-supervised learning for geospatial imagery
•	Point-, box-, and polygon-prompted segmentation architectures (e.g., adaptation of interactive segmentation models)
•	Test-time adaptation, domain adaptation, and few-shot learning under extreme label scarcity
•	Active learning, pseudo-labeling, and uncertainty-guided sample selection for low-resource mapping
•	Fine-tuning strategies for pre-trained geospatial backbone networks using minimal ground-truth data
Importance: Label scarcity represents one of the most critical operational bottlenecks in real-world geospatial applications. In time-sensitive scenarios, such as immediate post-disaster damage assessment, rapid flood monitoring, or tracking seasonal agricultural changes, waiting for manual pixel-level labeling is infeasible. Furthermore, regional distribution shifts, varying sensor resolutions, and dynamic atmospheric conditions often cause fully supervised models to fail when deployed in unseen environments. Developing data-efficient frameworks that achieve high precision with sparse, noisy, or minimal annotations is essential for scaling automated Earth observation pipelines globally and democratizing AI applications in low-resource geographic regions.
Complementarity to General and Special Themes: This session serves as a specialized, operational deep-dive that complements and grounds the broader theoretical scopes of SP.5 and A.2:
•	Complementing SP.5 (Foundation Models and Embeddings for Earth Observation): While SP.5 addresses the creation and pre-training of massive foundation models, this session explores their downstream utility under real-world low-data constraints. Specifically, it investigates how self-supervised, few-shot, active learning algorithms meaningfully help to evolve the representations of foundational model embeddings that can effectively be fine-tuned or prompted using ultra-sparse labels (e.g., single-point prompts or image-level tags) to achieve dense prediction tasks without requiring fully annotated target datasets.
•	Complementing A.2 (Foundation Models in Geoscience): A.2 broadens the scope of pre-trained architectures across Earth science domains. Our session complements this theme by focusing on the practical challenges of deploying these models under label-scarce conditions. It provides a targeted forum for evaluating zero-shot, few-shot, and parameter-efficient transfer methods, demonstrating how geoscientific foundation models can maintain robust generalization despite severe data scarcity.

Point clouds acquired by LiDAR, photogrammetry, and emerging 3D computer vision systems provide detailed three-dimensional representations of built and natural environments. Advances in sensing technologies, reconstruction algorithms, and artificial intelligence are expanding their applications in remote sensing, environmental monitoring, and infrastructure management. This session aims to bring together researchers developing point cloud acquisition, processing, reconstruction, and analysis methods, as well as those applying these technologies to address real-world geospatial challenges.

Topics of interest include, but are not limited to:

· Point cloud processing and analysis: registration, denoising, segmentation, classification, semantic understanding, geometric feature extraction, change detection, and uncertainty quantification.

· Surveys, benchmarks, and datasets: terrestrial, airborne, mobile, and spaceborne LiDAR surveys; annotated point cloud datasets; multimodal benchmarks; and evaluation methodologies.

· 3D reconstruction and computer vision: point cloud generation from multi-view imagery, structure-from-motion, multi-view stereo, neural rendering, Gaussian splatting, and learning-based 3D reconstruction, as well as the integration of point clouds as prior in these systems.
· Multimodal data fusion: integration of point clouds with optical and multispectral imagery, hyperspectral data, synthetic aperture radar (SAR), thermal imagery, and other geospatial data sources.

· Built environment monitoring: urban digital twins, building reconstruction, infrastructure inspection, construction progress tracking, structural deformation monitoring, road mapping, and urban change detection.

· Natural environment monitoring: forest structure and biomass estimation, vegetation dynamics, habitat characterization, terrain modelling, landslide detection, coastal erosion, flood assessment, and wildfire fuel mapping.

· Emerging systems and applications: foundation model for 3D reconstruction, Multimodal LLM systems for point cloud processing, AI for real-time 3D mapping, AI for robotics-assisted surveying, AI-based autonomous mobile mapping, and AI-driven geospatial analysis of point clouds.

Contributions addressing methodological advances, sensor integration, benchmark development, and operational applications are welcome. The session seeks to foster interdisciplinary collaboration across remote sensing, photogrammetry, computer vision, and geospatial artificial intelligence, advancing point cloud technologies for monitoring dynamic built and natural environments.

Polar orbiting environmental satellites in Low Earth Orbit (LEO) are critical for global monitoring of Earth and its environment. With three operational satellites in orbit, NOAA’s Joint Polar Satellite System is providing resilient and reliable Earth observations for operational meteorology and other mission critical applications. Two more satellites in the series are under development and the constellation is expected to provide backbone observations that support both short- and long-term weather forecast models well into the next decade. In collaboration with partner agencies such as NASA, ESA, EUMETSAT and JAXA, NOAA provides its stakeholders a large set of global observations in LEO. Additionally, there is a strong emerging commercial marketplace with several private sector companies providing new data that has the potential for operational use in Earth science and meteorological applications. The synergy between NOAA,  its partner missions and commercial vendors offers significant benefits to users such as improved global refresh of observations and complementarity of measurements in multiple regions of the electromagnetic spectrum to enable detailed monitoring of the Earth. In addition to providing timely and critical observations for extreme weather and disasters, the long historic data record from LEO satellites is also critical for climate reanalysis that is vital for training artificial intelligence and machine learning driven forecast models. This session invites presentations from international agencies to present the status of their current and future LEO missions that make routine observations to monitor the Earth and its environment for applications such as NWP that support decision makers. Subjects covered in this session will include applications and accomplishments from current operational missions as well as plans for new missions from space agencies. Presentations that demonstrate the societal and economic value of LEO observations to applications that are integral to decision making are also of high interest to this session.

Monitoring water distribution and dynamics in all its forms — from polar ice sheets and soil moisture to wetlands and flood events — is fundamental to understanding climate change and mitigating its effects. In this context, microwave remote sensing offers an unparalleled, all-weather, day-and-night monitoring capability. This proposed community-contributed theme builds upon these strengths by creating a focused forum for water monitoring. The session will cover the entire value chain: from the electromagnetic modeling that links sensor measurements to geophysical parameters; to observation techniques, with synergies and complementarities among different sensors; all the way down to the applications that deliver societal, humanitarian, and commercial impact.  

In the broader context of the IGARSS program, this session is conceived to foster synergies between foundational topics in Microwave Modeling (T20) and key application areas like Soils and Soil Moisture (L.8), Wetlands (L.9), and Inland Waters (L.10). 

The strong community demand for such a theme was demonstrated by the success of the first MAMIWATA workshop, organized by the IEEE GRSS MIRS Technical Committee in 2023, in Pasadena (CA), USA. ​The community showed continued interest in this theme at the 2nd edition of the MAMIWATA initiative, in the form of a four-day course+workshop held in Rome, Italy, in November 2025, with more than 50 attendees. This traction was followed by the successful Microwave-Based Water Monitoring CCT at IGARSS 2026. ​​Two key outcomes from the MAMIWATA gatherings and the CCT session at IGARS 2026 were the consensus on the need for a unified platform to advance the physical modeling of scattering processes that involve water in different forms, and the need for building a diverse network to bridge the gap between sensor-focused communities. This session aims to be that bridge, encouraging the cross-pollination of ideas to address scientific challenges in a more unified, application-driven way.  

​​We solicit contributions across the full spectrum of water monitoring. This includes advancements in modeling, such as polarimetric and interferometric models, direct problems and inverse models, and radiative transfer theory. We also invite papers on innovative techniques, including new retrieval algorithms, comparative performance analyses between different sensors, and novel strategies for multi-sensor data fusion. ​​​We also welcome studies of air–water interface processes, including surface winds, roughness, wave conditions, and energy and moisture exchanges over inland waters.​   Finally, we strongly encourage submissions on impactful applications, such as operational flood mapping, wetland carbon cycle monitoring, agricultural water management, and cryosphere change assessment.    Finally, we strongly encourage submissions on impactful applications, such as operational flood mapping, wetland carbon cycle monitoring, agricultural water management, and cryosphere change assessment. ​ 

​​By bringing together experts from these interconnected domains, this theme will ​     ​​ strengthen connections across the modeling, sensing, and application communities within IGARSS​ by facilitating a deeper understanding of retrieval uncertainties and paving the way for more robust algorithms. This holistic approach is essential for translating advanced modeling and sensor technology into tangible solutions for today’s most pressing environmental challenges.​ 

Radar remote sensing has become one of the most capable tools for observing Earth and planetary environments, leading to significant progress in retrieving bio-geophysical properties across rough surfaces of soil and ocean, forests, snow and ice, glaciers, and even planetary subsurface layers.
A fundamental requirement for these applications is the capability to model the scattering behavior of natural media. Forward modeling based on physical scattering mechanisms enables robust inversion of geophysical parameters and helps evaluate the sensitivity of radar observations under different system configurations and environmental conditions. By providing a rigorous link between measurements and physical processes, scattering models improve retrieval accuracy and also guide the design of novel radar sensors and observational strategies.
At the same time, data-driven approaches, particularly machine learning (ML), have gained substantial traction in radar remote sensing. While ML enables large-scale data exploitation, it often suffers from limited interpretability and robustness when confronted with out-of-distribution data. Physics-based scattering models can complement ML approaches by embedding physical constraints, thus creating hybrid frameworks that enhance reliability and explainability. This session will highlight advances in both theory and applications of scattering models, inversion methods, and physics-ML synergies in radar remote sensing.
This CCT is organized by the Modeling in Radar Remote Sensing (MRRS) Working Group of the Modeling In Remote Sensing Technical Committee (MIRS TC) of the IEEE Geoscience and Remote Sensing Society (GRSS). This CCT aims to bring together researchers across radar modeling, inversion, and hybrid methods, covering diverse environments and applications from environmental monitoring and climate studies to disaster management and planetary exploration.

The last years have substantially widened the range of spaceborne microwave observations available for land applications. ESA’s BIOMASS mission has provided the first P-band SAR observations from space, designed for the retrieval of forest above-ground biomass and structure; the NASA-ISRO NISAR mission is acquiring L- and S-band SAR data with global coverage and a 12-day repeat cycle; and ESA’s HydroGNSS Scout is testing dual-frequency, dual-polarization GNSS reflectometry for soil moisture, freeze/thaw, wetlands and forest biomass. These data complement the long-term records of Sentinel-1, SMAP, SMOS, AMSR2/3 and CYGNSS and the observations of ALOS-4 and other SAR missions, while ROSE-L and CIMR are in preparation.
The availability of observations at frequencies from P- to Ka-band, from active, passive and bistatic sensors with different spatial resolutions and revisit times, offers new opportunities for Earth observation, but also raises challenges in separating the contributions of soil and vegetation to the measured signals, taking into account the specific sensitivities of each frequency and observation technique. Moreover, the transferability of retrieval techniques between missions and different environments needs to be assessed.
Addressing these aspects requires both electromagnetic modeling and data-driven analyses. Scattering and emission models, from radiative transfer to full-wave and bistatic formulations, are needed to interpret the observations and to train or constrain machine learning based algorithms, which in turn can improve the accuracy and computational efficiency of the retrievals and support the generation of high-resolution products. The combination of microwave data with optical and thermal observations can provide further information on vegetation status.
The session is open to contributions on forward modeling of soil and vegetation and on the retrieval of soil moisture, vegetation water content, above-ground biomass and crop parameters from single- and multi-frequency microwave data, with applications to agriculture, drought monitoring, forest carbon and high-latitude environments. Particularly welcome are advances in retrieval techniques, such as physics-informed and machine learning methods, calibration/validation activities and early results of recently launched missions (e.g., BIOMASS, NISAR and HydroGNSS), as well as studies in preparation of upcoming missions such as ROSE-L and CIMR. The aim is to bring together mission scientists, modelers and algorithm developers to discuss the results obtained with the new datasets and their integration with existing ones.

Accurate monitoring of soil moisture (SM) and vegetation water content (VWC) is essential for understanding soil conditions and soil–vegetation-atmosphere interactions, as well as for supporting applications such as drought assessment, agricultural water management, and land degradation monitoring. Synthetic Aperture Radar (SAR) observations represent key resources for resolving fine-scale heterogeneity of SM and VWC. However, robust retrieval of these variables at high spatial resolution (0.1–0.3 km) remains challenging, as complex scattering processes and pronounced sub-kilometer variability confound the radar signal. 

Remote sensing has made significant progress in delivering reliable SM estimates. Microwave-based SM products at coarse spatial resolutions (e.g., 10–40 km) are now operational, while higher resolution SM products (e.g., 0.1–0.3 km) derived from SAR observations are transitioning from the research phase toward systematic generation. To maximize their scientific and operational value, these products must be accompanied by reliable uncertainty estimates, which are essential for data assimilation and for the integration of independent observational datasets. Significant efforts have therefore been devoted to establishing validation protocols, identifying reference measurements (RMs), and addressing the spatial mismatch between satellite SM products and RMs, which are typically point-scale measurements from hydrologic networks. 

The joint retrieval of other surface variables, such as VWC, can further enhance the scientific and societal impact of high-resolution SAR-based land variables. VWC is commonly measured through destructive field sampling, which is accurate but labor-intensive and unsuitable for operational monitoring over large agricultural regions. Satellite observations enable systematic and large-scale VWC monitoring. VWC can be inferred from microwave measurements through the retrieval of Vegetation Optical Depth (VOD), which quantifies microwave attenuation within the vegetation canopy. However, VOD products derived from radiometers and scatterometers are currently available only at coarse spatial resolutions (typically 10-50 km), limiting their applicability at the field scale. 

The growing availability of C-band data from Sentinel-1 and L-band observations from missions such as NISAR, SAOCOM, and ALOS-4, together with the future ROSE-L mission, offers unprecedented opportunities to advance scientific knowledge and assess their impact on applications.

By addressing key methodological challenges, including uncertainty characterization and scalability, this session encourages contributions presenting recent advancements in high resolution SM and VWC retrieval from current and upcoming multi-frequency EO missions. The session also solicits contributions demonstrating the added value of these products for end users and operational applications, including irrigation detection and irrigation volume mapping, improved root-zone SM and evapotranspiration estimation through data assimilation, drought monitoring, crop stress detection, and crop yield prediction.

Topics of interest include:
•	Methodological innovations in SM and VWC retrieval
•	Synergistic use of multi-frequency SAR data
•	Uncertainty characterization of SM and VWC products against in situ observations and reference datasets
•	Applications of SM and VWC products in scientific and operational disciplines

Greenhouse gas observation remains fragmented across atmospheric plumes, land and water surfaces, industrial infrastructure, and the subsurface. Yet emissions, subsurface storage performance, and climate mitigation outcomes depend on transport across these connected domains. This proposed Community Contributed Theme (CCT) will bring together geoscientists, remote-sensing specialists, atmospheric scientists, engineers, and data scientists to develop multi-interface observation frameworks that connect detection, source attribution, flux quantification, and verification.

The theme covers three complementary application areas. First, we invite methane monitoring and the use of co-emitted volatile organic compounds (VOCs) for tracing, differentiating, and characterizing emissions from oil and gas production and transportation, chemical and petrochemical facilities, livestock systems, rice cultivation, and waste management. VOCs are treated as co-pollutants and atmospheric chemistry indicators, not uniformly as direct greenhouse gases. Second, carbon capture, utilization, and storage (CCUS) monitoring includes subsurface CO2 plume behavior and containment verification through integrated atmospheric, near-surface, groundwater, wellbore, and reservoir observations, including early warning of possible leakage pathways. Third, the scope encompasses other anthropogenic climate-relevant gases, including nitrous oxide and sulfur hexafluoride (SF6), with attention to fit-for-purpose measurement approaches for diffuse agricultural sources, industrial installations, and electrical equipment.

We particularly welcome advances in satellite and airborne spectroscopy, unmanned aerial sensing, ground-based optical and laser instruments, soil-gas and groundwater measurements, geophysical monitoring, and InSAR-based deformation analysis. These observations should be linked through multimodal data fusion, atmospheric transport and inverse modeling, physics-informed learning, geospatial digital twins, and uncertainty-aware monitoring, reporting, and verification (MRV). Key challenges include reconciling spatial and temporal scales, detecting intermittent or low-concentration releases, separating anthropogenic signals from background variability, and validating observations against independent measurements.

Existing IGARSS 2027 themes address atmospheric chemistry (M.4), subsurface sensing (T.7), multimodal data fusion (T.18), and multi-platform Earth observation (S.6), while SP.6 emphasizes datasets and evaluation. None is dedicated to the end-to-end scientific problem of linking atmospheric observations with surface processes and subsurface integrity across multiple greenhouse gas classes and emission sectors. This CCT therefore complements those themes by establishing a shared, application-driven forum for cross-interface methodologies, comparable benchmarks, and actionable emission and storage verification.

Earth Observation (EO) increasingly relies on heterogeneous data from optical, SAR, infrared, hyperspectral, and LiDAR sensors to capture the complexity of terrestrial environments. Although multimodal learning and foundation models have substantially improved remote sensing perception, most existing approaches remain focused on predefined tasks and isolated prediction outputs. Transforming these capabilities into autonomous geospatial intelligence requires agents that can interpret diverse observations, integrate complementary evidence, reason about complex spatial relationships, and actively invoke analytical tools to accomplish high-level objectives.

This Community-Contributed Theme focuses on multimodal GeoAgents that bridge cross-sensor perception and grounded geospatial reasoning. The central objective is to investigate how autonomous agents can transform heterogeneous EO measurements into coherent, evidence-supported interpretations through coordinated perception, reasoning, planning, and tool use. Particular attention will be given to the interaction between multimodal representations, vision-language models, geospatial knowledge, and task-specific analytical models.

We welcome contributions on agent-driven multimodal information fusion; vision-language agents for remote sensing interpretation; cross-sensor semantic alignment and spatial grounding; agentic object detection, segmentation, and change analysis; reasoning over optical, radar, infrared, and point-cloud observations; retrieval-augmented geospatial understanding; and iterative perception-reasoning frameworks with uncertainty-aware tool selection. Research addressing complex scenes, incomplete observations, and conflicting multimodal evidence is particularly encouraged. Representative applications include urban monitoring, environmental assessment, infrastructure analysis, and disaster-related scene interpretation.

Topics of interest include, but are not limited to:
(1) Agentic multimodal perception and cross-sensor data fusion for Earth Observation.
(2) Vision-language models and multimodal foundation models for geospatial agents.
(3) Cross-modal alignment, spatial grounding, and evidence-based geospatial reasoning.
(4) Autonomous task planning, tool use, and iterative perception-reasoning workflows.
(5) Agent-driven object detection, semantic segmentation, and change analysis.
(6) Uncertainty-aware reasoning, cross-domain generalization, and trustworthy multimodal GeoAgents.

Scope:
Urban understanding requires connecting what can be observed from above, what is present in three dimensions, and how urban places are described and used. Satellite and aerial imagery capture spatial patterns and temporal change. LiDAR and reconstructed geometry describe the vertical structure of buildings, vegetation, and infrastructure. street-level imagery, maps, and language provide complementary information on appearance, semantics, and function. This theme brings these perspectives together through multimodal GeoAI, with Earth observation serving as a core source of spatial evidence. We invite methodological and application-oriented contributions that align heterogeneous urban observations, learn transferable representations, and support spatially grounded interpretation.

Importance:
Cities combine physical structure, human activity, and environmental processes within the same geographic space. Integrating these observations can support more complete representations of urban objects and their relationships, while also allowing discrepancies and missing information to be identified. Multimodal GeoAI therefore offers a pathway from individual mapping products to urban representations that can be queried, compared, and analyzed across scales. The proposed theme will connect remote sensing, 3D vision, language modeling, and urban science. It will encourage studies that demonstrate when complementary modalities improve urban inference, when they fail to transfer, and how predictions can be traced to observable evidence. Applications include urban inventories, infrastructure monitoring, urban renewal, heat and vegetation analysis, and assessment of urban spatial disparities. The emphasis is on measurable urban questions and robust evaluation, alongside advances in model design.

Rationale for Complementing General and Special Themes:
The closest General Theme is L.6: Urban and Built Environment. The proposal also connects to T.18: Multimodal Data Fusion, T.10: 3D Mapping, A.2: Foundation Models in Geoscience, and A.6: Digital Twins. Relevant Special Themes include SP.5: Foundation Models and Embeddings for Earth Observation and SP.6: Datasets, Benchmarks, and Evaluation Methodologies for Earth Observation.

Related Technical Committee:
GRSS Technical Committees that are most relevant include: Image Analysis and Data Fusion (IADF TC) and Earth Science Informatics (ESI TC). These committees cover multimodal fusion, data‐informatics workflows, visual‐language modeling applications for remote sensing, etc.

Recent developments in earth observation sensing technology have yielded new knowledge, methods, and tools for collecting and processing, and sensing. This session will provide a platform to discuss challenges and opportunities for multisource data fusion, AI methods for efficient and accurate remote sensing image classification, 3D mapping, and their applications, such as change detection for disaster management and assessment, land use and land cover analysis, environmental monitoring, and assessment.  

Various earth observation platforms (e.g., satellites, aircraft, and drones) and types of sensors (e.g., LiDAR, optical, SAR, hyperspectral), along with in-situ sensors, have noticeably increased during the last decade. As a result, massive amounts of rich data are systematically produced. This increase in the number and heterogeneity of Earth Observation data sources presents both opportunities and challenges. Opportunities result from the refinement of sensors and sensing methods, the ability to capture increasingly granular data via those improved sensors and methods, and the improved data storage, manipulation, analysis, and interpretive capacity of modern computational hardware and software, cloud computing, and IoTs. Challenges result from the difficulty in accurately fusing data, so that a coherent, comprehensive, fully detailed, and integrated understanding of data from multisensor. This session brings together state-of-the-art research and development on AI algorithms, techniques, and software tools to address the above-mentioned challenges and to better exploit spatial and temporal features and dependencies in RS data. The focus of the session will be on data-driven models (such as CNN, GAN, RNN, etc.)  that are highly adaptable to changes in large remote sensing datasets and can find unexpected patterns in multi-source and multi-temporal data.

Dedication. Prof. Luciano Alparone (1960–2026) spent his whole academic career at the University of Florence, where he became Full Professor and led the Signal Processing and Telecommunications Laboratory. He authored more than 300 papers on remote sensing data compression, multiresolution analysis, SAR processing and, above all, multisensor data fusion. His 2004 GRSL Prize Paper on the global quality measurement of pansharpened imagery set a reference for fusion assessment. He helped shape the GRSS Data Fusion Technical Committee and was among the organizers and evaluators of the first Data Fusion Contest in 2006. This theme gathers the community he helped build to present current work in the areas he pioneered.

Scope. The theme invites contributions on:
•	Pansharpening and multispectral and hyperspectral image fusion, from model-based to deep learning approaches;
•	Optical–SAR, thermal and multi-temporal fusion, and fusion for upcoming missions;
•	Quality assessment of fused products with and without reference, including consistency and synthesis protocols;
•	Compression, image co-registration, denoising, and SAR despeckling as components of fusion chains, including on-board processing;
•	Open datasets, benchmarks and reproducible toolboxes for image fusion;
•	A more natural and grammatically correct version would be:
•	The use of fused products for downstream applications, such as classification, semantic segmentation, object detection, change detection, biophysical parameter estimation, hazard monitoring, and so forth.

Importance. Fused images support applications in land cover, agriculture, urban environments, and hazard assessment, yet their evaluation remains contested: reduced-resolution protocols can be misleading, while full-resolution indexes may disagree. Deep networks and foundation models add new capabilities, but also raise questions about physical consistency and generalization across sensors. A focused forum that combines new methods with rigorous assessment is timely for the new generation of spaceborne hyperspectral and high-resolution sensors.

How it complements the General and Special Themes. T.18 Multimodal Data Fusion covers fusion across all modalities and tasks; SP.6 covers datasets and evaluation for Earth observation at large. This CCT sits at their intersection: pixel-level fusion together with the methodology for judging its quality. It links signal processing, sensor physics and machine learning, and draws contributors who would otherwise be spread across T.15, T.18, and SP.6. The organizers are ready to coordinate with any overlapping CCT proposals.

Unoccupied aerial vehicles (UAV) have become an established carrier for Earth observation instruments, yet the overwhelming majority of UAV payloads reported in the literature are miniaturised imagers whose output is a georeferenced raster. This theme addresses a different and under-represented class of work: payloads built around a new measurement principle, designed to quantify a physical or chemical parameter that is either inaccessible from orbit or accessible only through empirical proxies. Possible contributions span multi-wavelength LiDAR with matched absorption bands for leaf water content, filter-based separation of far-red sun-induced chlorophyll fluorescence from reflected radiation, radar concepts for quantitative soil moisture retrieval, on-board wet-chemical analysis of surface waters combined with image-guided adaptive sampling, real-time mass-spectrometric monitoring of volcanic gas composition, and contactless multi-sensor hydrometry for water level and velocity and bathymetry.

What unites these systems is not an application domain but a shared engineering problem. Each instrument must deliver a traceable, quantitative value under constraints that no laboratory or satellite instrument faces: a payload mass of a few kilograms, a flight-time budget of minutes, platform vibration and attitude changes, and a rotor-induced disturbance of the very medium being measured. The achievable detection limit, drift and measurement uncertainty are therefore not properties of the sensor alone but functions of the resources mass, energy and time per measurement allocated to it.

The theme complements rather than duplicates the existing programme. S.8 (Unmanned Aerial Systems and Payloads) is organised around platforms and integration, and its submissions are dominated by imaging payloads and processing chains; the proposed theme deliberately gathers the non-imaging and contact-based instruments that fall outside that focus. S.17 (Advanced Future Instrument Concepts) is the natural second anchor, since several of the systems presented are early-stage instrument concepts whose airborne demonstration is itself the result. Our theme also feeds the Special Themes: volcanic gas monitoring and quantitative hydrometry speak directly to SP.2 (Monitoring Rapidly Changing and Hazardous Environments), while cold-region vegetation, snow and soil applications connect to SP.1 (Geoscience and Remote Sensing in the Arctic). By bringing instrument developers from metrology, analytical chemistry, photonics and microwave engineering into the GRSS programme, the theme widens the community contributing to UAV-based Earth observation.

Context and Motivation:
The rapid evolution of remote sensing technology is shifting from traditional 2D imagery towards a paradigm of high-fidelity 3D environmental modeling. As the demand for precision grows—driven by the need for detailed environmental and urban monitoring, infrastructure health assessment, and fine-scale ecosystem analysis, the emergence of new very-high-resolution (VHR) sensor technologies is fundamentally changing our ability to capture the complex geometry and spectral signatures of the Earth's surface.

Scope of the Session:
This special session aims to bring together researchers, engineers, and industry experts to discuss the latest breakthroughs in very-high-resolution sensing technologies specifically tailored for 3D applications. We invite contributions covering a broad spectrum, including but not limited to:
•	Data Acquisition: New sensor capacities for enhanced 3D reconstruction.
•	Data Quality Assessment: operations to enhance and assess radiometric and geometric accuracy.
•	Algorithmic Advancements: Machine learning and deep learning approaches for point cloud processing, 3D semantic segmentation, and super-resolution techniques for topographic modeling.
•	Applications: High-precision applications in digital twin construction, disaster management (landslide/flood monitoring), precision agriculture, autonomous navigation/urban planning and land surface evolution monitoring (erosion/glaciers melting/volcanic activity).

Importance and Justification:
As IGARSS continues to lead the discussion on the future of Earth Observation, addressing the "resolution gap" in 3D modeling is imperative. While global coverage is well-established, the transition to "local-to-regional" high-fidelity 3D monitoring represents the next frontier. This session will bridge the gap between sensor capabilities and the computational intelligence required to process the resulting massive data volumes. 
By fostering this dialogue, this session will contribute directly to the conference’s objective: advancing the state of the art in the field of 3D reconstruction—incorporating new sensors capacities—and its practical implementation in addressing global challenges.

Earth Observation (EO) is increasingly being used to support civil-security operations, policy-making and decision-making, covering application areas such as crisis and disaster management, border surveillance, maritime awareness, critical infrastructure monitoring and situational awareness. At the same time, emerging technologies such as Artificial Intelligence, foundation models and agentic AI, Digital Twins, multimodal data fusion, cloud and edge computing and new commercial EO constellations are rapidly transforming and enhancing geospatial intelligence (GEOINT) capabilities.

This rapid development has led to a proliferation of multiple isolated technologies, creating an increasingly complex operational environment in which identifying the most appropriate data sources, technologies, and their integration and transition from R&I activities to real-world operational capabilities has become a significant challenge. 
This Community Contributed Theme (CCT) addresses the need to effectively integrate relevant geospatial emerging technologies into operational environments, highlighting how their combination can enable next-generation GEOINT capabilities for civil security. By bringing together research, industry, and operational perspectives, the theme complements the General and Special Themes by providing a cross-cutting focus on the application and integration of technologies for operational needs, rather than their individual development. 

Expected contributions include studies, products or real-world use cases demonstrating EO applications across diverse technological and scientific domains, such as:
-	AI and foundation models for automated image analysis and change detection in operational settings and crisis scenarios;
-	Agentic and multimodal systems that combine EO data with other sources (e.g., social media, in-situ, AIS) to enrich situational awareness;
-	Digital Twins and simulation environments for borders, coastlines, critical infrastructure and other security-relevant scenarios.
-	Cloud and edge computing architectures enabling near-real-time EO exploitation in crisis and disaster response;
-	On-board AI and edge processing enabling near-real-time monitoring and early-warning capabilities.
-	Demonstrative cases of transitions from technological innovation to operational use in civil security scenarios.

The session welcomes a variety of contributions demonstrating the application of EO and emerging geospatial technologies across diverse civil-security domains. Particular attention will be given to the transition from technological innovation to operational use, including integration, interoperability, validation, trust and interaction with existing analytical workflows.

In recent years, nighttime remote sensing has gained prominence as a powerful tool for observing human activities and environmental conditions in a rapidly changing world. Applications range from identifying wildfires, detecting power outages following natural disasters, and analyzing changes during the COVID-19 pandemic to examining the impacts of wars, borders, poverty, urbanization, light pollution, and cultural and social events. Nighttime observations have also provided new perspectives on environmental and societal processes, including air pollution, urban heat, animal behavior, vegetation dynamics, and human health. Together, these applications demonstrate the value of nighttime remote sensing for observing both long-term transformations and rapid or unexpected changes in human and natural systems, while also demonstrating its growing role in Earth Action by translating observations into information that can support planning, response, and decision-making.

Nighttime remote sensing is fundamentally distinct from daytime remote sensing. Artificial illumination, thermal emissions, and other nighttime signals reveal dimensions of human activity and the environment that are difficult to observe during the day, while also introducing unique methodological challenges related to observation geometry, atmospheric conditions, temporal variability, and sensor differences. Long-term observations from systems such as VIIRS-DNB and NASA’s Black Marble products have enabled increasingly detailed time-series analysis of nighttime changes, while advances in satellite and sensing technologies are expanding nighttime observations toward multispectral and meter-scale imagery. Emerging nighttime thermal observations, ground-based imaging systems, artificial intelligence, multimodal data integration, and time-series methods further create opportunities to move from static observations toward continuous monitoring and detection of change, and ultimately toward more timely and actionable information.

This session aims to provide a platform for researchers to discuss new nighttime remote sensing data, methods, monitoring frameworks, and applications. We particularly welcome research on time-series analysis; continuous and near-real-time monitoring; new sensors, data sources, and analytical frameworks; rapidly changing cities, regions, and countries; and applications examining urbanization, socioeconomic and environmental change, disasters, conflicts, infrastructure and power outages, and other disruptions and recovery processes. We also welcome research that advances Earth Action by demonstrating how nighttime observations can be translated into timely, actionable information that supports real-world needs and informed decision-making. Through these perspectives, the session seeks to explore how nighttime remote sensing can help observe, understand, and monitor an increasingly changing world, while strengthening the connection between Earth observations and action.

Space-based microwave imager with low-to-high frequency channels is a strong tool for monitoring water-related variables inside of clouds and Earth’s surface through clouds on a daily basis. The Advanced Microwave Scanning Radiometers (AMSR) are a series of Japanese microwave imagers that have unique capabilities in their high spatial resolution with large-sized main reflector (~2-m diameter), dual polarization channels for all frequency bands from 6.9 to 89 GHz, and frequent global observation with wide swath more than 1500-km. The series of instruments provide continuous global water cycle observations for more than 23 years by previous two missions, AMSR-E on board the NASA’s Aqua satellite launched in 2002 and AMSR2 on board the Global Change Observation Mission - Water (GCOM-W) launched in 2012 and currently in operation. Data from the AMSR series is widely used in meteorological agencies in Japan, U.S., Europe, Australia, etc., and is also used for monitoring of sea ice in polar oceans. It also contributes to producing satellite-based merged precipitation products, such as GSMaP, IMEGR, CMORPH, etc., to monitor hourly global rainfall distribution in a near-real-time basis. 
The latest instrument, AMSR3 on board the Global Observing SATellite for Greenhouse gases and Water cycle (GOSAT-GW) launched in June 2025 and released data to public in June 2026, succeeds the afternoon orbit observation of AMSR2, and expands its observation capability to respond to new user needs. Major improvements in AMSR3 are new high-frequency channels (166- & 183-GHz) for snowfall retrievals and water vapor analysis in numerical weather predictions, and additional 10.25-GHz channels with better Noise Equivalent Delta Temperature (NEDT) for robust sea surface temperature (SST) estimates.
In this session, we will focus on the observation results and outcomes obtained by the past and current AMSR series, including the latest AMSR3, covering calibration, inter-comparison with other microwave imagers, ground validation, algorithm development, and outcomes in science and applications.

Accurate, high-resolution observations of ocean surface currents are essential for understanding upper-ocean dynamics, air–sea interactions, and marine transport. Satellite synthetic aperture radar (SAR) Doppler measurements and wide-swath altimetry offer complementary opportunities to advance these observations across diverse ocean environments. However, major challenges remain in reconstructing two-dimensional current vectors from radar line-of-sight velocities, separating current signals from wind-wave contributions, and achieving reliable retrievals during extreme weather.
This session invites contributions on emerging methods and applications for observing ocean surface currents, emphasizing three interconnected themes: vector current reconstruction from SAR Doppler measurements; retrieval under strong winds and extreme sea states; and integration of SAR with Surface Water and Ocean Topography (SWOT) observations to investigate eddies, fronts, and submesoscale variability. Contributions spanning retrieval theory, algorithm development, observational validation, and oceanographic applications are welcome.
Methodological topics include dynamically constrained inverse methods; physics-guided machine learning; correction of wave-induced Doppler signals, including wave-breaking effects; multi-platform and multi-geometry observations; and uncertainty characterization. Particular attention will be given to the limitations of conventional retrieval approaches during tropical cyclones, extratropical storms, and rapidly evolving sea states, together with strategies for improving robustness and physical consistency.
A central focus is the complementary information provided by SAR-derived surface velocities and SWOT-derived sea-surface-height and geostrophic-current estimates. Contributions are encouraged that explore how their integration can help distinguish geostrophic and ageostrophic variability, characterize fine-scale eddies and frontal structures, and investigate wave–current interactions and cross-scale energy transfers. Studies linking these observations to upper-ocean mixing, material transport, and air–sea coupling are also welcome.
The session will address validation using surface drifters, high-frequency radar, uncrewed observing platforms, and ocean analyses, including challenges associated with differences in sampling, spatial resolution, and the physical quantities observed. Reproducible processing, intercomparison exercises, quality-controlled datasets, and open-source tools are additional priorities.
By bringing together radar remote sensing, satellite altimetry, physical oceanography, and machine learning, this session aims to connect sensor-specific retrieval advances with multi-sensor understanding of ocean dynamics. It will foster international collaboration toward more reliable surface-current observations and their use in ocean forecasting, pollutant tracking, search and rescue, and maritime operations. 

Recent advances in Earth observation (EO) foundation models and multimodal representation learning have significantly enhanced our ability to extract information from large-scale satellite observations. However, a fundamental challenge remains: how can data-driven models move beyond learning statistical patterns in remote sensing imagery toward representing physically meaningful Earth surface properties, understanding their temporal dynamics, and predicting their evolution? Addressing this challenge requires bridging the gap between remote sensing measurement physics, geospatial artificial intelligence, and dynamic Earth surface modeling.
Earth surface states, including vegetation structure, soil moisture, freeze–thaw conditions, surface deformation, and anthropogenic changes, are observed indirectly through sensors with distinct measurement principles, viewing geometries, and spatiotemporal characteristics. Consequently, observed signal variations do not always reflect actual changes in surface conditions. Developing physically grounded representations that distinguish observation-induced variability from genuine geophysical processes is therefore essential for advancing EO from conventional image interpretation toward reliable state estimation and predictive modeling.
This Community Contributed Theme (CCT) aims to foster interdisciplinary research on physics-grounded representation, reconstruction, and prediction of Earth surface states. It welcomes contributions addressing the integration of physical observation models and domain knowledge into multimodal representation learning; multi-timescale modeling of persistent surface properties and dynamic changes; physics-aware signal processing and inverse modeling; object-centric and process-aware state representations; physically consistent generative modeling and spatiotemporal prediction; and benchmark methodologies for evaluating physical consistency, generalization, and predictive reliability. Applications spanning terrestrial ecosystems, cryospheric environments, agricultural landscapes, and urban infrastructure are particularly encouraged.
The proposed theme complements existing IGARSS General and Special Themes by establishing a focused connection between physical measurement principles, learned geospatial representations, and dynamic state understanding. In particular, it extends SP.5: Foundation Models and Embeddings for Earth Observation beyond general-purpose feature learning toward physically meaningful and verifiable state representations. It also bridges related themes in remote sensing inversion, multimodal data fusion, temporal analysis, and benchmarking by emphasizing the physical interpretation and predictive utility of learned representations. By bringing together researchers from remote sensing physics, machine learning, and Earth system science, this session seeks to advance more interpretable, physically consistent, and reliable approaches to Earth observation intelligence.

Machine Learning (ML) has transformed remote sensing by offering powerful data-driven tools capable of analyzing vast datasets and solving complex problems. However, traditional ML models may struggle to generalize across novel datasets or environmental conditions when underlying physical processes are poorly understood. Moreover, machine learning has often been criticized for the difficulty of model interpretation and the lack of physical consistency. The proposed CSS aims to explore the potential of Physical Model Informed Machine Learning (PIML) methods, which integrate the strengths of physical models—rooted in electromagnetic theory, sensing characteristics, and geophysics—with the data-driven flexibility of machine learning. By embedding fundamental physical constraints, PIML aims to develop robust and interpretable models that generalize well even with sparse data. This hybrid approach is particularly relevant in applications where remote sensing data may be limited or unrepresentative, such as climate monitoring and disaster prediction, as well as in cases involving the solution of forward and inverse problems. Topics include, but are not limited to the following
• Physics-Informed Neural Networks • Hybrid Data and Physics-Based Modeling • Physics-regularized Deep Learning • Data Augmentation with Physics-Based Simulations
• Physics-Guided Inversion Techniques
• Physics-Driven Surrogate Modeling
• Integrate Data Models and Assimilation Methods
• Explainable AI (XAI) with Physical Constraints
• Physics-guided uncertainty quantification

Radio Frequency Interference (RFI) has been having an increasingly detrimental impact on microwave remote sensing. Interference can corrupt passive and active microwave measurements and reduce the ability of retrieving relevant geophysical measurements over many regions of the globe. This is particularly harmful for passive instruments, which measure the very weak natural emission from the Earth surface and atmosphere that cannot be detected in the presence of manmade emissions of several order of magnitude strong.  Many innovative technical advances, both in software development and hardware design, have been made by the Earth observing microwave remote sensing community to improve identification of interference and mitigation of its negative effects. The problem of increasing occurrence of Radio Frequency Interference in measurements is closely related to the ever growing spectrum needs of commercial interests, particularly those of the telecommunication industry. The ongoing development of mega-constellations of telecommunication satellites to provide global Internet connectivity is a clear example of how the demand for spectrum is putting an enormous pressure on frequency bands utilized for microwave remote sensing. For this reason, the remote sensing community has also been working closely with spectrum managers to protect the frequencies of interest for science applications.

The sessions in this Community Contributed Theme (CCT) will showcase various interference detection and mitigation techniques developed within the passive and active remote sensing communities and reports on the status of current and upcoming missions with regards to dealing with RFI, and also discuss spectrum management issues facing remote-sensing instruments.  

This CCT proposal is submitted by the Frequency Allocations in Remote Sensing (FARS) Technical Committee. The FARS TC is asking for three oral sessions, one focused on RFI in passive microwave remote sensing, one on RFI in active microwave remote sensing and one on spectrum management.  

This session invites papers that relate to remote-sensing based efforts to bridging the gap between technology and tangible socio-economic improvements, alongside the mitigation of natural and anthropogenic hazards, as outlined in the United Nations Sustainable Development Goals (SDGs). As the world bears the escalating impacts of climate change—from extreme weather events to gradual environmental degradation—there is a pressing need to address the associated environmental challenges through innovative solutions and by advancing the breadth and depth of Earth Observation (EO). This session aims to gather and showcase cutting-edge research that embodies this critical philosophy, demonstrating how EO can transition from data acquisition to actionable intelligence for humanity’s benefit.

We welcome submissions on recent advances in Earth observation, including both methodological developments and technological solutions that support evidence-based sustainable management of terrestrial, coastal, and marine resources. The session scope encompasses all environmental spheres, including the cryosphere, biosphere, atmosphere, and geosphere, recognizing the interconnected nature of Earth's systems. The common thread for all submissions should be a clear and demonstrable relevance to sustainable development targets. We are particularly interested in techniques that leverage data from optical, lidar, and SAR systems—utilizing various imaging modes such as interferometry, polarimetry, tomography, multi-frequency analysis, optical indices, electromagnetic (EM) models, and lidar waveforms—along with advanced methods for quantifying the resulting environmental products and their uncertainties. 

We welcome work positioning remote sensing as the foundational data layer for Digital Twins, decision-support platforms, and cross-sectoral modelling frameworks—for example, water–energy–food (WEF) Nexus systems, urban digital twins, or climate services—where EO feeds directly into simulation, forecasting, and policy scenarios. Papers that combine satellite imagery with sparse in-situ sensors, meteorological reanalysis, census and socio-economic datasets, or provincial/national statistics are also encouraged. Submissions addressing multi-hazard interactions, rapid-onset or short-lived phenomena, and reevaluation of established risk paradigms are welcome. In short, we invite contributions that not only push the technical boundaries of remote sensing but also clearly articulate the pathway from data to impact, helping to realize the full potential of Earth Observation for a sustainable and resilient future.

Wildfires are becoming more frequent, larger, and more destructive across many parts of the world due to climate change, land use shifts, and evolving fuel and vegetation dynamics. This Community Contributed Theme (CCT) seeks to unify the remote sensing, geospatial modeling, and decision support communities around the wildfire life cycle: from early detection and risk assessment (including fuel moisture and pre‑conditions), through fire spread and behavior modeling, to burned area/emission quantification, post‑fire ecological recovery, and informing policy and land management strategies. Contributions may exploit optical and thermal sensors, SAR, LiDAR, UAV/drone platforms, IoT/in‑situ networks, data assimilation, machine learning / deep learning, new satellite constellations, and multisource data fusion.

The importance of this theme is manifold: it addresses disaster risk reduction, public safety, ecosystem integrity, carbon cycle and climate feedbacks, air quality, and land stewardship. Faster detection and more accurate modeling of wildfire evolution provide the means to save lives, guide evacuations, optimize resource deployment, support prescribed burns, calculate emissions, and plan effective post‑fire rehabilitation.

IGARSS 2025 already includes SP.1: Natural Disasters & Disaster Management among its Special Scientific Themes. This theme strongly overlaps with SP.1 in its concern for disasters, monitoring, and risk reduction. However, SP.1 is broad (encompassing many types of natural hazards) and does not always focus explicitly on integrating emerging sensor technologies, real‑time / near‑real‑time modeling for wildfires, cross‑scale data fusion, and decision support systems tailored to the wildfire phase transitions (pre‑fire, active fire, post‑fire). Our proposed CCT would complement SP.1 (and related General Themes such as Forest & Vegetation: Application & Modelling, Change Detection, Data Fusion) by concentrating specifically on wildfires and the full chain from sensing to actionable modeling and policy/management integration.

We anticipate strong community engagement, with at least five indicative submissions: e.g. early detection via multispectral and thermal imagery, fuel moisture estimation using microwave and optical fusion, spread forecasting with physics‑based + AI hybrid models, burn severity and recovery using LiDAR / hyperspectral techniques, emissions estimation integrated into atmospheric and health models. The session format may include both oral and poster presentations, accommodating methodological innovations and applied case studies. This CCT aims to bring together sensor technologists, modelers, ecologists, fire managers, and policy actors to foster interdisciplinary collaboration and push forward the state of wildfire remote sensing and modeling.

Coastal regions are among the world's most dynamic and densely populated environments, where rapid urbanization, intensive resource use, ecosystem change, and climate-related pressures interact. Sustainable coastal development requires reliable and timely information on urban growth, land-use transitions, coastal wetlands, biodiversity, water quality, environmental degradation, and natural hazards. Remote sensing provides essential observations for understanding these interconnected processes across spatial and temporal scales.

Recent advances in Earth observation have expanded the range of data available for coastal monitoring, including spaceborne and airborne optical imagery, synthetic aperture radar (SAR), hyperspectral imagery, LiDAR, and complementary in-situ observations. Their integration enables more comprehensive characterization of coastal environments and supports applications ranging from urban expansion and land reclamation mapping to wetland dynamics, ecosystem assessment, pollution monitoring, and coastal hazard analysis. However, challenges remain in integrating observations with different spatial and temporal resolutions, exploiting complementary spectral and scattering characteristics, handling cloud-contaminated or incomplete observations, and translating remote sensing products into actionable information for sustainable development.

This Community-Contributed Theme aims to bring together researchers working on remote sensing methods, Earth observation data, and interdisciplinary applications for coastal sustainability. We welcome original contributions addressing coastal urbanization, land-cover and land-use change, coastal wetlands and vegetation, marine and coastal environmental quality, reclamation and aquaculture, biodiversity, climate change impacts, and natural hazards. Contributions on multisource data fusion, time-series analysis, machine learning, foundation models, uncertainty assessment, and scalable monitoring approaches are also encouraged when they advance understanding or observation of coastal systems.

The proposed theme complements the IGARSS 2027 General Themes. It provides an interdisciplinary focus connecting advances in remote sensing techniques with the environmental and societal challenges of coastal regions. The theme builds on our experience organizing coastal sustainability sessions at IGARSS 2023, 2024, 2025 and 2026. These activities have provided a platform for research exchange among participants working on coastal environments from different methodological and geographical perspectives. We aim to continue fostering international collaboration and discussion on how emerging Earth observation capabilities can support sustainable and resilient coastal development.

Remote sensing is increasingly central to oil and gas exploration and production and to environmental monitoring across production, storage, and transportation operations. Satellite constellations now provide global coverage and generate terabytes of freely available multispectral and hyperspectral data daily, while optical and radar sensors are achieving ever finer spatial and spectral resolutions—down to centimeter-scale for optical imagery and submetric-scale for radar. These capabilities enable new opportunities for surface characterization on land and at sea. Onshore, remote sensing supports structural and mineralogical mapping, surface deformation quantification, vegetation-change detection, and identification of soil alteration associated with hydrocarbon seeps. Offshore, it enables oil-film detection and oil-type characterization. Mapping oil slicks is valuable both for hydrocarbon exploration - by helping identify natural seeps - and for emergency response to anthropogenic pollution. This session aims to bring together academic and industrial communities applying remote sensing to mineral and petroleum research and to provide a broad overview of current and emerging applications. We invite contributions on the following topics:

Onshore Applications: 
•	Hyperspectral and multispectral imaging for geological and mineral mapping
•	Detection of surface liquid and gaseous hydrocarbons and hydrocarbon-induced soil alteration
•	Monitoring of mineral and oil and gas field development operations
•	Surveillance of gas storage facilities (CO₂, CH₄, H₂) and associated plant infrastructure
•	Interferometric techniques for quantifying surface deformation
•	Recent advances in big data analytics, particularly deep learning

Offshore Applications
•	Detection of surface hydrocarbons
•	Identification of natural oil seeps for hydrocarbon exploration
•	Monitoring of anthropogenic oil spills in support of emergency response
•	Remote sensing for subsurface mapping and monitoring
•	Recent progress in big data analysis, with emphasis on deep learning for oil-slick pattern recognition

Scope
This CCT invites contributions that advance forest evapotranspiration (ET) estimation through remote sensing of canopy structure, plant physiology, and water dynamics. Contributors may use optical/thermal, SAR/InSAR, passive microwave, LiDAR/GEDI, and SIF observations, with possible validation against FLUXNET/ICOS, T/ET partitioning, and ecohydrological models. Topics include:
·	Canopy Structure and ET Modeling — linking canopy structural properties, such as height, stratified LAI, and vertical foliage profiles, to improved ET modeling and T/ET partitioning. Studies using LiDAR, GEDI, TomoSAR, or passive microwave are encouraged [1].
·	Plant Physiology and Water Status Retrieval — retrieving plant physiological and water-status information and drought response, such as stem water content, vegetation water content, and canopy water potential. SAR, passive microwave, and related observations are welcome [2].
·	ET Dynamics and Multi-Sensor Validation — multi-sensor time-series retrieval of ET components, cross-technique integration, and validation with FLUXNET/ICOS observations. Studies addressing drought-induced ET dynamics and T/ET partitioning are particularly solicited [3].

Importance
Forests dominate terrestrial ET. However, operational forest ET evaluation still relies on cloud-sensitive optical/thermal data and often treats canopies as homogeneous “big leaves.” Microwave, LiDAR/GEDI, and Sun-Induced Chlorophyll Fluorescence (SIF) observations provide complementary information on energy, water status, structure, and physiology. Still, they remain poorly integrated for large-scale forest ET modeling, retrieval, and validation. Big-leaf and simplified remote sensing approaches are prone to uncertainties of up to 10-fold in assimilation/transpiration and 300% in vertical gas exchange [4]. Although LiDAR/GEDI/TomoSAR can retrieve canopy height and stratified LAI, their links to improved ET estimates and large-scale validation remain limited. Additionally, stem water content, a key regulator of transpiration, also remains difficult to retrieve quantitatively and routinely. A systematic multi-sensor framework that integrates canopy structure, plant water status, and ET dynamics—and explicitly accounts for vertical structure and physiological constraints—is therefore needed.

Complementarity with the General and Special Themes
·	L.2 (Land Cover Dynamics): further enhance all-weather time series of forest water dynamics and sub-canopy sensitivity.
·	L.4 (Forest and Vegetation: Biomass and Carbon Cycle): add water–carbon coupling through dynamic ET monitoring and FLUXNET-based T/ET partitioning.

References
[1]	J. Bu and J. Xiao, “Upscaling eddy covariance measurements of carbon and water fluxes to the continental scale by incorporating GEDI-derived canopy structural complexity metrics,” Remote Sensing of Environment, vol. 329, p. 114930, 2025.
[2]	T. Leistner, A. R. Monteith, J. Gutierrez Lopez, and L. M. H. Ulander, “Estimating stem water content from tower-based L-band tomographic radar using a single-scattering model of a uniform layer: A first in situ experiment,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 19, pp. 3477–3487, 2026.
[3]	M. M. Mueller et al., “Sentinel-1 backscatter time series for characterization of evapotranspiration dynamics over temperate coniferous forests,” Remote Sensing, vol. 14, no. 24, p. 6384, 2022.
[4]	P. J. Bennet, A. R. Monteith, and L. M. H. Ulander, “Temporal variations in L-band reflectivity profiles of a boreal forest during growth and drought,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 19, pp. 16299–16313, 2026.

The IEEE Geoscience and Remote Sensing Society (GRSS) Standards for Earth Observation (GSEO) Technical Committee (TC) is focused on facilitating the development of protocols and standard practices for coordinated remote sensing data acquisition and assimilation by industry, academia, and governments. The scope of the GSEO is to help achieve full interoperability of hyperspectral, passive microwave, synthetic aperture radar (SAR), GNSS reflectometry (GNSS-R), and future remote sensing technologies data by empowering virtual constellations to enhance the quality of Earth observation data while minimizing the revisit time. The goals of our standardization initiatives include improving disaster management and early warning as well as optimizing natural resources, such as agriculture productivity. The GSEO takes a holistic approach that incorporates not only technological considerations but also policymaking, economics, and societal implications. We think that this vision can generate new insights within a geopolitically divided planet that is being burdened by the impacts of climate change.

The standardization landscape is highly dynamic, with many relevant stakeholders working on a diverse range of topics. There is a need to integrate different perspectives, and we believe that the standardization process that the GSEO uses under the IEEE Standards Association (SA), and our technical operational philosophy, provide the required next step to leverage global efforts in a coordinated and cost-effective scientific manner. The GSEO continues to foster the creation of standards that address identified needs. This is being accomplished by liaising with a variety of GRSS TCs for leveraging the full technical capabilities of the GRSS. We understand this effort through a long-term perspective, keeping in mind key areas of interest, such as natural hazard identification and monitoring, emergency response and coordination, recovery and reconstruction, and sustainable economic growth based on well-organized use of our natural resources.

This session serves as a scientific forum, including needs and thoughts coming from research institutes, universities, space agencies, e.g., National Aeronautics and Space Administration (NASA) and European Space Agency (ESA), as well as from private companies, which are becoming a key actor within the New Space era.

The growing integration of Artificial Intelligence (AI) in Earth Observation (EO) is transforming our capacity to monitor, understand, and address environmental and societal challenges. However, as the influence of AI expands, so too does the need for responsibility, transparency, and fairness in its design, deployment, impact assessment, and role in fostering efficient policies. These principles become even more vital in high-stakes contexts such as emergency responses to natural hazards, where timely, accurate, and trustworthy AI-driven insights are essential to mitigate risks and protect vulnerable communities. This session, hence, will focus on Responsible AI for EO, exploring both methodological foundations and practical applications that ensure AI-driven EO systems operate ethically and inclusively. This initiative aims to consolidate community efforts on topics such as AI for social good, bias detection and mitigation, data and model transparency, FAIR and open science principles, uncertainty-aware and explainable/interpretable AI systems, AI safety and robustness, ethical frameworks for global sustainability, and the role of AI on measuring, informing and supporting public policies, stakeholder engagement for AI systems, and the risks AI with EO applications can bring to society.
Participants will share insights into best practices, standards, and frameworks that promote responsible innovation across both the methodological and application dimensions of AI for Earth Observation. The session will explore how principles of transparency, accountability, fairness, and reproducibility can be integrated throughout the AI lifecycle - from data curation and model development to deployment and impact assessment. Emphasis will be placed on fostering open, explainable, transparent, and trustworthy approaches that advance scientific rigor while ensuring societal benefit.

Relevance and Objectives:
* Highlight the growing importance of Responsible AI as the backbone for trustworthy EO solutions.
* Showcase practical methodologies for detecting and mitigating bias, ensuring fairness, and maintaining transparency in EO-based AI pipelines.
* Encourage open science practices by adopting FAIR principles and reproducible benchmarking initiatives.
* Encourage the development of explainable, interpretable, and transparent EO-based AI systems.
* Encourage uncertainty and quality transparency of EO-based AI products. 
* Foster interdisciplinary collaboration among scientists, policymakers, and industry and society stakeholders to ensure AI for EO serves the collective good.
* Build community consensus toward establishing guidelines and ethical frameworks aligned with international standards and sustainability goals.

The session is part of the activities of IADF through the working group on Responsible AI (WG-RAI), and will be promoted through IADF TC newsletters, GRSS communication channels, and professional networks, encouraging broad participation across global regions. By focusing on Responsible AI, the session will promote inclusivity, ensuring that EO-derived insights and AI models reflect the realities and needs of underserved and marginalized communities. Through discussions on bias mitigation, data accessibility, and open-source collaboration, this effort aims to reduce inequities in digital and scientific participation while fostering ethical, transparent, and socially responsible use of AI in Earth Observation, and stimulating the discussions on the risky uses of AI with Earth Observation data.

Protected bands used for microwave radiometry are under increasing pressure from spillover from adjacent bands, intentional jamming, and growing demand for terrestrial and space-based connectivity.

At the same time, a new generation of microwave radiometers is emerging that observes both protected and unprotected parts of the spectrum over very wide bandwidths. Such measurements offer new opportunities for geophysical retrievals, but they also make the robust detection, characterisation, and mitigation of radio-frequency interference (RFI) an integral part of the observing system.

Missions will need to use fine spectral sampling and RFI detection and mitigation techniques to retrieve their respective geophysical products, such as sea-state parameters or atmospheric profiles. Missions using wideband retrievals have been proposed by space agencies worldwide, including JAXA’s SAMRAI, NASA’s PolarRad, and ESA’s CryoRad, as well as commercially through Spire’s HyMS mission. Although ESA’s upcoming CIMR mission will observe within the protected part of the spectrum, it will need to perform onboard RFI processing for all channels and polarisations. RFI processing is therefore becoming increasingly important for radiometers operating within protected bands, illustrating that interference mitigation is becoming a system-level consideration for future passive microwave missions.

This Community-Contributed Theme aims to bring together the microwave radiometry, signal processing, instrument design, and spectrum regulation communities to discuss the state of the art and future directions in RFI processing for ultra-wideband passive microwave sensing. Topics of interest include RFI monitoring and characterisation; onboard and ground-based mitigation techniques; artificial intelligence and machine-learning methods for interference suppression; calibration challenges; and implications for future satellite architectures and observing systems.
The theme complements the general and special themes of IGARSS by addressing a cross-cutting challenge that affects a broad range of current and future microwave remote-sensing missions.

Mission Overview
The Copernicus Imaging Microwave Radiometer (CIMR) is one of six Copernicus Expansion Missions being implemented by the European Space Agency on behalf of the European Commission. It will deliver high-spatial-resolution microwave radiometry measurements and derived products, with global coverage and sub-daily revisit over the polar regions and adjacent seas.
Mission Configuration and Coverage
The mission will comprise two satellites launched sequentially. Each satellite is designed for a 7.5-year operational lifetime and will carry sufficient fuel for up to 12 years, potentially providing approximately 20 years of continuous observations. The first launch is anticipated in 2028/29.
CIMR will operate in a dawn–dusk orbit coordinated with MetOp-SG-B1, enabling collocated observations over the polar regions within ±10 minutes. Its conically scanning radiometer will use a large, 8 m-diameter deployable mesh reflector at an incidence angle of 55°, providing a swath of approximately 2,000 km. This configuration will enable 95% daily global coverage with a single satellite, without a coverage gap at the poles.
Instrument Capabilities
The instrument will acquire dual-polarised measurements at L-, C-, X-, K- and Ka-band, with the following effective spatial resolutions:
•	L-band: <60 km
•	C-band: ≤15 km
•	X-band: ≤15 km
•	K- and Ka-band: <5 km, with a goal of 4 km
Measurements will be collected along both forward- and backward-scan arcs. On-board processing will strengthen resilience to radio-frequency interference and support the calculation of modified third and fourth Stokes parameters for all channels.
Geophysical Products
CIMR will support the retrieval of Level-2 geophysical products across the cryosphere, ocean, land and atmosphere domains, including:
•	Sea-ice concentration, thickness, drift, type and surface temperature
•	Sea-surface temperature and salinity
•	Ocean-surface wind vectors
•	Snow parameters
•	Soil moisture, land-surface temperature and vegetation indices
•	Atmospheric water parameters
Session Focus
In preparation for the CIMR mission, this session will promote international and cross-disciplinary collaboration across the cryosphere, ocean, land and atmosphere communities.

The RIT Open Community eXperiment (ROCX) 2025 brought together more than 60 researchers representing 23 organizations and six countries for a collaborative remote sensing field campaign at RIT’s Tait Preserve near Rochester, New York, USA. The campaign coordinated remote sensing data collections with 20 diverse ground experiments, creating a rich, well-characterized dataset intended for open distribution and use by the broader research and education community.

ROCX 2025 acquired multispectral, hyperspectral, polarimetric, and lidar data from ground-based, UAS, airborne, and satellite platforms. Ground experiments addressed a wide range of applications and research questions, including methane detection, sub-pixel target detection, humanitarian demining, sensor calibration, and wetland characterization.

This community theme invites ROCX participants and collaborators to share datasets, ground experiments, methodologies, lessons learned, and scientific results emerging from the 2025 campaign. Contributions demonstrating how ROCX data are being used for algorithm development, validation, education, and new research applications are also encouraged.

The session will also look ahead to ROCX 2027, presenting plans for the next field campaign and inviting the IGARSS community to help shape future experiments, data collections, collaborations, and research opportunities.


ROCX is well suited as a Community-Contributed Theme because it cuts across multiple IGARSS General Themes, bringing together hyperspectral remote sensing, multi-platform data collection, sensor calibration and performance, algorithm development, education, and open science. Much of the campaign data was acquired using UAS, airborne, and satellite hyperspectral sensors, with several experiments focused on radiometric and spectral calibration. The resulting openly accessible datasets will support remote sensing research, algorithm development, validation, and education. As IEEE GRSS, through its Geoscience and Spaceborne Imaging Spectroscopy (GSIS) Technical Committee, was a primary sponsor of ROCX 2025, IGARSS provides a natural forum for sharing campaign results and building community engagement toward ROCX 2027.

This Community Contributed Theme (CCT) will be highly interdisciplinary and will focus on the way, in which SAR sensors can be used for the surveillance of changing marine coastal environments. Presenters will demonstrate how these sensors can detect and quantify processes and phenomena that are of high relevance for the local fauna and flora, for coastal residents and local authorities, and for a better quantification of hazards caused by global change and other anthropogenic impacts. Examples include:
·	Coastline changes and coastal morphodynamics
·	Coastal run-off and marine pollution
·	Wind fields and storm events
·	Hazardous surface currents and waves 
·	Illegal, unreported and unregulated fishing
·	Marine litter
·	Operational use of SAR and other sensors for coastal zone applications.
Although the focus of this CCT is on the application of SAR, the CCT organizers encourage submissions of non-SAR presentations that complement SAR and lead to an interdisciplinary and a more comprehensive solution, and to address multiple end-user needs for sustainable development of the marine coastal zone in Arctic regions and worldwide. 
To achieve Diversity, Equity, Inclusion, Access & Belonging (DEIAB) values, the CCT chairs will strive to:
·	Encourage submissions from early-career scientists to help them gaining experience in presenting to an international audience
·	Organize the session with contributions from industry, academia, and government, to the extent possible
·	Ensure:
o	adherence to presentation time-limits so that all presenters have equal time to present their work
o	respectful intervention if an inexperienced presenter is struggling, with the aim to help the presenter get back on-track
o	constructive questions to the presenter and curtail destructive questions that seek to undermine the presenter

Importance
Coastal marine environments, being invaluable ecosystems and host to many species, are under increasing pressure caused by anthropogenic impacts such as, among others, growing economic use, coastline changes and recreational activities. Continuous monitoring of those environments is of key importance for the identification of natural and anthropogenic hazards, for an understanding of oceanic and atmospheric coastal processes, and eventually for a sustainable development and use of those vulnerable areas. Here, Synthetic Aperture Radar (SAR), because of its high spatial resolution along with its independence of day and night and its all-weather capabilities, is one sensor of choice.

Complement to General and Special Themes
This CCT will complement General and Special Themes, because it will particularly address coastal erosion and coastline changes (SP.2; Rapidly Changing Environments), marine pollution of different kinds (SP.2; Hazardous Environments, or Environmental Hazards), with special emphasis on coastal marine environments (O.5; Coastal Zones). The CCT will emphasize the use of radar sensors for environmental monitoring applications, hence it will complement to the given General and Special Themes through a clear focus on a class of remote sensing sensors that can be, and already are, widely used for the monitoring of the marine environment, particularly in high latitudes (SP.1; Remote Sensing in the Arctic).

Surface deformation is driven by diverse natural and anthropogenic processes, including groundwater extraction, fossil fuel exploitation, mining, and cryospheric degradation (permafrost, snow, and glacier melt). Anomalous deformation patterns frequently precede or accompany seismic activity and volcanic eruptions. Over extended temporal scales, subtle ground movements can induce detectable anomalies across both terrestrial and atmospheric parameters. When severe, surface deformation manifests as ground fracturing, faulting, and soil liquefaction, posing critical threats to civil infrastructure such as roads, dams, and buildings. High-altitude mountain environments are exceptionally vulnerable to these geomechanical shifts; catastrophic events like the Chamoli disaster and the Gorkha, Nepal earthquake triggered deadly floods and mass movements that impacted thousands of lives and erased entire villages. Modern spaceborne remote sensing—notably advanced Interferometric Synthetic Aperture Radar (InSAR) techniques and state-of-the-art satellite missions such as NISAR—now enables the precise detection of slow-moving surface precursor signals. These spaceborne observations provide vital early-warning capabilities for impending geohazards, including earthquakes, volcanic unrest, and glacial lake outburst floods (GLOFs). Contributions to this special issue are welcome on topics related to sea-level rise, coastal subsidence, and their combined impacts along global coastal margins.

Electromagnetic interactions with geophysical media govern the backscatter, bistatic reflection, and thermal emission observed by microwave remote sensing missions. Sensors that are impacted by these effects include scatterometers, altimeters, synthetic aperture radars, global navigation satellite system-reflectometry, and microwave radiometers. Theories of random medium and rough surface scattering and emission couple stochastic descriptions of the structure of a geophysical medium and the roughness of its interfaces with electromagnetic theory. Solutions are determined analytically under certain approximations or with the use of computational electromagnetic methods applied in a Monte Carlo simulation over medium realizations. Refinements in methods occur as new and more accurate descriptions of the geophysical medium itself are developed or as improved descriptions of the electromagnetic scattering process are incorporated. Applications include the remote sensing of Earth's oceans, atmosphere, and land surface including scattering from hydrometeors, snow, ocean waves, vegetation, and others. The resulting theories not only support the development of retrieval algorithms for translating sensor measurements into geophysical parameters, but also the development of insights into electromagnetic interactions that form the basis of new sensing concepts and technologies.

For more than 40 years, Prof. Leung Tsang of the University of Michigan has led the state-of-the-art in advancing the theory of scattering and emission from random media and rough surfaces, so that a session honoring his contributions is highly appropriate for IGARSS 2027. The session will include papers reporting ongoing advancements in the area across multiple application domains building upon the past and present works of Prof. Tsang. As a community contributed session, papers contributing to the theory of microwave interactions with statistically described media for remote sensing applications will be welcome for inclusion. 

Commercial and institutional operators now fly constellations of tens of small satellites that carry SAR, hyperspectral, and optical payloads. These constellations provide sub-meter resolution, revisit intervals of a few hours, and on-demand tasking. A constellation differs from a single flagship mission in ways that affect geophysical retrieval. Each satellite has a different instrument response and calibration history. A time series often contains images from several satellites with different orbits and viewing geometries. The size, weight, and power of a small satellite limit the antenna or telescope aperture, the radiometric performance, the on-board storage, and the downlink rate. At sub-meter resolution, the image statistics are frequently non-Gaussian, because a small number of strong scatterers or bright objects can dominate a resolution cell. Many detection and estimation methods were developed for coarser data and do not account for these properties.

The number of these satellites increases each year, and more scientists use their data in research and operations. However, no current IGARSS theme examines how the design of these satellites affects the quality of the science products.

This theme is for papers on these topics:
- Design of compact SAR, hyperspectral, and optical payloads for small satellites, including phased-array and reflector antennas, polarimetric modes, and compact imaging spectrometers.
- Radiometric, spectral, geometric, and phase calibration of small-satellite sensors, and consistency between the satellites of one constellation.
- On-board processing and data compression, and their effect on image quality and on derived products such as interferograms.
- InSAR, coherent and incoherent change detection, and moving target indication with data from more than one satellite or constellation.
- Statistical, physics-based, machine learning, and quantum computing methods for high-resolution, high-revisit data, including the fusion of SAR, hyperspectral, and optical data.
- Applications that need frequent revisit or fast tasking, for example flood mapping, volcano and landslide monitoring, sea-ice monitoring, and maritime monitoring.

Other IGARSS 2027 themes include parts of this subject. S.1, S.4, S.12, and S.13 include missions and instruments. T.2, T.14, and T.18 include InSAR, change detection, and data fusion. A.4 includes on-board processing, and SP.2 includes hazard monitoring. These themes do not include the full path from the small-satellite sensor to the geophysical result. For example, a compression algorithm can increase the phase noise in an interferogram, and a calibration difference between two satellites can cause a false change in a change detection product. This theme puts sensor engineers, calibration engineers, algorithm developers, and Earth scientists in the same session, so that each group can see how its work affects the work of the other groups. The theme supports the IGARSS 2027 theme, "Global Vision for a Changing Planet," because frequent observation is necessary to monitor fast changes on the surface of the Earth. Iceland and the Arctic have many of these fast changes, for example volcanic eruptions, glacier retreat, and changes in sea ice.

Energy transfer and exchanges of water and carbon fluxes in the soil–vegetation–atmosphere system need to be modelled and quantified. Imbalances of these impact directly the global climate indicated by stress on water resources, urban heat islands, floods and fires, drought periods, sea level rise, land & ocean heat waves, retreat of glaciers and sea-ice extent. Most of these can be measured accurately provided that realistic information is obtained from a list of Essential Climate Variables (ECV) among which are land surface temperature (LST), land surface emissivity (LSE) and evapotranspiration (ET). LST is defined as the radiative skin temperature and is broadly considered in agriculture (plant growth, water stress, crop yield, precision farming, early warning of vegetation stress, freezing scenarios), in hydrology (water cycle, catchment water, etc), and in meteorology. 
Whereas thermal observations at the global scale are largely carried out by moderate resolution sensors, a new generation of spaceborne platforms will measure the Thermal InfraRed (TIR) radiometry at a high spatial resolution - between 30 m and 100 m typically - with a frequent global coverage at noon and night time. The precursor is ECOSTRESS (ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station) flying today aboard the International Space Station (ISS), followed by TRISHNA (Thermal infraRed Imaging Satellite for High-resolution Natural Assessment), EAGLE-TIR and LSTM (Land Surface Temperature Monitoring). A design driver of these foreseen space missions is the management of water use in agriculture which represents 70 % of global resources, making sustainable irrigation a key issue. Automatic detection and mapping of irrigated farmland area is vital for many services in charge of water management. This will rely heavily on cutting-edge spaceborne TIR measurements that bring new insights, additionally to visible and near infrared observations, on irrigated areas that display lowest LST values at the peak of growth. Indeed, the global change imposes an implementation of more efficient irrigation practices at the scale of an agricultural plot for more sustainable water management practices. The increase of moisture within the soil after irrigation can be evaluated from the surface moisture estimated by radar but TIR observations are far more appropriate for monitoring vegetation water stress and irrigation at the agricultural field scale to adapt the irrigation practices to individual crops.  It is thus essential to develop new satellite TIR observation systems that reconcile high spatial resolution and high revisit capabilities into the same mission scenario as well as ensuring their interoperability to further mutualise benefits of these missions
This special session will present the specificities of the forthcoming space missions devoted to the measurement of the thermal infrared signal of the entire Earth system with the successive launches of TRISHNA, LSTM & EAGLE-TIR. Since these Earth observing systems will overlap in time, joint efforts are set up between space agencies (CNES, ISRO, ESA & NASA) to develop mutual research work and harmonize mission characteristics in terms of the user delivered products and associated Cal/Val strategy. This special session will permit to highlight the latest developments in this area through TIR mission presentations and related application domains.

This session provides the remote sensing community an update on government, commercial, and other space-based imaging spectrometers (hyperspectral imagers). This session creates an opportunity to inform the community on current missions providing data, for example, NASA's EMIT (Earth Surface Mineral Dust Source Investigation), ESA’s solar-induced fluorescence mission FLEX (Fluorescence Explorer), the German Aerospace Center’s DESIS (DLR Earth Sensing Imaging Spectrometer) and EnMAP (Environmental Mapping and Analysis Program), ASI’s PRISMA (PRecursore IperSpettrale della Missione Applicativa), JAXA’s HISUI (Hyperspectral Imager Suite), and China’s GaoFen-5 AHSI (Advanced Hyperspectral Imager). The session is also a venue to highlight newly launched sensors designed to operate alone on small satellites, for example, Planet’s Tanager, on larger independent platforms, on the space station, or as part of a constellation, for example Orbital Sidekick’s GHOSt-1 and -2 (Global Hyperspectral Observation Satellite). This venue also facilitates communication of the latest developments on forthcoming missions, such as ESA’s CHIME (Copernicus Hyperspectral Imaging Mission for the Environment) and NASA’s EAGLE-VSWIR (Explorer for Artemis Geology, Lunar, and Earth-Visible to Shortwave InfraRed). As data usage and societal impact are key goals of Earth observation missions, this forum also invites presentations on data systems, tools for data visualization and analysis, user needs and experiences, outreach to underserved communities and co-development of research and applications. As the pace of global change has increased the demand for science quality data to address new challenges, this session also seeks presentations on novel and emerging science and applications which have exceeded core objectives of imaging spectrometer missions. Discussions of leveraging national missions of varied measurement modalities, international cooperations, and partnerships between government, commercial, and other groups are encouraged.

Abstract:
As the volume of Earth observation data expands, Digital Twins of the Earth (DTE) and Ocean (DTO) have emerged as essential frameworks for understanding and predicting planetary systems. However, the true potential of these digital replicas relies heavily on global interoperability and robust architectural standards. This session issues a call for community progress, inviting presentations on the latest discoveries surrounding the creation, management, and scaling of digital twins globally.
Spearheaded by the IEEE P3501 working group, thhttps://2027.ieeeigarss.org/cct_proposal.php#is session will bring together pioneer architects, scientists, and users from national and international initiatives (including DestinE, DTO, NASA CZDT, and NOAA KM). We will explore successful pilot programs, architectural definitions, and the practical application of interoperability standards. By fostering novel partnerships across international borders this session aims to bridge the gap between technical machine-to-machine connections and human-level community sharing. Attendees will leave with actionable insights into how recommended practices can accelerate earth system modeling, improve coverage, and empower users to actively engage in building a more connected, sustainable planet.

One of our fundamental global imperatives in standard development is the continuous pursuit of opportunities to refine, stress-test, and grow the standard. This session serves as a deliberate forum to do exactly that for Earth and Ocean Digital Twins.
We are calling for presentations from the global community that showcase new discoveries, pilot executions, and operational successes in digital twin infrastructure and efforts. IEEE P3501 is uniquely positioned to bridge this conversation because our membership features the earliest experts who helped define the foundational interoperability and architecture for DTEs broadly. We have driven the concept from theory into executed pilots across national initiatives, and our rapidly expanding membership includes diverse participation across multiple continents and UN organizations.

Contributors will present on key themes & topics including:
Architectural Innovations: New approaches to the creation, lifecycle management, and federation of digital twins.
Interoperability in Action: Case studies from pilots demonstrating the value of standardized data sharing.
Standardizing the Future: Evaluating the current coverage of the IEEE P3501 standard including what it has considered, what it recommends, and where it must evolve based on active user feedback.
Community and Capacity Building: Tools and frameworks that foster active user engagement, enabling seamless information sharing at both the human and machine levels.

Target Audience: 
Earth observation scientists, systems architects, data engineers, climate modelers, and policymakers involved in national or global digital twin initiatives.

The Surface Water and Ocean Topography (SWOT) satellite mission, launched in December 2022, is a groundbreaking effort to observe inland, ocean, and coastal waters worldwide. It marks a significant advancement in hydrological sciences as the first satellite designed to investigate surface water in the global water cycle, and it provides an unprecedented, comprehensive view of Earth's freshwater bodies from space. Using Ka-band radar interferometry, SWOT delivers, for the first time, simultaneous, high-resolution maps of water surface elevation and inundation extent in rivers, lakes, reservoirs, and wetlands globally. The mission has a 21-day repeat cycle, offering multiple revisits, ranging from two to more than seven, depending on latitude.

For over a decade, the hydrologic remote sensing community has developed new methodologies and scientific frameworks to fully leverage the potential of SWOT data, which are now advancing our understanding of global water fluxes and fundamentally changing how we perceive and analyze surface water dynamics. Early investigations with this new dataset are already uncovering valuable insights into hydrologic processes across rivers, lakes, reservoirs, wetlands, as well as the terrestrial cryosphere, with performance exceeding the mission’s science requirements for some variables .

SWOT’s innovative capabilities, building on existing altimetry and imaging satellite missions, are unlocking new opportunities to deepen our understanding of water storage, river discharge, and flood dynamics.  These advances are expanding our ability to inform water resource management and assess the impacts of climate change. This session invites abstracts that present recent breakthroughs and novel discoveries, demonstrating how SWOT is advancing new frontiers in hydrology and enriching our understanding of Earth’s surface water systems.

Quantum technologies are increasingly being explored for Earth Observation (EO), with potential applications in machine learning, optimization, inverse problems, and scientific data analysis. However, significant challenges remain in understanding when and how quantum methods can provide practical benefits under realistic hardware, resource, and data constraints. This session will bring together researchers working on the theory, algorithms, benchmarking, and application of quantum computing for EO.

Topics of interest include quantum machine learning, quantum optimization, hybrid quantum-classical workflows, quantum kernels and feature maps, explainable and interpretable quantum AI, benchmarking methodologies, resource estimation, error mitigation, and hardware-aware algorithm design. Particular emphasis will be placed on rigorous evaluation frameworks that enable fair comparisons between classical and quantum approaches, as well as on methods for understanding how EO information is represented, processed, and exploited within quantum models. This includes emerging approaches to interpreting quantum embeddings, quantum feature spaces, and learned representations of remote sensing data.

The proposed session complements the IGARSS 2027 Special Theme on Computing Paradigms and Quantum Technologies for Earth Observation by focusing on the scientific foundations and assessment frameworks required to translate quantum computing advances into practical EO capabilities. The session will foster collaboration between quantum scientists, computer scientists, and EO researchers, helping to establish a shared understanding of opportunities, limitations, and future research directions.

As an emerging and highly interdisciplinary research area, the topic addresses the growing need for rigorous, reproducible, and physically grounded research in quantum computing for EO. The session will provide a venue for discussing pathways toward quantum utility, improving the interpretability of quantum models, identifying impactful EO use cases, and strengthening collaboration across the remote sensing and quantum technology communities.

Satellite SAR has been playing an important role in earth observation for decades. However, due to constrains in orbit directions, revisit time, altitudes, etc., space-borne SAR systems are difficult to meet observation requirements in same cases. UAV-borne SAR systems are developing rapidly in recent years, which provide an important complementary observation tool for earth observation.
Geohazards happen frequently in China and all over the world. The Jilong mudflow disaster happened at the border region of Himalaya between SW China and Nepal on Aug. 26, 2026, causing heavy losses of human lives and properties. Common remote sensing techniques were difficult to acquire data due to weather condition, rugged mountainous terrain, and satellite orbit geometries and revisit intervals. However, UAV-borne SAR systems could fly at any place in any time with arbitrary flight directions and flexible flight parameters, providing observation data in real-time. These advantages could meet most requirements proposed by geoscientists. 
Much attention on UAV-borne SAR has been paid by Chinese authorities, academic communities and industries. Several organizations and authorities such as National Natural Science Foundation of China (NSFC), Ministry of Science and Technology of China (MOST) have set up projects on UAV-SAR system development and applications. Commercial activities in this field are also promising. Many UAV-borne SAR systems are developed by enterprises in China, along with import of some foreign UAV-borne SAR systems.
The purpose of this proposed theme is to invite the UAV-borne SAR researchers and geoscientists who are interested in this topic or users from China and abroad to share their experiences and recent advance in system design, algorithms development, and applications in geosciences.

SAR systems on UAVs, high-altitude long endurance (HALE) unmanned aircraft, and mobile mapping platforms, such as automobiles, have increasingly gained attention also within the geoscience community. Small SAR systems deployed on such platforms offer complementary properties in terms of revisit time, operational flexibility, and observation capabilities compared to spaceborne and conventional airborne SAR systems. On the other hand, compared to stationary terrestrial radar/SAR systems, the increased synthetic aperture size of UAV/mobile mapping SAR systems allows to obtain a higher spatial cross-range resolution also for quasi-terrestrial observation geometries.
These complementary properties of UAV-based and mobile mapping SAR systems open a large field of potential applications, some of which are addressed within the scope of this session including high-resolution DInSAR based measurements of surface displacements, monitoring of vegetation / agricultural crop, change detection.
From a system point of view, these agile SAR platforms require not only new compact SAR system designs, but also compact and innovative high-performance navigation using smaller INS/GNSS systems, in some cases combined with vision systems, as well as adequate SAR imaging algorithms and DInSAR processing chains adapted to the potentially non-linear sensor trajectories and partial aperture synthesis common to UAV/mobile mapping SAR systems and application.
UAV-borne SAR systems enable experimental formation flying, making them an important tool for developing and testing bistatic and multistatic SAR mission concepts, including synchronization for future spaceborne SAR missions.
This community-contributed session aims at giving an insight into recent state-of-the-art UAV- and mobile-mapping-based SAR systems, methods, and applications developed with a focus on geoscience applications.
After our successful invited/community-contributed sessions on this topic during IGARSS 2021/2022/2023/2024/2025 and IGARSS 2026, we would like to keep track of this topic, providing insight into the latest technological developments with small SAR systems on UAV, HALE/HAPS, and mobile-mapping platforms.
The session typically covers a number of novel systems and UAV/mobile mapping platforms of different sizes, types (fixed-wing and VTOL UAVs, cars), and a range of methods, and applications such as repeat-pass differential SAR interferometry for displacement measurements, change detection, and tomographic configurations.
We believe (and this has been confirmed by repeatedly reaching a large audience in our sessions) that our session topic, UAV/mobile mapping-based SAR systems, methods, and applications, is of very high interest to the geoscience and remote sensing community already now and will continue to be so in the future.

Scope
 
This Community Contributed Theme (CCT) focuses on intelligent perception, semantic understanding, and structured representation of urban geographic scenes. It aims to connect urban objects with their spatial relationships, functional roles, and contextual semantics, bringing together remote sensing, geographic information science, computer vision, and urban research. Contributions may use satellite and UAV imagery, street-view observations, vector maps, and complementary geospatial information. Topics include urban scene perception and object geometry; spatial relation modeling and remote sensing scene graphs; automatic relation annotation and geographic knowledge constraints; urban functional understanding through graph learning and spatial reasoning; vision-language and foundation models for geographically grounded scene understanding; hierarchical scene graphs and geographic knowledge graphs; and datasets and evaluation methods addressing spatial consistency, cross-city generalization, uncertainty, and interpretability.

Importance. 

Urban scene interpretation requires more than identifying individual objects. Buildings, roads, vegetation, and open spaces form spatial configurations that provide essential evidence for understanding neighborhood characteristics and urban functions. This theme explores how observations can be transformed into structured representations that explicitly capture these relationships and support spatial reasoning. By integrating scene graphs, geographic knowledge, and multimodal models, it seeks to advance the interpretation of urban form, function, and context beyond object-level recognition. Key challenges include reliable spatial relation annotation, geographic consistency, cross-source and cross-scale integration, and the evaluation of scene-level representations. The resulting methods can support urban land-use analysis, environmental assessment, urban renewal, sustainable planning, and digital twin applications.

Complementarity with General and Special Themes. 

This CCT complements L.6 Urban and Built Environment by emphasizing structured scene representations and spatial reasoning rather than urban applications alone. While T.11 focuses on object recognition and T.18 on multimodal fusion, this theme examines how object-level outputs can be organized into spatially consistent relationships and meaningful scene-level semantics. It also complements SP.5 Foundation Models and Embeddings for Earth Observation and SP.6 Datasets, Benchmarks, and Evaluation Methodologies for Earth Observation by emphasizing geographic grounding, explicit spatial structure, and the evaluation of urban scene representations. We welcome coordination with related CCT organizers to consolidate overlapping topics and strengthen thematic coherence.

Scope: 
Earth observation (EO) supports rapid mapping of hazard extent and disaster damage. Yet similar exposure can produce different outcomes because buildings, infrastructure, population distribution, and urban functions shape vulnerability and recovery. Hazard observations and urban representations are often developed separately, limiting their use across the disaster management cycle. This CCT seeks contributions that connect measurements of hazardous events with representations of the cities they affect. It considers how satellite and airborne imagery, SAR, LiDAR, street-level observations, and ancillary records can be analysed with AI to describe urban assets and their relationships. Urban digital twins offer a setting for combining these descriptions with hazard simulations and population information, testing alternative actions, and updating assessments as new observations arrive. Research should address decisions before, during, or after an event. Topics include: 
•	Estimation of risk-relevant building properties and neighbourhood context, including structural characteristics, materials, occupancy, and spatial configuration.
•	Fusion of hazard measurements, urban geometry, and semantic information to model damage mechanisms and dependencies between urban assets.
•	Analysis of how local damage translates into population exposure, loss of access, interruption of services, and unequal societal consequences.
•	Observation-driven city models for preparedness scenarios, emergency operations, and tracking the restoration of buildings and urban functions.
•	Validation across cities and hazards, with uncertainty carried from data acquisition to impact estimates; synthetic observations for poorly sampled environments.
•	Coordinated analytical workflows that connect data acquisition, model execution, and interpretation, including agent-based approaches to complex geospatial tasks.

Importance: 
Disaster management requires more than identifying where damage occurred: it must explain who and what is at risk, how urban characteristics influence impacts, and which interventions improve resilience. High-resolution EO, multimodal AI, and 3D urban modelling create opportunities to answer these questions at building and neighbourhood scales. Progress requires collaboration among remote sensing, urban modelling, hazard science, and social vulnerability research, with evaluation against observed impacts and practical decision needs.

Complementarity with General and Special Themes:
The closest theme is D.7 (Risk and Disaster Management). This CCT adds a focused interdisciplinary question: how does EO-derived knowledge of a city improve predictions and explanations of disaster consequences? L.6 (Urban and Built Environment) contributes urban characterisation, T.18 (Multimodal Data Fusion) contributes integration methods, and A.6 (Digital Twins) contributes computational representations. Their intersection is evaluated through damage, disruption, and recovery. SP.2 addresses hazardous environments, whereas the proposed theme centres on the urban systems affected. SP.3 and SP.5 contribute agentic and foundation-model approaches where these demonstrably advance urban disaster analysis.

Wildfires emerge from coupled interactions among climate, vegetation, topography, land use, and human activities, and their consequences extend well beyond burned areas. Advances in Earth observation (EO), geospatial data science, and environmental modeling are creating new opportunities to investigate wildfire processes and improve management across diverse ecosystems and communities. Yet information remains fragmented across sensors, spatial and temporal scales, scientific disciplines, and stages of fire activity. This session brings together the IGARSS community to explore how multi-source geospatial observations and analytical methods can advance wildfire science and management, linking physical processes, predictive understanding, practical interventions, and long-term environmental and societal outcomes.

The scope extends across the wildfire lifecycle, including fuel characterization and fire-prone landscape mapping, ignition susceptibility, active-fire detection, risk and spread forecasting, smoke emissions and air-quality impacts, and post-fire ecological and community recovery. Fire management forms an important part of this scope, encompassing fuel treatments and prescribed fire, preparedness and resource allocation, suppression support, evacuation and escape-route planning, and assessment of management effectiveness. Relevant studies may operate from hours and days for monitoring and emergency response, through seasonal and annual risk outlooks, to decadal and century-scale projections of fire regimes and climate-related change, recognizing that forecasting and long-term scenario analysis require different methods and validation strategies.

Geospatial evidence may include optical, thermal, radar, hyperspectral, and LiDAR observations from satellites or airborne platforms, together with weather data, climate projections, vegetation and fuel inventories, topography, ground observations, infrastructure, population distribution, and socio-environmental indicators. Research directions include data quality assessment, multimodal integration, spatiotemporal harmonization, mapping and change detection, benchmark development, and uncertainty-aware data products. The methodological scope encompasses GIS and spatial statistics, process-based fire and ecosystem modeling, classical and deep machine learning, probabilistic and Bayesian inference, and hybrid physical and data-driven approaches.

Across these applications, the session encourages further work on interpretability and explainable AI, uncertainty quantification, probability calibration, robustness, and generalization across regions and changing environmental conditions. It also highlights regional and community inequalities in wildfire exposure, access to warnings and relevant information, emergency response capacity, and post-disaster recovery. Research involving Indigenous and local communities, including community knowledge, participatory evaluation, and data sovereignty, can broaden the scientific and social relevance of geospatial wildfire studies.

The session aligns directly with IGARSS General Theme D.7, Risk and Disaster Management, and complements areas in AI and Big Data, multimodal data fusion, forest and vegetation applications, and atmospheric remote sensing. It also connects with Special Theme SP.3, Artificial Intelligence Ethics and Governance in Remote Sensing, through research on transparency and equity. Rather than privileging a particular algorithm or stage of wildfire activity, the session offers an interdisciplinary setting for connecting advances in observation, modeling, and management across geographic and temporal scales.