On October 15-16, 2026, the Global Food-and-Water Security-support Analysis Data (GFSAD) Project Workshop was held at the U.S. Geological Survey (USGS) Flagstaff Science Center in Flagstaff, Arizona. The workshop brought together scientists, researchers, and collaborators from the USGS, Bay Area Environmental Research Institute (BAERI), San Diego State University (SDSU), and Northern Arizona University (NAU) to review accomplishments, discuss ongoing research, showcase new scientific developments, evaluate products and publications, and chart future directions for the GFSAD program.

The GFSAD project is a major NASA-funded Earth Action Agriculture initiative, formerly supported through NASA's Making Earth System Data Records for Use in Research Environments (MEaSUREs) Program, with complementary support from the USGS. The project aims to develop global agricultural information products at 30-meter and increasingly 10-meter spatial resolutions using satellite remote sensing, cloud computing, advanced machine learning, and artificial intelligence. Its overarching goal is to provide scientifically rigorous geospatial information on multiple cropland products, irrigation, crop water productivity, and agricultural dynamics to support global food security, water security, climate adaptation, and sustainable resource management.

The workshop highlighted the remarkable progress made by the GFSAD program over the past decade. A major foundational achievement has been the development of the Global Cropland Extent Product (GCEP30), the first global cropland map at 30-meter resolution derived from Landsat time-series data. This effort involved multiple international mapping teams and independent validation groups, resulting in numerous NASA Earthdata releases, journal articles, a USGS Professional Paper, and a body of research that has received thousands of scientific citations worldwide. Building on these foundations, the project has expanded into a new generation of products addressing cropland extent, crop intensity, irrigated and rainfed agriculture, crop water productivity, and AI-enabled Earth observation analytics.

One of the major themes of the workshop was the transition from traditional satellite image analysis toward geospatial foundation models and artificial intelligence. Multiple presentations demonstrated how Google DeepMind's AlphaEarth Foundation satellite embeddings, consisting of 64-dimensional AI-generated representations at 10-meter resolution, are transforming cropland mapping. These embeddings integrate information from Landsat, Sentinel-1, Sentinel-2, ALOS-PALSAR, elevation models, climate datasets, and other Earth observation sources into compact, analysis-ready geospatial information layers. Workshop discussions emphasized that foundation models represent a paradigm shift in agricultural monitoring, enabling more accurate, scalable, and transferable mapping of cropland extent, irrigation status, crop intensity, and crop types worldwide.

Significant attention was devoted to the development of next-generation 10-meter agricultural products. Researchers presented ongoing work on Global Cropland Extent Product Version 2 (GCEP10), which seeks to improve upon earlier 30-meter products by reducing omission and commission errors while better capturing fragmented smallholder agriculture and rapidly changing agricultural landscapes. In this context, the Landsat-Derived Rainfed and Irrigated Area Product at 30 m for the Conterminous United States for the Year 2020 Version 2 (LGRIP30 CONUS 2020 V2) was discussed. Participants highlighted both the scientific advances embodied in the product and the recognition it received through the ASPRS ESRI Award for Best Scientific Paper in GIS.

The workshop also celebrated the release of the Landsat-derived Global Crop Intensity Product (LGCIP30) for 2020. These products quantify single-, double-, triple-, and continuous-cropping patterns across multiple world regions and provide important insights into agricultural intensification and land-use dynamics. Results from Brazil, southwestern South America, North America, and Africa demonstrated encouraging classification accuracies while also revealing important challenges associated with annual agricultural change, validation datasets, and temporal transferability of machine-learning models.

Africa emerged as a major focus area for advancing agricultural mapping from 30-meter to 10-meter resolution through the use of AI-derived Google DeepMind AlphaEarth Foundation (AEF) data. Researchers highlighted efforts to synthesize multiple existing cropland products, quantify mapping uncertainties, and develop AI-assisted classifications using AlphaEarth embeddings and machine-learning approaches. Two major Africa products were completed and presented: (1) the AlphaEarth Foundation Derived Global Cropland Extent Product at 10 m for Africa for the Year 2020 Version 2 (AEF-GCEP10 Africa 2020 V2), and (2) the AlphaEarth Foundation Derived Global Rainfed and Irrigated Cropland Product at 10 m for Africa for the Year 2020 Version 2 (AEFGRIP10 Africa 2020 V2). These efforts are laying the foundation for globally consistent 10-meter cropland, irrigated-area, and rainfed-area products by integrating existing datasets, AI-derived embeddings, cloud computing, machine learning, and rigorous validation approaches. Results from Africa demonstrated the strong potential of these methods for improving agricultural and water-use assessments, particularly in regions dominated by fragmented smallholder farming systems.

South Asia emerged as another major focus area. Presentations detailed the development of SouthAsiaCEP10 and SAIR10 products, which use AlphaEarth embeddings, Sentinel imagery, and advanced machine-learning methods to map cropland extent and distinguish irrigated from rainfed agriculture at 10-meter resolution. Researchers highlighted the challenges of irrigation mapping in South Asia, including monsoon cloud cover, mixed cropping systems, paddy rice cultivation, fragmented farm ownership, seasonal irrigation practices, and highly heterogeneous agricultural landscapes. Preliminary results indicate substantial improvements in identifying cropland and irrigation patterns while reducing uncertainties associated with earlier 30-meter products.

Another important area highlighted during the workshop was crop water productivity (CWP), often referred to as "crop per drop." The WGSC team examined methods for integrating field measurements, satellite remote sensing, and evapotranspiration estimates to quantify agricultural productivity relative to water consumed. Studies conducted in California's Central Valley demonstrated the use of PlanetScope, Sentinel-2, Landsat, and OpenET datasets to characterize crop water productivity and estimate potential water savings. Findings indicated that improvements in crop water productivity could lead to substantial agricultural water conservation while providing a scientific basis for evaluating future water-banking opportunities.

The workshop also highlighted the project's commitment to open science, product dissemination, and stakeholder engagement. Presentations described the GFSAD project website, croplands visualization portal, NASA Earthdata distribution systems, and data-access infrastructure maintained jointly by USGS and NASA. Participants emphasized the importance of making scientifically validated products freely available to researchers, water managers, agricultural agencies, policymakers, and international organizations worldwide.

Looking ahead, workshop participants identified several strategic priorities. These include completion of global 10-meter cropland extent and irrigation products, expanded utilization of AI-driven geospatial foundation models, development of annual and seasonal agricultural monitoring systems, improved crop-intensity and crop-type mapping, enhanced crop-water-productivity assessments, and broader integration of Earth observation datasets for food-and-water security analysis. Planned outputs include peer-reviewed publications, NASA Earthdata releases, algorithm theoretical basis documents, user guides, and enhanced web-based visualization and dissemination platforms.

Overall, the workshop demonstrated that the GFSAD project has evolved into one of the world's most comprehensive Earth observation initiatives focused on agriculture, irrigation, and agricultural water use. Its growing scientific importance and global relevance were underscored by the widespread use of GFSAD data products, with users from 94 countries accessing GFSAD resources during the period September 2025 through August 2026. Through the integration of satellite remote sensing, artificial intelligence, cloud computing, field observations, and open-science principles, GFSAD is providing critical information needed to monitor agricultural systems, improve water-use efficiency, strengthen food security, and support evidence-based decision-making from local to global scales.

Western Geographic Science Center (WGSC), United States Geological Survey (USGS) Flagstaff

Prasad Thenkabail

Itiya Aneece

Daniel Foley

Adam Oliphant (also SDSU)

San Diego State University (SDSU)

Trent Biggs

Violet Navarrete

Sirena Rodriguez

Zoe Goughan

Adam Oliphant (also USGS)

Bay Area Environmental Institute (BAERI) and NASA AMES

Pardhasaradhi Teluguntla

Richard McCormick

Sumalika Biswas

Northern Arizona University (NAU)

Jake Lawton (NAU and BAERI)

Cummins-Johnson, Nicholaus R (NAU and BAERI)