Satellite Eyes on Water: How Earth Observation and AI Supports Agriculture
Machine learning delivers field-scale soil moisture maps from satellite data.
Imagine trying to manage irrigation across hundreds of hectares with only a rough guess about where your soil is actually dry. This is the challenge many European farmers face today. Existing satellite-based soil moisture maps work at resolutions of 1 kilometer or larger—meaning an entire farm appears as just a few pixels. For farmers trying to optimize water use and crop yields, this simply isn’t detailed enough to make field-by-field decisions about when and where to irrigate.
Researchers at the National Observatory of Athens are addressing this challenge by combining data from multiple satellites with ground measurements from 113 open-access monitoring stations across Europe and soil moisture sensors that will be deployed through the UNIVERSWATER project. Using the latest artificial intelligence technology to analyze patterns in the satellite data, the team has developed algorithms that estimate soil moisture at 10-meter resolution—fine enough to distinguish between individual fields and even different zones within the same field. This research is going to explore foundation models, i.e. models pre-trained on large satellite datasets, to further improve accuracy and expand the system’s capabilities across different agricultural regions. The system combines radar signals (i.e. Sentinel-1), optical satellite images (i.e. Sentinel-2) that capture vegetation health, and weather data tracking recent rainfall and temperature variations. Testing across diverse European landscapes—from croplands to grasslands—showed the model could accurately predict soil moisture levels, helping farmers identify which areas need water and which don’t.
Within the UNIVERSWATER project, this work provides practical pathways for pan-European soil moisture monitoring systems for croplands. As climate change makes water resources more precious, knowing exactly where and when to irrigate becomes increasingly important. This research contributes to UNIVERSWATER’s mission of optimizing sustainable water resource management for agriculture through Earth Observation and Artificial Intelligence.
