Product Updates: Streamflow

Streamflow forecasting traditionally requires gauges or costly and basin-specific hydrological modeling. We are developing machine learning models to forecast streamflow in any basin given weather and landscape characteristics. We are blown away by the early results: as shown in the graphs below, our early models are already predicting streamflow in ungauged basins with higher accuracy than physical models! We are continuing to iterate and calibrate these models, with a streamflow forecasting release this spring. 

Product Updates: Land Cover 

Land cover change causes significant impacts for ecosystems and communities. With our land cover classifications, users can analyze over 20 classes to monitor changes in land use and landscape conditions, such as developed land or deforestation, as shown in the image below. 

• Reimagining the Potential of Earth Observations for Ecosystem Service Assessments
• Designing an Environmental Impact Bond for Wetland Restoration in Louisiana

• Thinking Ahead in a Thirsty State 💧
• Measuring What Matters: Verifiable forest monitoring and REDD+ 🌳
• Denoising Sentinel-1 Radar Images with Neural Networks 🤖