Using satellite-based data to support agricultural decision-making in Rwanda

Addressing the grand challenge of achieving global food security for a growing population within land and other resource constraints requires significant advances in production efficiency and sustainability. Systems relying on satellite-based data have the near-term potential to provide stakeholders with information on crop production, status, and predicted yield that is timely, covers large areas, captures spatial and temporal detail, and can be obtained at low cost, potentially contributing to better decisions. We develop methods and tools to acquire and process geospatial information from multiple data streams and apply to analyses of Rwandan agriculture. We combine satellite imagery with images collected using unmanned aerial vehicles and field survey data and apply machine learning techniques to provide rapid, consistent image processing and analysis. These data are used to estimate area cultivated, identify major crops, and assess crop growth over time during the growing season.