Each year, billions of dollars are invested in agricultural systems with the aim of reducing poverty, raising productivity, and closing yield gaps. Our ability to improve these systems, however, is limited by our ability to assess the effectiveness of the policies or programs put into place to accelerate their development.  Unfortunately, assessing the impacts of new investments or policies on agricultural performance is complicated by the spatial diversity of growing conditions, and the heterogeneity of management systems. To overcome these challenges CIAT and USAID have developed an innovative workflow and computational framework for generating weather-adjusted, site-specific benchmarks for estimating the impact of agricultural programs. Their approach leverages the power of data-rich remote sensing and weather observations, and the pattern recognition capacity of convolutional neural networks.  It is designed for maximum flexibility, can be adapted to identify impacts across a broad range of geographies and agricultural systems, and can be crafted to work with nearly any EO platform.  Moreover, it can be applied anywhere, including in areas deemed too difficult to reach, or in areas which might otherwise be too dangerous for traditional impact assessment tools.

 

In this presentation we will provide a brief overview of the System for Evaluation of Impact using Remote Sensing workflow, describe the results associated with several development programs in Senegal and Nepal, and identify future applications for the new technology.