Huge uncertainties remain when it comes to where and when food is grown, how much will be produced, early warning of shortfalls, yield gaps and drivers, impacts of current land cover change and conflict on production. Yet big-dollar food security decisions are constantly being made that could benefit from enhanced, timelier and more accurate crop information. The new era for satellite technology capabilities and integration into existing monitoring systems and agricultural decision-making processes across sectors offer big promise in resolving most of these uncertainties. The project will start with a prototype in Kenya and an initial study in Mexico before being set up as an open-source, portable and cloud-agnostic machine learning tool, dubbed EO-FARM, that can be deployed in other regions.

The goal of this work is to develop semi-automated, scalable remote sensing-based datasets and information products for maize and wheat conditions and yield assessment. First working with Kenya’s Ministry of Agriculture, Livestock, and Fisheries (MoALF) and county governments, the team will use the EO-derived dataset (within-season crop maps) required for yield assessments to guide sample design to significantly enhance crop condition and yield assessments. This will not only significantly reduce the cost of yield assessment, it will accelerate the process to ensure early pay-outs to the now estimated 300,000 beneficiaries of the Kenya Crop Insurance Programme and will support its further expansion. In order to test the generalizability of the mapping and monitoring approach in a very different small holder context, the approach developed in Kenya will then be adapted and evaluated through a pilot study in Mexico, working closely with the International Maize and Wheat Center (CIMMYT).

Furthermore, to enable scaling and deployment of the approach in additional regions, our team will work with the support of policymakers in Kenya and Mexico to define the financial investment required, to estimate the costs and savings associated with this new method compared to traditional crop assessment methods, to analyze the benefits of timely and accurate tools for policy implementation, and to articulate the value of agricultural and insurance technology development. We will also leverage to the extent possible relevant and complimentary projects through the GEOGLAM and other NASA Harvest networks in order to bolster capacity development.