Using the SPEI to Estimate Food Production in East Africa

Abstract:

The Famine Early Warning Systems Network (FEWS NET) monitors critical environmental variables that impact food production in developing countries. Due to a sparse network of observations in the developing world, many of these variables are estimated using remotely sensed data. As scientists develop new techniques to leverage available observations and remotely sensed information there are opportunities to create products that identify the environmental conditions that stress agriculture and reduce food production.

FEWS NET pioneered the development of the Climate Hazards Group InfraRed Precipitation with stations (CHIRPS) dataset, to estimate precipitation and monitor growing conditions throughout the world. These data are used to drive land surface models, hydrologic models and basic crop models among others. A new dataset estimating the reference evapotranspiration (ET₀) has been developed using inputs from the ERA-Interim GCM. This ET₀ dataset stretches back to 1981, allowing for a long-term record, stretching many seasons and drought events. Combining the CHIRPS data to estimate water availability and the ET₀ data to estimate evaporative demand, one can estimate the approximate water gap (surplus or deficit) over a specific time period. Normalizing this difference creates the Standardized Precipitation Evapotranspiration Index (SPEI), which presents these gaps in comparison to the historical record for a specific location and accumulation period.

In this study we evaluate the SPEI as a tool to estimate crop yields for different regions of Kenya. Identifying the critical time of analysis for the SPEI is the first step in building a relationship between the water gap and food production. Once this critical period is identified, we look at the predictability of food production using the SPEI, and assess the utility of it for monitoring food security, with the goal of incorporating the SPEI in the standard monitoring suite of FEWS NET tools.