Agricultural drought analysis and famine early warning with the FEWS NET land data assimilation system

Global and regional changes related to water resources and agriculture affect food and fresh water security. To mitigate and adapt to these changes it is important to quantify how climate variability and change has impacted agricultural production and water resources. This research examines trends in supply and demand for moisture availability in rain-fed agro-pastoral regions. With a focus on the Sahel region of Africa we ask the following two questions: (1) Do land surface models, forced with remotely sensed data, detect the spatio-temporal patterns of agricultural drought over the past 30 years? (2) How have these trends impacted agricultural productivity and food security? To explore implications of hydro-climatic (e.g. precipitation and potential evapotranspiration (PET)) change on agriculture, we use the Famine Early Warning Systems Network Land Data Assimilation System (FLDAS) forced with rainfall from the University of California Santa Barbara Climate Hazards Infrared-Precipitation with Stations (CHIRPS) dataset (1981-present) and 10 km meteorological data (wind, temperature, radiation, humidity) from Cheney and Sheffield, released in 2012, for continental Africa north of 10S (1979-2008). We examine trends in model outputs (e.g. soil moisture and evapotranspiration (ET)), as well as composite indices, such at the evapotranspiration-rainfall ratio and water requirement satisfaction index (WRSI). We compare these results to the Normalized Difference Vegetation Index (NDVI) and microwave soil moisture. Finally, we examine how the different model outputs and composite indices relate to reported trends in agricultural production. Preliminary results show that the FLDAS Noah3.2 and geoWRSI models accurately estimate near surface (0-40cm) soil moisture anomalies as defined by microwave and in-situ observations across the Sahel. With respect to ET, the literature reports that vegetation biomass, as indicated by NDVI, has increased in conjunction with rainfall (i.e. ';re-greening' of the Sahel). However, at least one study has reported a downward trend in modeled ET in the Sahel. Preliminary results indicate that the spatial and temporal patterns of transpiration in Noah3.2 and geoWRSI are highly sensitive to their respective vegetation parameterizations. Our model runs explore the timing and magnitude of ';crop' vegetation parameters, such as LAI and green vegetation fraction, to assess agricultural drought trends and confirm findings from previous work.