A New Global Reference Evapotranspiration Reanalysis Forced by MERRA2: Opportunities for Famine Early Warning, Drought Attribution, and Improving Drought Monitoring

Recent work has demonstrated that improving the representation and analysis of atmospheric evaporative demand (E₀; the “thirst of the atmosphere”) permits vital early warning of emergent drought conditions, improves ongoing drought monitoring, and allows explicit attribution of drought into its evaporative drivers, with further benefits accruing in sub-seasonal to seasonal drought forecasting. The FEWS NET science community in general and NOAA in particular have been globalizing these E₀-related efforts to improve drought and famine early warning in food-insecure countries, starting with the development of a new, long-term reference evapotranspiration reanalysis as an estimator of E₀ at fine space and time scales across the globe.

We estimate E₀ using the FAO-56 model of Penman-Monteith reference evapotranspiration, driven by radiative and meteorological forcings from NASA’s MERRA-2 reanalysis dataset. The E₀ output is then spatially downscaled to match sub-grid variability provided by the International Water Management Institute (IWMI) monthly climatologic potential evaporation grids. The resulting E₀ reanalysis spans the globe at a 0125-degree spatial resolution, and is daily from January 1, 1980, to within a month of the present. The E₀ reanalysis verifies well against station-derived observations across Africa.

As drought is driven by, and/or reflected in, variations in E₀, the question arises, “what drives E₀ variability across the globe?” We answer this by decomposing the temporal variability of E₀ into all of its physical drivers through a mean-value, second-moment uncertainty analysis in which both the sensitivity of E₀ to its drivers and their observed variabilities are accounted for. Amongst other things, this rigorous variability analysis informs us as to which drivers are the most important to represent well in estimating E₀.

We use the new E₀ reanalysis in estimating the Evaporative Demand Drought Index (EDDI), which has been shown to provide early warning of agricultural and hydrologic drought across the continental US. Uniquely, EDDI leverages relationships of a fully physical E₀ and actual evapotranspiration, measuring E₀’s physical response to surface drying due to land/atmosphere interactions, without need for precipitation or surface moisture data. EDDI provides a perspective of drought from the atmospheric demand side previously missing from the “convergence of evidence” approach to drought monitoring and provides a valuable planning window for land and water resource managers and agricultural producers.

The new E₀ reanalysis permits explicit attribution of changes in the drought signal in E₀ to its individual drivers. We outline the methodology and results of this uncertainty analysis and attribution study across the space-time domains of established droughts, demonstrating the drivers of E₀ variability and the evaporative drivers of crop stress and food insecurity, concentrating on Famine Early Warning Systems Network (FEWS NET) food-insecure countries during established drought periods.

Here, we present these developments so far, including the development of a new global E₀ reanalysis and its verification, and the rigorous variability analysis across the globe. Further, we demonstrate the power of E₀ in (i) driving EDDI for famine early warning, and (ii) attributing the drivers of drought from the evaporative perspective.