Recents declines in potential evapotranspiration over South Africa: potential causes and implications for maize yield and irrigation demand

Abstract:

Recent work has identified a 31-year (1979-2010) decline in potential evapotranspiration (PET) during the maize growing season in South Africa, the world's 9th largest producer of that crop. Using a newly-developed, bias-corrected meteorological forcing dataset, we apply an attribution analysis to identify the relative role of four key physical drivers (temperature, net radiation, vapor pressure, and windspeed) in reducing atmospheric demand for water. We conduct a statistical analysis to correlate changes in these four key drivers to potential causal mechanisms, including atmospheric aerosol concentration and changes in the extent of irrigated cropland, which we identify using a novel, high accuracy landcover dataset. Finally, we use the DSSAT maize model, together with counter-factual climate scenarios, to investigate the implications of the PET decline on maize yields and maize irrigation demand. This study illustrates how improved meteorological data, better landcover maps, and crop simulation can be combined to 1) improve understanding of the linkages between the land surface and atmosphere, and 2) help inform crop and irrigation management under changing climates.