Among the pool of climate variables affected by anthropogenic climate change, air temperature increases are some of the most well-established and most likely outcomes. Air temperatures are well-represented by the current generation of climate models, and the various models tend to agree reasonably well: during the 21st century many (but not all) areas will experience substantial increases in air temperatures. For example, increases are likely to be 1.5 times greater in Africa than at the global level (IPCC AR4). Air temperatures are also easily measured using thermometers, and tend to be well-correlated at long distances; this make observing changes in air temperatures relatively straight forward.
However, potential evapotranspiration is not so clear-cut. PET is a very important hydrologic variable which strongly impacts the rate at which water evaporates from the surface of the earth or transpires through the stomata of plant leaves. For a given amount of precipitation, a higher PET will tend to result in less available water. For applications interested in estimating the evaporating power of the atmosphere across diverse landcover types, reference evapotranspiration (ETo) is a common replacement for PET. This is because ETo calculations refer to the PET from the same reference surface and so can be directly compared (Allen 1998). The international standard for ETo calculations is the Food and Agriculture Organization of the United Nations Paper No. 56 Penman-Monteith equation (FAO56-PM). FAO56-PM uses only climatic parameters, relating net radiation, wind speed, temperature, and relative humidity. The equation is composed of one component driven by net radiation (mm∙day-1) and a second component driven by vapor pressure deficit (mm∙day-1); the sum of the components, accumulated over the day, provide ETo estimates in mm∙day-1.
Given the need to project regional and seasonal changes to atmospheric water demand, this study identifies where temperature is a suitable driver of monthly ETo estimates. In this study, reference evapotranspiration is calculated using the CIMIS version of the Food and Agriculture Organization of the United Nations Paper No. 56 Penman-Monteith equation (FAO56-PM) (Allen 1998; Walter 2000; CIMIS 2009). Using 3-hourly GLDAS 0.25 degree resolution climate parameters, daily ETo estimates for 2001-2009 are calculated. The net radiation (Rn) and vapor pressure deficit (VPD) components of ETo are held separate. Correlations between temperature and each of these components initially identify where (spatially) and how (to which parameters) temperature-driven simulations may be acceptable. Tight coupling between vapor pressure deficit and temperature and a substantial year round proportion of the ETo totals being accounted for by the VPD component, indicate high simulation potential in the Sahel and other semi arid regions. The vapor pressure component is further broken down into the saturation vapor pressure (SVP) and water vapor pressure parameters. The theoretical relationship between temperature and SVP, and statistically-derived relationship between temperature and water vapor at a grid cell basis (2001-2009) drive simulations of the VPD component of ETo. Simulations are compared to the historical estimates and where skillful, IPCC AR4 temperature simulations (1950-2000) and projections (2001-2050) are used to estimate regional and seasonal changes to atmospheric water demand.