Evaluation of Different Methodologies to Estimate Reference Evapotranspiration from PRISM Dataset in the Southeastern USA.

The demand for spatial climate datasets has grown in the past years due to the need to have widely spread information. However, the weather variables availability is restricted in these datasets, restricting their use. Reference evapotranspiration (ET_O) is an important agrometeorological parameter used in climatological studies, water resources planning, and irrigation scheduling. FAO considers the Penman-Monteith equation as the standard equation to estimate ET_O. However, this methodology utilizes weather variables not always available in gridded datasets. A dataset must have adequate spatial and temporal resolutions, in addition to the required variables to properly estimate ET_O. The main goal of this study was to determine the best equation to estimate ET_O in the Southeastern USA using the Parameter-elevation Regressions on Independent Slopes Model (PRISM), which is a spatial dataset with daily temperature for the USA. Daily ET_O calculated between 2006 and 2013 by FAO Penman-Monteith (FAO-56 PM) using observed weather data was compared to ET_O estimated by six empirical equations: Priestley and Taylor, Turc, Makkink, Stephens-Stewart, Valiantzas, and Hargreaves using temperature from PRISM. The required data to estimate FAO-56 PM were obtained from 92 automated weather stations in Florida, Georgia, and North Carolina. PRISM data for the locations of each weather station were extracted and used to estimate ET_O. For the radiation-based equations, the Donatelli–Bellocchi methodology was used to estimate daily solar radiation. For daily estimative, the Valiantzas method presented the best results, showing higher correlation and lower error. After calibrating the models for the region, all methods except Hargreaves had similar results. The results of this study indicate that the Valiantzas is the best method to estimate ET_O for the Southeast USA using PRISM data, since this method does not require calibration for the region and has shown improved performance in similar studies for Florida and other humid regions.