An Experimental Investigation of Evapotranspiration Estimation Methods

Abstract

Using meteorological data collected during the 1978, 1979, and 1980 growing seasons at St. Paul, Minnesota, estimates of daily potential evapotranspiration were made using thirteen different calculation methods. These estimates were then compared to measurements of actual evapotranspiration from a cropped surface obtained from the University of Minnesota's weighing lysimeter. The crop was soybeans in 1978 and 1979, and alfalfa in 1980. Simpie linear regression techniques were used to compare the estimated and measured values of evapotranspiration. Scatter plots showing the relationships between the predicted and observed values and histograms of the differences between the two are presented. Summary statistics for each regression and set of difference values are reported. It was found that pan evaporation was most successful at predicting daily evapotranspiration. The methods which combine energy balance and aerodynamic functions were the next best, followed by the methods using radiation and temperature as inputs. The methods based on temperature alone performed most poorly.