Experimental, verification of a mechanistic model to partition evapotranspiration into soil water and plant evaporation

A model that separates evapotranspiration (ET) into soil water evaporation (E) and plant transpiration (T), and calculates parameters associated with the water and energy balance of the soil and crop surfaces is required to u nderstand the fate of water in different cropping systems. ENWATBAL, a mech anistic ET model, is such a model. Our objective was to test the model by c omparing calculated values of ET, E, T, net irradiance (R-n), soil heat flu x (G), and;oil surface temperature (T-s) with measured ones. Three field ex periments were conducted throughout 1993-1995 in a structureless sandy soil either bare or planted with sorghum in Tottori, Japan. E was measured by l ysimeter using both micro- and an in situ large weighing lysimeter, ET was measured with the weighing lysimeter, and T was calculated by difference (T = ET-E). In these experiments, R-n, G, and T-s were also measured using st andard methods. Three statistical tests, i.e., bias, RMSE and linear regres sion analysis were used to evaluate model performance. In all three experim ents, measured and calculated values of E and T from both a bare soil and a cropped surface were in agreement with a bias close to 0.0 mm day(-1) and RMSE <1.0 mm day(-1). In these comparisons, slopes and intercepts were not significantly (P = 0.05) different from 1.0 and 0.0, respectively. Furtherm ore, experimental results indicated that ENWATBAL correctly calculated dail y ET with a bias = -0.60 mm day(-1), RMSE = 0.89 mm day(-1) and a slope and intercept not significantly different from 1.0 and zero. Measured and calc ulated values of R-n and G over a bare soil were also in agreement. However , there are differences between the midnight and diurnal values of measured and calculated T-s. Therefore, we concluded that the ENWATBAL model calcul ated ET, E, T, R-n, and G within an acceptable accuracy and range for the s oil and environmental conditions of our test. It is suggested that for our soil, model accuracy on the calculation of E and T-s could be improved by i ncluding the formation of a dry topsoil layer. (C) 1999 Elsevier Science B. V. All rights reserved.