Date of this Version
Transactions of the ASABE, Vol. 54(3): 915-930
Estimation of actual evapotranspiration (ET), especially its partitioning into plant transpiration (T) and soil evaporation (E), in agricultural fields is important for effective soil water management and conservation and for understanding the interactions between ET, T, and E with the management practices. Direct field measurements of ET, T, and E rates are difficult and costly; hence, mathematical models are used for estimating them. The objective of this study was to evaluate the practical applicability of the Shuttleworth‐Wallace (S‐W) model to estimate and partition ET in a subsurface drip‐irrigated soybean (Glycine max L. Merr.) field with partial residue cover. While its performance has been studied for various surfaces, the performance evaluation of the S‐W model for such surface has not been carried out. An integrated approach of calculating bulk stomatal resistance (rsc) as a function of soil water content (Θi) was incorporated into the model to allow simulation of T over a range of Θi, and a residue decomposition function was introduced to account for surface residue decay over time to more accurately account for the actual residue cover in field conditions. The model performance was evaluated for different plant growth stages during the 2007 and 2008 growing seasons at the University of Nebraska‐Lincoln,
South Central Agricultural Laboratory near Clay Center, Nebraska. The sum of estimated T and E was compared to the Bowen Ratio Energy Balance System (BREBS)‐measured actual ET on a daily time‐step. The model was able to capture the trends and magnitudes of measured ET, but its performance differed for various plant physiological growth stages. The root mean square difference (RMSD) values between the model‐estimated and measured ET values for the growing season (day after emergence until physiological maturity) were 1.26 and 1.03 mm d‐1 for 2007 and 2008, respectively. Best performance was observed during the mid‐season during full canopy cover with a two‐year average r2 of 0.87, average RMSD of 0.94 mm d‐1, and average mean biased error (MBE) of 0.30 mm d‐1. Estimates for both initial and late season growth stages where E was dominant had the least agreement with BREBS measurements. The proportion of T and E in the estimated ET varied with growth stage. The S‐W‐estimated seasonal total ET and BREBS measurements were equal in 2007 (S‐W model ET = 496 mm and BREBS ET = 498 mm), and in 2008 the model underestimated by only 8.2% (S‐W model ET = 452 mm and BREBS ET_= 489 mm). While, in general, the model was successful in tracking the trends and magnitude of the BREBS‐measured ET, further re‐parameterization of the T module of the model can improve its accuracy to estimate ET, especially T, during the initial and late season (before full canopy cover and after physiological maturity) for a subsurface drip‐irrigated soybean canopy. Other enhancements needed in the model for improved estimation of the E component include accurate determination of soil surface resistance coefficients and accounting for direct evaporation of intercepted rainfall on the canopy.