Effect of Temporal Scales on Estimating Evapotranspiration with Apex.

The Agricultural Policy/Environmental eXtender (APEX) model has been widely applied to determine the impacts of climate and land use changes, and management of water resources. APEX is the scientific basis for the Nutrient Tracking Tool (NTT), which is used by USDA to estimate nutrient losses to the environment associated with alternative management practices. Evapotranspiration (ET), is a major component of hydrologic budgets and requires accurate estimation. Any errors in the estimation of ET is propagated to water quality predictions. Therefore, the objectives of this study were to: 1) calibrate and validate the APEX model for estimating monthly and annual ET in dryland cropping systems in the semi-arid Texas High Plains; and 2) determine the impact of temporal scales on model performance. Lysimetric data (2001–2010) measured for cotton, sorghum, and soybean were used for model calibration (2001–2005) and validation (2006–2010). The model was built using the NTT interface, and Hargreaves ET equation was used. A sensitivity analysis was performed using APEXSENSUN tool. Study results indicated that the sensitive parameters included soil evaporation plant cover, Hargreaves equation parameter, and Hargreaves PET exponent. The model performed best at the monthly time step. The model simulated monthly ET very well with Nash Sutcliffe efficiency (NSE) values of 0.82 and 0.81 and percent bias (PBIAS) values of ±0.6% and ±9% during the calibration and validation periods, respectively. The model simulated annual ET very well during the calibration period with NSE and PBIAS values of 0.86 and ±0.6%, respectively. However, during the validation period the model did not capture the inter-annual variability within an acceptable range (NSE=-0.10). This implies that the calibrated annual parameter values are not robust enough to simulate temporal variability of ET. Therefore, we recommend parameterizing and simulating ET on a monthly temporal scale in cases with limited annual data.