Improving Parameterization of a Regional Irrigation Scheduling Program for Evapotranspiration Estimation with Eddy Covariance Measurements

Abstract

Actual evapotranspiration (ETa) is an essential variable in linking energy cycles, carbon, and water, yet challenging to measure. Input uncertainty and deficiencies in the key elements of hydrologic models are fundamental challenges for optimizing model performance. Furthermore, the performance of land surface model-based ETa, reanalysis, and remote sensing products varies with spatiotemporal scales. Here, we evaluate sources of bias in the regional Wisconsin Irrigation and Scheduling Program (WISP) model and develop a correction using eddy covariance (EC) observations. ETa, observations were made for five years (2018–2022) using EC systems in agricultural fields in Wisconsin. WISP model ETa bias was linked to underestimation of net long wave radiation (LWnet) that was traced to incorrect specification of effective clear sky atmospheric emissivity ( ). Applying a correction to the  led to reduced WISP model percent bias (pbias) and error for both LWnet and ETa. The calibrated model more accurately represented observed ETa. The results indicate that explicit treatment of the LWnet balance decreases the uncertainty of model parameters and improves the WISP model performance at independent sites. Applying this improved model parameterization reduced the bias of LWnet radiation from 62.9% to -6.2%, which improved the Nash-Sutcliffe Efficiency (NSE) from -0.08 to 0.52 for ETa on training sites. Additionally, overall pbias was significantly reduced (p=0.035) for validation sites after WISP correction. Hence, WISP performance improved for different crop types when optimal regional parameters were used, confirming the physical parameters' reliability. Our results highlight that model development should focus on energy balance parameterizations to improve ET simulation to improve the accuracy of hydrologic and climatic simulations for understanding critical processes underlying hydrologic and climatic variability and change over land.

Keywords: Actual evapotranspiration, Eddy covariance, Longwave radiation, emissivity, percentage bias