A crop coefficient-based water use model with non-uniform root distribution

Authors: Schwartz, Robert; DOMINGUEZ, ALFONSO - University Of Castilla-La Mancha(UCLM); PARDO, JOSE - University Of Castilla-La Mancha(UCLM); Colaizzi, Paul; Baumhardt, Roland - Louis; BELL, JOURDAN - Texas A&M Agrilife

Submitted to: Agricultural Water Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/1/2019
Publication Date: 11/13/2019
Publication URL: https://handle.nal.usda.gov/10113/6755814
Citation: Schwartz, R.C., Dominguez, A., Pardo, J.J., Colaizzi, P.D., Baumhardt, R.L., Bell, J.M. 2019. A crop coefficient-based water use model with non-uniform root distribution. Agricultural Water Management. https://doi.org/10.1016/j.agwat.2019.105892.
DOI: https://doi.org/10.1016/j.agwat.2019.105892

Interpretive Summary: Yield and profitability of irrigated crop production depends on accurately evaluating crop water requirements. However, crop water use estimated using the standard model can be unreliable for deeply rooted crops because all water in the rooting zone is assumed to be available. In addition, the standard model is often inaccurate during drought when crops are severely water stressed. Therefore, scientists from USDA-ARS (Bushland, TX), University of Castilla La Mancha (Spain) and Texas A&M AgriLife Research expanded upon the standard method to estimate crop water use by modifying the stress response function and by limiting root water extraction deeper in the soil profile. The model was calibrated to predict corn water use and grain yield over a wide range in crop water deficits. Seasonal water use of dryland to fully irrigated corn was predicted with an average error of one inch. Using a nonuniform root distribution reduced the prediction error by 18% compared with the standard model. Grain yield was estimated with a relatively large average prediction error of 42 bushels per acre because of heat stress that caused yield declines and differing farming practices. Because of the limited input requirements and robustness over a wide range in crop water stress levels, the model would be suitable for evaluating deficit irrigation strategies.

Technical Abstract: Uncertainties in the estimation of evapotranspiration (ET) using the crop coefficient (Kc)-reference ET method arise for deeply rooted crops and severe water stress. We expanded upon the crop coefficient–based model by modifying plant available water via a nonuniform root distribution that limited deep water extraction using daily estimated soil profile water contents. The model was calibrated to predict maize (Zea mays L.) ET over a wide range in crop water deficits. In addition, maize grain yield was calibrated with model-predicted ET using a multiplicative water production function. The calibrated model with optimized crop and stress response coefficients predicted actual maize ET for a wide range in water deficits with a daily and growing season prediction root mean square error (RMSE) of 1.16 mm d-1 and 25.6 mm, respectively. A nonuniform root distribution functioned similarly to stress response coefficients reducing soil water extraction deeper in the profile with a resultant 18% reduction in the prediction RMSE compared with the optimized stress response conventionally used with the Kc approach. The largest uncertainties in predicted crop ET resulted from an underestimation of runoff and an overestimation of crop water use during stress-induced early senescence. Measured and predicted soil water contents averaged over the entire rooting depth agreed closely, however root water extraction was overestimated deeper in the profile. Calibration of the water production function using data exhibiting a wide range in measured grain yield resulted in a RMSE of 2.1 Mg ha-1. Including an additive high temperature stress response expression improved the calibration. Because of the limited input requirements and robustness over a wide range in crop water stress levels, the model would be suitable for evaluating deficit irrigation strategies.