Location: Soil and Water Management ResearchTitle: Assessing SWAT plant stress algorithms using full and deficit irrigation treatments Author
|Chen, Yong - Texas A&M University|
|Marek, Thomas - Texas Agrilife Research|
|Xue, Qingwu - Texas Agrilife Research|
|Brauer, David - Dave|
|Srinivasan, Raghavan - Texas A&M University|
Submitted to: Agronomy Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/10/2018
Publication Date: 12/14/2018
Citation: Chen, Y., Marek, G.W., Marek, T.H., Xue, Q., Brauer, D.K., Srinivasan, R. 2018. Assessing SWAT plant stress algorithms using full and deficit irrigation treatments. Agronomy Journal. https://doi.org/10.2134/agronj2018.09.0556.
DOI: https://doi.org/10.2134/agronj2018.09.0556 Interpretive Summary: Decades of pumping with minimal recharge has resulted in declining levels of the Ogallala Aquifer in the Central High Plains. Irrigation strategies for extending limited water resources are commonly evaluated using models such as the Soil and Water Assessment Tool (SWAT), a widely used hydrologic model. However, concerns about the ability of the plant stress algorithms in SWAT to simulate plant response to limited irrigation treatments have tempered results. Scientists from ARS and Texas A&M AgriLife compared simulated plant growth and yield data with measured values from full and limited irrigation studies in Texas and Colorado. Results showed that the plant stress algorithms were unable to represent plant growth response to limited irrigation treatments, prompting the need for revision of the plant stress algorithms in SWAT.
Technical Abstract: Intensively irrigated agriculture in the Central High Plains of the United States has led to the depletion of groundwater levels of the underlying Ogallala Aquifer. Decreased well capacities along with concerns over groundwater conservation have increased interest in simulating crop responses to deficit irrigation strategies to evaluate sustainable irrigation management for profitable crop production. However, the ability of widely used simulation models to accurately represent crop responses to deficit irrigation is not thoroughly evaluated. Therefore, the primary objective of this research was to evaluate the efficacy of the plant stress algorithms of Soil and Water Assessment Tool (SWAT) to simulate leaf area index (LAI), biomass, and yield responses of corn (Zea Mays L.) to limited irrigation treatments (approximately 50% evapotranspiration requirement) when using crop parameters calibrated from full irrigation treatments for two study sites in the Central High Plains. Results showed that simulated corn LAI, biomass, and yield under full irrigation scenarios matched measured data reasonably well at both study sites. However, clear reductions in model performance for corn LAI simulations were found under the limited irrigation scenarios for both sites (Nash-Sutcliffe efficiency; NSE values less than 0.31 and percent bias; PBIAS greater than 22%). In addition, considerable overestimation of corn yield occurred in the limited irrigation scenarios for both sites (PBIAS greater than 30%). However, the aboveground biomass simulations were comparable to measured values (NSE greater than 0.86 and PBIAS within plus or minus15%) for the limited irrigation scenarios. The unsatisfactory results from simulations of both corn LAI and yield under the limited irrigation scenarios suggested potential deficiencies in the plant stress algorithms in SWAT. Two apparent limitations of the plant stress algorithms were 1) the equation for computing actual plant growth using a singular stress factor, determined by the maximum value of the four plant stress factors of water, temperature, nitrogen, and phosphorus and 2) the computed actual plant growth only adjusting potential daily accumulations of LAI and biomass rather than modifying the shape of the LAI-based crop growth function by adjusting related parameters. Further investigation into the source code of the plant stress algorithms in SWAT is needed for development and testing of alternative algorithms.