Modeling arid/semi-arid irrigated agricultural watersheds with SWAT: Applications, challenges, and solution strategies

Authors: SAMIMI, MARYAM - Oklahoma State University; MIRCHI, ALI - Oklahoma State University; Moriasi, Daniel; AHN, SORA - Texas A&M Agrilife; ALIAN, SARA - Oklahoma State University; TAGHVAEIAN, SALEH - Oklahoma State University; SHENG, ZHUPING - Texas A&M Agrilife

Submitted to: Journal of Hydrology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/10/2020
Publication Date: 9/10/2020
Citation: Samimi, M., Mirchi, A., Moriasi, D.N., Ahn, S., Alian, S., Taghvaeian, S., Sheng, Z. 2020. Modeling arid/semi-arid irrigated agricultural watersheds with SWAT: Applications, challenges, and solution strategies. Journal of Hydrology. 590:125418. https://doi.org/10.1016/j.jhydrol.2020.125418.
DOI: https://doi.org/10.1016/j.jhydrol.2020.125418

Interpretive Summary: The Soil and Water Assessment Tool (SWAT) is a watershed-scale hydrologic and water quality model that has been widely used to simulate the impacts of land management, climate and land use change on soil and water resources and to determine adaptive water resources management. A realistic representation of regional water budget components such as evapotranspiration and agricultural water management practices that affect them is essential to obtaining realistic SWAT model outputs in arid/semi-arid irrigated agricultural areas. In this study, a review of SWAT applications in arid/semi-arid areas was carried out to determine model hydrologic and water quality components studied, modeling challenges and possible solutions, and calibration strategies in order to provide recommendations for modeling in dryland agricultural areas. A majority of these studies focused on water quantity/availability concerns. The main modeling challenges we determined are lack of observational data, poor data quality, concerns about the accuracy of predicted model outputs, and limitations in the equations used to represent various processes in the model. To address these challenges, SWAT model users combined data from different sources with those in existing SWAT databases, used available tools to estimate missing data and evaluate model performance, and, when possible, improved equations of specific processes based on new advancements in science. An example SWAT application in the Elephant Butte Irrigation District near the New Mexico-Texas border, United States, showed that calibrating the model for streamflow, evapotranspiration, and groundwater recharge (multi-component) instead of only streamflow, better captured the regional water budget analysis in arid and semi-arid irrigated agricultural lands. Model users are encouraged to explicitly report the limitations, discuss the regional relevance of model parameter values, and calibrate the model based on multi-components to ensure reliable outcomes used by irrigation water resources managers.

Technical Abstract: Watershed models, including the widely applied Soil and Water Assessment Tool (SWAT), inform adaptive water resources management by facilitating quantitative analysis of different components of the water budget within a watershed. A realistic representation of regional hydrologic fluxes and agricultural water management practices that affect them is essential for meaningful SWAT applications to arid/semi-arid irrigated agricultural watersheds. We provide a state-of-the-practice review of SWAT applications in these watersheds, modeling challenges, and calibration strategies. The main modeling challenges are lack of observational data, poor data quality, concerns about simulation accuracy, and technical limitations of the model despite numerous advancements in the last two decades. To deal with these challenges, review results showed that modelers (i) combined data from different sources with those in existing SWAT databases, (ii) used supplemental tools to estimate missing data and evaluate model performance, (iii) sacrificed simulation accuracy when the major aim of the study was not undermined, and (iv) developed modular codes, tools, and algorithms to expand the model’s capabilities and improve process representations. We applied SWAT to an agricultural watershed in the desert southwest, U.S., to show the precautions necessary for credible simulations of streamflow, evapotranspiration, and groundwater recharge, three key water budget components governing regional water availability in irrigated drylands. Our application illustrates potential advantages of multi-component calibration, as opposed to the common practice of streamflow-based calibration, for SWAT-based regional water budget analysis in arid and semi-arid irrigated agricultural lands. Modelers are encouraged to explicitly report the limitations and regional relevance of model parameter values. Future advancements, such as conjunctive use of surface water and groundwater and dynamic annual land use, would significantly enhance the flexibility and performance of the model in terms of regional water budget analysis in the context of dryland agricultural areas.