Simulating internal watershed processes using multiple SWAT models

Authors: APOSTEL, ANNA - The Ohio State University; KALCIC, MARGARET - The Ohio State University; DAGNEW, AWOKE - University Of Michigan; EVENSON, GREY - The Ohio State University; KAST, JEFFREY - The Ohio State University; King, Kevin; MARTIN, JAY - The Ohio State University; LOGSDON MUENICH, R - Arizona State University; SCAVIA, DONALD - University Of Michigan

Submitted to: Science of the Total Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/29/2020
Publication Date: 12/3/2020
Citation: Apostel, A., Kalcic, M., Dagnew, A., Evenson, G., Kast, J., King, K.W., Martin, J., Logsdon Muenich, R., Scavia, D. 2020. Simulating internal watershed processes using multiple SWAT models. Science of the Total Environment. 759. Article 143920. https://doi.org/10.1016/j.scitotenv.2020.143920.
DOI: https://doi.org/10.1016/j.scitotenv.2020.143920

Interpretive Summary: Hydrology and water quality models are generally validated and used to guide watershed scale agricultural policy and programs. However, the models may or may not capture field scale dynamics where practices and programs are implemented. This disconnect can lead to practice promotion and adoption that may not be effective. Thus understanding the effects of spatial resolution on representing field scale dynamics is important. Assessment of three models varying in upland spatial resolution indicated that regardless of spatial resolution, surface runoff and tile drainage discharge were over-predicted. In contrast, phosphorus was under-predicted in tile drainage while nitrogen was under-predicted in surface runoff. These findings highlight the need for caution when trying to extend watershed scale model results to field scale management recommendations and should be useful for agricultural educators and policy makers that depend solely on watershed scale model outputs to inform conservation practice promotion.

Technical Abstract: The need for effective water quality models to help guide management and policy, and extend monitoring information, is at the forefront of recent discussions related to watershed management. These models are often calibrated and validated at the basin outlet which ensures that models are capable of evaluating basin scale hydrology and water quality. However, there is a need to understand where these models succeed or fail with respect to internal process representation, as these watershed-scale models are used to inform management practices and mitigation strategies which are incorporated at upland points. We evaluated an ensemble of models—each calibrated to in-stream observations at the basin outlet—against discharge and nutrient observations at the farm field scale to determine the extent to which these models capture field scale dynamics. While all models performed well at the outlet, upstream performance varied. Models tended to over predict discharge through surface runoff and subsurface drainage, while underpredicting tile components of phosphorus and surface components of nitrogen. Our study suggests that while models may be applied to predict basin scale management, care should be taken in applying the models to evaluate field scale management and processes in the absence of data incorporation at that scale, even with the use of multiple models.