A SWAT-based optimization tool for obtaining cost-effective strategies for agricultural conservation practice implementation at watershed scales

Authors: LIU, YAOZE - State University Of New York (SUNY); GUO, TIAN - Heidelberg University; WANG, RUOYU - University Of California, Davis; ENGEL, BERNARD - Purdue University; Flanagan, Dennis; LI, SIYU - State University Of New York (SUNY); PIJANOWSKI, BRYAN C - Purdue University; COLLINGSWORTH, PARIS - Purdue University; LEE, JOHN - Purdue University; WALLACE, CARLINGTON - Interstate Commission On The Potomac River Basin

Submitted to: Science of the Total Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/11/2019
Publication Date: 7/12/2019
Citation: Liu, Y., Guo, T., Wang, R., Engel, B.A., Flanagan, D.C., Li, S., Pijanowski, B., Collingsworth, P.D., Lee, J.G., Wallace, C.W. 2019. A SWAT-based optimization tool for obtaining cost-effective strategies for agricultural conservation practice implementation at watershed scales. Science of the Total Environment. 691:685-696. https://doi.org/10.1016/j.scitotenv.2019.07.175.
DOI: https://doi.org/10.1016/j.scitotenv.2019.07.175

Interpretive Summary: Water quality problems in the Great Lakes are increasingly being identified as caused by nonpoint source pollutants, particularly nitrogen (N) and phosphorus (P) originating on agricultural fields. Some of the worst problems with algal blooms have been seen in Lake Erie, in response to soluble and total phosphorus loadings from the Maumee River. Different ways to reduce these large-scale algal blooms involve various land management practices on farms that reduce the amount of P leaving fields. These practices include things such as reduced tillage or no-till, buffer strips at the edges of fields, and nutrient management, and are collectively known at BMPs (Best Management Practices). In this research a computer modeling tool was developed that determines the most cost-effective BMPs to implement within a watershed. The tool was applied to a watershed in northeastern Indiana (contained with the Maumee River Basin), and various tests were conducted with it. Simulations with different BMPs applied to the entire watershed were compared with simulations where BMPs were targeted to critical areas. We found that the optimized scenario could reduce spring (March-July) P losses by 40.0%, which is the target for P reductions for Maumee River water entering Lake Erie. The optimization tool can help stakeholders determine the optimal types, quantities, and spatial locations of BMPs that can maximize reductions in pollutant loadings with the lowest BMP costs. This research impacts farmers, landowners, conservation agency personnel, extension agents, and others involved in efforts to reduce water pollution of the Great Lakes and other water bodies.

Technical Abstract: To address the harmful algal blooms problem in Lake Erie, one solution is to determine the most cost-effective strategies for implementing agricultural best management practices (BMPs) in the Maumee River watershed. An optimization tool, which combines AMALGAM (optimization algorithm), SWAT (Soil and Water Assessment Tool), and MLSOPT framework (computational efficient method), was created to optimally apply agricultural BMPs at watershed scales. The optimization tool was demonstrated in the AXL watershed, an agricultural watershed in the Maumee River basin. The initial implementation of BMPs with low expenditures greatly reduced pollutant loadings; beyond certain levels of pollutant reductions, additional expenditures resulted in less significant reductions in pollutant loadings. Compared to optimization for the entire watershed, optimization in critical areas (25% of the watershed with the greatest pollutant loadings per area) can greatly reduce computational time and obtain similar optimization results for initial reductions in pollutant loadings, which were 10% for Dissolved Reactive Phosphorus (DRP) and 38% for Total Phosphorus (TP); however, for greater reductions in pollutant loadings, critical area optimization was less cost-effective. To reduce each type of pollutant load, optimization should be conducted individually with the objective function of reducing the corresponding pollutant. The optimized scenario, which can reduce spring (March-July) DRP and TP losses by 40.0% and 51.1%, respectively, attains the goal of simultaneously reducing spring DRP/TP losses by 40%. The optimization tool is capable of helping stakeholders obtain optimal types, quantities, and spatial locations of BMPs that can maximize reductions in pollutant loadings with the lowest BMP costs.