Category: Planning & Management

394714 - Comprehensive Hydrologic Calibration of SWAT using Many-Objective Evolutionary Algorithm

Tuesday, Jun 5
8:30 AM – 5:30 PM

Comprehensive hydrologic calibration is of critical importance towards developing a robust model that is capable of simulating observed flow magnitudes, flow duration curve and water balance components with a reasonable accuracy. In this study, a hydrologic model was developed for Kankakee River watershed to assess water supply availability under changing climate and water use conditions. The watershed, which is located in Illinois and Indiana, is predominantly agricultural with a total drainage area of approximately 13,300 square kilometers. The model was developed using the Soil and Water Assessment Tool (SWAT) coupled with a many-objective evolutionary algorithm known as NSGA-III for comprehensive multi-variable, multi-site hydrologic calibration of the watershed model. To perform automatic calibration of daily streamflows, flow duration curves (FDCs) and water balance components at four gauging sites, a many-objective optimization problem was formulated. The FDC at each site was further divided into five segments with different probability exceedance as additional objectives. A nonlinear groundwater module was incorporated into the SWAT model to improve low flow simulations. Error index statistics of stream flow, FDCs, and water balance components were used as objective functions. For both calibration and validation periods at all sites, the minimum Nash-Sutcliffe Efficiency (NSE) value obtained for daily flow simulation was 0.6, with a maximum percent bias (PBIAS) of 8%. Ratio of root mean square error to standard deviation of observed data (RSR) for all FDC segments ranged from 0.1 to 0.6 for all sites. The average annual surface runoff, groundwater and evapotranspiration were simulated with a maximum PBIAS of 15%. Results indicated the model’s good performance in simulating daily flows, flow durations and water balance components at all sites. The many-objective optimization algorithm was effective in identifying the optimal parameter sets for all hydrologic variables and sites, isolating unique watershed characteristics for further fine-tuning of model parameters.

Co-Authors: Sangeetha Chandrasekaran, Champaign, IL – Illinois State Water Survey, Prairie Research Institute, UIUC

Elias Getahun

Research Hydrologist
Illinois State Water Survey at Prairie Research Institute, University of Illinois Urbana-Champaign
Champaign, Illinois

Dr. Elias Getahun is a research hydrologist with Illinois State Water Survey at Prairie Research Institute, University of Illinois Urbana-Champaign. He received his PhD in civil engineering from Southern Illinois University. Dr. Getahun (Bekele) has more than 10 years of experience in water resources systems analysis, watershed modeling and decision support systems for non-point source pollution control. He has published his research work in peer-reviewed journals, technical reports and conference proceedings.