|Larose, M - PURDUE UNIVERSITY|
Submitted to: International Congress on Modeling and Simulation Proceedings
Publication Type: Proceedings
Publication Acceptance Date: August 1, 2007
Publication Date: December 10, 2007
Citation: Heathman, G.C., Larose, M. 2007. Influence of Scale on SWAT Model Calibration for Streamflow [abstract]. In Oxley, L. and Kulasiri, D. (eds) MODSIM 2007 International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, December 10-13, 2007. p. 2747-2753. Interpretive Summary: There are several issues to consider in the application of watershed scale hydrologic modelling, one of which is the influence of scale on model calibration parameters. This is especially true when using the model as an environmental assessment tool or as a decision-support system for soil and water resource management. The objective of this study was to determine to what extent the influence of scale effects streamflow estimates in the St. Joseph River Watershed, a large scale agricultural watershed. Based on our results, the SWAT model shows greater sensitivity between daily and monthly streamflow estimates at the smaller watershed scale and more uncertainty associated with estimates at the larger scale. However, the actual scale at which the model was calibrated did not have a substantial impact on model results based on comparisons with observed streamflow at gauging locations representing each scale.
Technical Abstract: The Soil Water Assessment Tool (SWAT) was implemented in the 281,000 ha St. Joseph River Watershed (SJRW) to investigate the influence of multiple scales on stream flow model calibration parameters. The relationship between model calibration parameters and associated hydrological response units (HRU) between different scales is not well understood. In this investigation, two scales were used within the SJRW where such factors as land use, soil type, topography and management practices are considered similar, thus, conforming to the concept of downscaling rather than regionalization. The model was calibrated for stream flow in the SJRW. Critical parameters optimized for calibration were: 1) CN2, curve number, 2) ESCO, soil evaporation compensation factor, and 3) SOL_AWC, available water holding capacity. Using optimized parameters at the SJRW scale, stream flow estimates were evaluated in the 70,820 ha Cedar Creek Watershed (CCW), the largest subbasin and tributary in the SJRW. These same parameters were then optimized at the CCW scale and stream flow estimates for the CCW and the SJRW were evaluated. Modeled and measured stream flow data were statistically analysed at both scales based on their respective calibrations. Based on our results, the SWAT model shows greater sensitivity between daily and monthly streamflow estimates at the smaller watershed scale and more uncertainty associated with estimates at the larger scale.