SWAT Literature Database for Peer-Reviewed Journal Articles

Title:Analysis of the spatial variation in the parameters of the SWAT model with application in Flanders, Northern Belgium 
Authors:Heuvelmans, G., B. Muys and J. Feyen 
Journal:Hydrology and Earth System Sciences 
Volume (Issue):8(5) 
Article ID: 
URL (non-DOI journals): 
Broad Application Category:hydrologic only 
Primary Application Category:regionalization of input parameters 
Secondary Application Category:calibration, sensitivity, and/or uncertainty analysis 
Watershed Description:25 watersheds ranging in size from 2 to 210 km^2, located in the Schelde River in northern Belgium. 
Calibration Summary: 
Validation Summary: 
General Comments:The goal of the study was to determine zones in which key parameters can be transferred, with minor loss of accuracy. Calibration was initially performed for each watershed, which were located in one of 8 zones (8 larger subwatersheds in the Scheidt River Basin). For each of the 25 subwatersheds, SWAT was run with local optimal parameters and average values to reflect whole Flanders region (containing the Scheidt River Basin) parameters based on zonal characteristics. The use of region-wide (Flanders) average parameters resulted in much worse results. Better results were obtained with clusters of parameters than using a single parameter approach. 
Abstract:Operational applications of a hydrological model often require the prediction of stream flow in (future) time periods without stream flow observations or in ungauged catchments. Data for a case-specific optimisation of model parameters are not available for such applications, so parameters have to be derived from other catchments or time periods. It has been demonstrated that for applications of the SWAT in Northern Belgium, temporal transfers of the parameters have less influence than spatial transfers on the performance of the model. This study examines the spatial variation in parameter optima in more detail. The aim was to delineate zones wherein model parameters can be transferred without a significant loss of model performance. SWAT was calibrated for 25 catchments that are part of eight larger sub-basins of the Scheldt river basin. Two approaches are discussed for grouping these units in zones with a uniform set of parameters: a single parameter approach considering each parameter separately and a parameter set approach evaluating the parameterisation as a whole. For every catchment, the SWAT model was run with the local parameter optima, with the average parameter values for the entire study region (Flanders), with the zones delineated with the single parameter approach and with the zones obtained by the parameter set approach. Comparison of the model performances of these four parameterisation strategies indicates that both the single parameter and the parameter set zones lead to stream flow predictions that are more accurate than if the entire study region were treated as one single zone. On the other hand, the use of zonal average parameter values results in a considerably worse model fit compared to local parameter optima. Clustering of parameter sets gives a more accurate result than the single parameter approach and is, therefore, the preferred technique for use in the parameterisation of ungauged sub-catchments as part of the simulation of a large river basin. 
Keywords:hydrological model, regionalisation, parameterisation, spatial variability