Sensitivity of the SWAT model to the soil and land use data parametrisation: a case study in the Thyle catchment, Belgium

Romanowicz, Agnieszka [UCL] Vanclooster, Marnik [UCL] Rounsevell, Mark [UCL] La Junesse, I. [UCL]

The sensitivity of the distributed hydrological SWAT model to the pre-processing of soil and land use data was tested for modelling rainfall-runoff processes in the Thyle catchment in Belgium. To analyse this sensitivity, 32 different soil and land use parameterisation scheme were generated and evaluated. The soil input data sources were a generalised soil association map at a scale of 1:500,000, a detailed soil map at a scale of 1:25,000 and the soil profile analytical database AARDEWERK. These soil data were combined with a detailed and a generalised land use map. The results suggest that the SWAT model is extremely sensitive to the quality of the soil and land use data and the adopted pre-processing procedures of the geographically distributed data. The resolution and fragmentation of the original map objects are significantly affected by the internal aggregation procedures of the SWAT model. The catchment size threshold value (CSTV) is thereby a key parameter controlling the internal aggregation procedure in the model. It is shown that a parabolic function characterises the relationship between the CSTV and the hydrological modelling performance of the uncalibrated model, suggesting that optimal uncalibrated modelling results are not obtained when the CSTV is minimised. The hydrological response of the SWAT model to the calculated soil properties is significant. Therefore preference should be given to the calculation of the derived hydrologic soil properties prior to averaging of the profile data. Finally some general guidelines are suggested for parameterising soil and land use in the SWAT model application. (c) 2005 Elsevier B.V. All rights reserved.