Keywords
hydrological modelling, downscaling, land use change, uncertainty analysis
Start Date
1-7-2006 12:00 AM
Abstract
Semi-distributed hydrological models generally have the advantages of short calculation times, comparative low calibration needs and high model efficiency, but lack the ability to consider localization effects of land use change. A regionalisation of these models allows a sensitivity analysis of the localization effects. HBV-D, a conceptual hydrological model is used in this study. The regionalization for the German watershed Parthe (˜317 km2) is coded in the framework of SME (spatial modeling environment) which allows a fast grid based regionalization of the model. Additional complexity at the finer scale is handled by downscaling of calibration parameters fromthe semi-distributedmodel by using auxiliary information (soil, relief). This allows a better representation of the heterogeneity in the watersheds without the need of grappling with hundreds of calibration parameters. A Monte-Carlo analysis is used to simulate the effects of the different spatial pattern of land use changes on discharge. This allows a better forecasting of land use change effects and can be used to generate uncertainty estimates for existing semi-distributed models. We focus here on the following major questions: 1. how can we downscale the calibration parameters from the semi-distributed model to the distributed model, 2. how do downscaling approaches differ, 3. how does land use composition and configuration influence discharge and 4. how do these results depend on catchment characteristics?
Localization effects of land use change on hydrological models
Semi-distributed hydrological models generally have the advantages of short calculation times, comparative low calibration needs and high model efficiency, but lack the ability to consider localization effects of land use change. A regionalisation of these models allows a sensitivity analysis of the localization effects. HBV-D, a conceptual hydrological model is used in this study. The regionalization for the German watershed Parthe (˜317 km2) is coded in the framework of SME (spatial modeling environment) which allows a fast grid based regionalization of the model. Additional complexity at the finer scale is handled by downscaling of calibration parameters fromthe semi-distributedmodel by using auxiliary information (soil, relief). This allows a better representation of the heterogeneity in the watersheds without the need of grappling with hundreds of calibration parameters. A Monte-Carlo analysis is used to simulate the effects of the different spatial pattern of land use changes on discharge. This allows a better forecasting of land use change effects and can be used to generate uncertainty estimates for existing semi-distributed models. We focus here on the following major questions: 1. how can we downscale the calibration parameters from the semi-distributed model to the distributed model, 2. how do downscaling approaches differ, 3. how does land use composition and configuration influence discharge and 4. how do these results depend on catchment characteristics?
