Modelling the spatial and temporal variations of the water balance for the Weser catchment 1965–1994

Abstract

This study describes the application of the physically based SVAT model PROMET (PRocess Oriented Model for EvapoTranspiration) in the mesoscale catchment of the Weser (Northern Germany, approx. 37,500 km 2) utilizing a 30 years time series of meteorological input data. To enable a representative analysis of the spatial and temporal variations of the water balance components, the modelization is performed continuously without re-initialization of the state variables or specific calibration. Therefore, PROMET is expanded with the one-layer snow model ESCIMO (Energy balance Snow Cover Integrated MOdel) to provide an integrated model structure for continuous simulations of the water cycle. All necessary input data fields are integrated in a four-dimensional GIS data structure with a raster grid spacing of 1 km: a DEM, soil texture information derived from digitized maps, landuse distribution computed by unmixing a time series of NOAA/AVHRR satellite images and meteorological input data fields which are spatially and temporally interpolated using data provided by the standard measurement network of the German Weather Service (DWD). Spatially non-distributed physical soil and plant parameters are either derived from measurements or taken from literature. The study presents the structure of ESCIMO and its validation at the point and the catchment scale. Then, the modelled mean annual evapotranspiration, aET, as obtained by application of the linked models PROMET/ESCIMO is compared with the corresponding term ET calculated by inserting the measured precipitation and gauged runoff into the water balance equation. It is started from the assumption that for the 30 years period, the overall underground storage change Δ S is negligible. The mean annual deviation aET–ET over the 30 years period is 10.9 mm, indicating that the results represent a valid long-term description of the water balance. The patterns of the simulated water balance components are discussed with respect to the physical–geographical properties of the environment and their interaction with the predominant meteorological conditions. The study then addresses smaller spatial and temporal scales: for the subcatchment scale, the variation of aET–ET and thus the model error is considerably larger; for shorter periods, aET–ET is used for the estimation of Δ S. The simulation accuracy as depending on the length of the aggregation period and the mean monthly evolution of the storage are discussed.