Freiburger Schriften zur Hydrologie

[ << Band/volume 16 ] [ Freiburger Schriften zur Hydrologie ] [ >> Band/volume 18 ]

Band/volume 17: ARMBRUSTER V. (2002):

Grundwasserneubildung in Baden-Württemberg

The objective ofthis study was the development of a method to determine the lang term mean groundwater recharge from precipitation für large scale areas. This method was supposed to account für the influence of climate, land use, soil and groundwater level on the goveming processes of evapotranspiration and runoff generation. Olle focus of the study was the determination of groundwater recharge in hard rock regions. This implied to improve the description of fast lateral runoff components in the macro scale. The method was applied in the federal state of Baden-Württemberg für the time period 1961-1990. Groundwater recharge GWR is determined according to following equation:

GWR = (P-ET)' (Dbas/Dtot)

In a first step the sink term Evapotranspiration ET is subtracted from Precipitation P. In a second step fast lateral runoff components are considered by multiplying the resulting total discharge with the ratio of baseflow divided by total discharge Dbas/Dtot.
The detailed evapotranspiration model TRAIN-GWN (TRAnspirationINterzeption-GrundWasserNeubildung) was newly developed in this study. Its application provides precipitation and evapotranspiration für the whole study area. The model is based on the evapotranspiration model TRAIN (MENZEL 1997a, 1999). TRAIN-GWN consists of conceptual and physically based modules. It works on a daily time step on a 500 m x 500 m grid and simulates evapotranspiration and percolation out of the rooted soil zone. The snow module is based on the degree dar method. The soil module according to the HBV -model and capillary rise according to the 'Bodenkundliche Kartieranleitung' (AG BODEN 1994) are newly integrated. Another new module is the interpolation of meteorological input data, based on a combination of Inverse Distance Weight and a regression on altitude. Radiation calculation, interception according to the theory of latent heat transfer over droplets, and evapotranspiration according to the Penman-Monteith-relation are strongly physically based components. Extended sensitivity analysis were conducted. The lang term mean groundwater recharge is sensitive to different land use parameters, especially the leaf area index and to the two soi! parameters. Among the meteorological input data, precipitation hag a large impact on recharge.
For the land use agriculture and grass the model was validated on measured evapotranspiration and percolation of Olle weighable and twenty non-weighable lysimeters. A regionalization function für the empirical soil parameter BETA was derived and groundwater recharge was simulated on a weekly and monthly basis. The results confmn the concept of the soi! module, accounting für percolation before maximum soil water content is reached. The mean ready groundwater recharge could be determined without systematic deviation, showing a mean deviation of 8%. In addition, the inter-annual variation could be reproduced weIl. A comparison of the evapotranspiration with results of the model VEKOS (KLÄMT 1988) showed good agreement of ready totals, while considerable deviations occurred in winter times. This might be due to the fact, thai TRAIN-GWN does not simulate soil evaporation separately.
In a next step TRAIN-GWN was validated on measured discharge of tell mainly forested, mountainous catchments. The systematic overestimation of the modeled lang term discharge amounts to 8%. Furthermore, catchment evapotranspiration and precipitation were compared to values from the Hydrological Atlas of Germany. Evapotranspiration is weIl comparable, with TRAIN-GWN showing only 5% higher values. Precipitation does not show systematic deviation, while single catchments show differences ofup to 7%. This demonstrates the difficulties of precipitation determination in mountains, which presents a main source of erraT für groundwater recharge simulations. Overall the results show, that precipitation is slightly overestimated.
The spatial distribution of the lang term mean evapotranspiration is influenced by climate, land use, soils and groundwater levels. The mean evapotranspiration für the whole study area amounts to 555 mm, the mean für agricultural areas and grasslands is 533 mm, while it is 635 mm für forests.
In areas with fast lateral runoff components this sink term is accounted für by the lang term mean ratio of baseflow divided by total discharge (Dbas/Dtot). Base flow was determined from time series of discharge, using a modified Wundt/Kille-method, the Demuth-method. This results in baseflow during low flow conditions, which is not the same as baseflow, defmed by source areas and flow paths. In the mountainous crystalline test catchment Brugga in the Black Forest, für example, the baseflow according to the Demuth-method originates from slow inter flow from the debris cover of slopes in addition to groundwater from hard rock aquifers. Since the used Demuth-method is based on discharge data, the data base is large enough to allow the regionalization of the ratio with a multiple linear regression. 105 test catchments were selected. Their determined Dbas/Dtot ranges from 16 to 80% and is scale independent. The regression model was developed based on a calibration data set of 70 catchments and their characteristics. The model consists of eleven catchment parameters, which are all plausible to describe runoff generation. They characterize soils, geology, hydrogeology and drainage density. Parameters describing precipitation and topography were statistically insignificant. Model performance is satisfying with a coefficient of determination of 0.76. Its validation on an independent data set of 35 catchments yield a coefficient of determination of 0.74, confirming the developed model.
In a next step Baden- Württemberg was delineated into areas without fast lateral runoff components, mainly weIl permeable alluvial aquifers, and into areas with lateral components. The latter Olles were further subdivided into 1669 sub catchments. On the basis of their characteristics Dbas/Dtot was regionalized. The spatial distribution of Dbas/Dtot plausibly reflects the physiographic differences of the regions.
Groundwater recharge results from the combination of the lang term mean total discharge with the regionalized Dbas/Dtot. Precipitation and Dbas/Dtot strongly influence the large scale variability of the groundwater recharge, while the small scale variability is mainly govemed by evapotranspiration and its influencing factors. The mean evapotranspiration in Baden-Württemberg amounts to 237 mm, 86% of the area shows values between 50 and 400 mm. Due to the high temporal resolution, TRAIN-GWN can simulate the inter-annual distribution of groundwater recharge, in addition to values of single years. As the latter vary strongly, the lang term mean groundwater recharge can only be considered as a first estimate für single years.
The combination of the detailed evapotranspiration model with the lang term mean Dbas/Dtot presents a new approach in the modeling of groundwater recharge from precipitation. It contains an improved method to ac count für fast ateral runoff components in the macro scale.. This method is based on several catchment characteristics, being plausible für runoff generation. Required input data sets are available für most large scale study areas. The approach is applicable in alluvial aquifers as weIl as in hard rock regions. With its application in BadenWürttemberg spatially detailed groundwater recharge, determined with a unique method, is available in the whole federal state für the fIrst time. The approach presents an adequate tool für different practical questions ofwater management. Long term mean groundwater recharge is the basis für the sustainable management of groundwater resources, while monthly time series are needed für transient groundwater models. Impacts of climate change on groundwater recharge and its inter annual variation can also be estimated.

PDF-Download (5,3 MB)