Top Model for calculating groundwater Recharge (TMR) based on vegetation LAI, soil properties and groundwater depth coupled with 3-D dynamic groundwater modeling using PCRaster-Modflow

Extensive afforestation since the mid-19th century has contributed to the desiccation of former wetland and mire ecosystems in Europe. Restoration of wet conditions is expected from transformation of evergreen coniferous forest characterized by high interception and transpiration to deciduous forest and low vegetation. The quantification of the spatial effects of such forest transformation on groundwater levels is difficult because evapotranspiration is usually calculated from atmospheric parameters only, while transpiration by plants and trees is partly determined by available soil moisture and groundwater. The aim was to develop a transient 3-D model that can calculate the effects of land use changes on groundwater levels and fluxes in time and space. For this purpose, a transient 3-D groundwater model (Modflow) per time step was linked to a 1-D top Model for Recharge (TMR). Recharge was calculated here from precipitation and reference evaporation, interception and evaporation related to vegetation (season LAI), transpiration depending on available soil moisture in the root zone and groundwater level per time step. Negative recharges were calculated at water levels just below or above ground level partly due to water losses by overland flow. The TMR model has been validated with time series (> 30 years) of groundwater level observations at various locations. A Modflow model of 5 model layers, cell size 10 x 10 m, time step 1 day, during 10 years has been built of a 55 km² large pilot area in Niedersachsen (DE) using Python and the PCRaster-Modflow (https://pcraster.geo.uu.nl/) platform. The LAI is classified from available forest stand data and land use maps. Soil parameters are based on the soil map of Germany (1:50k). The developed TMR-MF model has been validated for the period 2007-2016 by comparing calculated groundwater levels with measured levels at 86 locations. Mean deviation was 0.038 m. (Stdev.: 0.309 m., R2=0.9929). This model was used to quantitatively spatially analyse the effectiveness of forest conversion scenarios on eco-hydrological restoration of a dried-up stream valley lowland bog.

Acknowledgments: This project was created in collaboration with Niedersächsische Landesforsten (NLF) Fachbereich Entwicklung & Innovation and the Abteilung Wasserbewirtschaftung und Wasserrechte Oldenburgisch-Ostfriesischer Wasserverband (OOWV). Meteorological data were obtained from the Deutscher Wetterdienst.