First results of using convection-permitting WRF simulations to drive the LARSIM hydrological model for the Rhine River catchment

Tuesday, 16 December 2014
Sebastian Knist1,2, Klaus Goergen1,2, Enno Nilson3, Peter Krahe3, Bastian Klein3 and Clemens Simmer1,4, (1)University of Bonn, Bonn, Germany, (2)SimLab TerrSys, Jülich Supercomputing Centre, Jülich Research Centre, Jülich, Germany, (3)Federal Institute of Hydrology, Koblenz, Germany, (4)Centre for High-Performance Scientific Computing in Terrestrial Systems, Geoverbund ABC/J, Jülich, Germany
Convection-permitting regional climate model (RCM) simulations more accurately resolve small-scale surface features such as land-use patterns or topography and are less dependent on convection parameterizations. They have been shown to provide added value e.g. in the reproduction of the intensities, the diurnal cycle and the spatial distribution of precipitation or dominant interaction processes like soil moisture-precipitation and temperature feedbacks, offering the potential for an improved reproduction of the hydrological cycle. The Rhine River catchment in Central Europe covers 185000 km² from the Alps in the South to the North Sea with many population centres of high economic importance along its major rivers.
In this study, first results are presented from 3 km convection-permitting Weather Research and Forecasting model (WRF) hindcast simulations for a central European model domain that are used as offline forcing data for the LARSIM (Large Area Runoff Simulation Model) hydrological model at 5 km resolution over the Rhine River catchment. The goal is to investigate how well hydrometeorological conditions during recent flood and drought events can be reproduced and how the higher resolution atmospheric forcing data affect the performance of the hydrological model.
The WRF model is run in a one-way double-nesting setup, where the 3 km model domain is inscribed into the official pan-European Coordinated Regional Downscaling Experiment (CORDEX) EUR-11 model grid (about 12 km). It is driven by ERA-Interim reanalysis as part of validation runs of a bigger regional climate change experiment. The simulated time span from 2009 to 2013 covers distinct hydrometeorological and hydrological conditions in parts of the Rhine catchment (flood: 2013; hydrological droughts: 2011). The RCM data are compared to high-resolution gridded observations. LARSIM is run using 12 km and 3 km RCM data as input and its results are validated with observed discharge data at gauging stations. In addition, performance metrics are derived to analyse whether an added value of the higher resolution RCM forcing as opposed to coarser standard resolution products can be seen in the simulated hydrometeorological and hydrological data during contrasting events (low- vs high-flow).