ALBERT

Description: The land surface and the atmospheric boundary layer are closely intertwined with respect to the exchange of water, trace gases, and energy. Nonlinear feedback and scale-dependent mechanisms are obvious by observations and theories. Modeling instead is often narrowed to single compartments of the terrestrial system or bound to traditional viewpoints of definite scientific disciplines. Coupled terrestrial hydrometeorological modeling systems attempt to overcome these limitations to achieve a better integration of the processes relevant for regional climate studies and local-area weather prediction. This study examines the ability of the hydrologically enhanced version of the Weather Research and Forecasting model (WRF-Hydro) to reproduce the regional water cycle by means of a two-way coupled approach and assesses the impact of hydrological coupling with respect to a traditional regional atmospheric model setting. It includes the observation-based calibration of the hydrological model component (offline WRF-Hydro) and a comparison of the classic WRF and the fully coupled WRF-Hydro models both with identically calibrated parameter settings for the land surface model (Noah-Multiparametrization; Noah-MP). The simulations are evaluated based on extensive observations at the Terrestrial Environmental Observatories (TERENO) Pre-Alpine Observatory for the Ammer (600 km2) and Rott (55 km2) river catchments in southern Germany, covering a 5-month period (June–October 2016). The sensitivity of seven land surface parameters is tested using the Latin-Hypercube–One-factor-At-a-Time (LH-OAT) method, and six sensitive parameters are subsequently optimized for six different subcatchments, using the model-independent Parameter Estimation and Uncertainty Analysis software (PEST). The calibration of the offline WRF-Hydro gives Nash–Sutcliffe efficiencies between 0.56 and 0.64 and volumetric efficiencies between 0.46 and 0.81 for the six subcatchments. The comparison of the classic WRF and fully coupled WRF-Hydro models, both using the calibrated parameters from the offline model, shows only tiny alterations for radiation and precipitation but considerable changes for moisture and heat fluxes. By comparison with TERENO Pre-Alpine Observatory measurements, the fully coupled model slightly outperforms the classic WRF model with respect to evapotranspiration, sensible and ground heat flux, the near-surface mixing ratio, temperature, and boundary layer profiles of air temperature. The subcatchment-based water budgets show uniformly directed variations for evapotranspiration, infiltration excess and percolation, whereas soil moisture and precipitation change randomly.

Description: Coupled atmospheric–hydrologic systems are increasingly used as instruments for flood forecasting and water management purposes, making the performance of the hydrologic routines a key indicator of the model functionality. This study's objectives were (i) to calibrate the one-way-coupled WRF-Hydro model for simulating extreme events in Cyprus with observed precipitation and (ii) to evaluate the model performance when forced with WRF-downscaled (1×1 km2) re-analysis precipitation data (ERA-Interim). This set-up resembles a realistic modelling chain for forecasting applications and climate projections. Streamflow was modelled during extreme rainfall events that occurred in January 1989 (calibration) and November 1994 (validation) over 22 mountain watersheds. In six watersheds, Nash–Sutcliffe efficiencies (NSEs) larger than 0.5 were obtained for both events. The WRF-modelled rainfall showed an average NSE of 0.83 for January 1989 and 0.49 for November 1994. Nevertheless, hydrologic simulations of the two events with the WRF-modelled rainfall and the calibrated WRF-Hydro returned negative streamflow NSE for 13 watersheds in January 1989 and for 18 watersheds in November 1994. These results indicate that small differences in amounts or shifts in time or space of modelled rainfall, in comparison with observed precipitation, can strongly modify the hydrologic response of small watersheds to extreme events. Thus, the calibration of WRF-Hydro for small watersheds depends on the availability of observed rainfall with high temporal and spatial resolution. However, the use of modelled precipitation input data will remain important for studying the effect of future extremes on flooding and water resources.