Hydro-meteorological systems are widely applied to support the management of water resources as well as for flood protection purposes. Coupling of meteorological and hydrological models to predict streamflow is still challenging, mostly due to weather variables predictability. Uncertainty associated to the modeling chain play also a crucial role, especially in mountainous catchments, where many hydropower plants are located in order to exploit the available water resources and the gross hydraulic head.
Weather Research and Forecasting (WRF) model reanalysis data of past years are now becoming available at hourly time step and spatial resolution of about 2 km2 for the entire Alpine region, covering periods with different flow regimes, driven either by relevant snowmelting contributions or by intense convective precipitations in summer. To understand the WRF predictability in mountainous region and for prolonged time period, and as a consequence its suitability to drive hydrological models, a comparison between WRF-simulated and observed temperature and precipitation fields is performed in the Passirio river basin, a 400 km2 watershed located in the Upper Adige river catchment (Northern Italy), during the period 2013-2016. Furthermore, the effects of adopting such meteorological forcing on streamflow predictions is evaluated by means of two hydrological models adopting different schemes for simulating snowmelt and streamflow aggregation processes.
The two models are calibrated against observed streamflow time series using both observed and WRF meteorological data as input forcing. This comparison aims to understand the potential benefit on streamflow simulations due to a direct calibration using simulated meteorological variables instead of observed data as well as the effects of considering a uniform temperature field versus taking into account its spatial variability.
Preliminary results show a consistent performance of the WRF model in simulating temperature and precipitation fields throughout the year, confirming its suitability as meteorological input provider for hydrological models not only when dealing with time scales typical for flood simulation.
Streamflow predictions with both hydrological models reveal a good agreement with observed values while also forcing them with WRF data, which might be an attractive alternative to calibrating hydro-meteorological models in catchments not extensively covered by weather stations.