Does the Danube exist? Versions of reality given by various regional climate models and climatological data sets

[1] We present an auditing (intercomparison and verification) of several regional climate models (RCMs) nested into the same run of the same atmospheric global circulation model (AGCM) regarding their representation of the statistical properties of the hydrological balance of the Danube river basin for 1961-1990. We also consider the data sets produced by the driving AGCM, by the European Centre for Medium-Range Weather Forecasts (ECMWF) and National Centers for Environmental Prediction (NCEP) National Center for Atmospheric Research (NCAR) reanalyses. The hydrological balance is computed by integrating the precipitation and evaporation fields over the area of interest. Large discrepancies exist among RCMs for the monthly climatology as well as for the mean and variability of the annual balances, and only few data sets are consistent with the observed discharge values of the Danube at its Delta, even if the driving AGCM provides itself an excellent estimate. We find consistently that, for a given model, increases in the resolution do not alter the net water balance, while speeding up the hydrological cycle through the enhancement of both precipitation and evaporation by the same amount. Since the considered approach relies on the mass conservation principle and bypasses the details of the air-land interface modeling, we propose that the atmospheric components of RCMs still face difficulties in representing the water balance even on a relatively large scale. Their reliability on smaller river basins may be even more problematic. Moreover, since for some models the hydrological balance estimates obtained with the runoff fields do not agree with those obtained via precipitation and evaporation, some deficiencies of the land models are also apparent. The driving AGCM greatly overperforms the NCEP-NCAR and ECMWF 40-year (ERA-40) reanalyses, which result to be largely inadequate for representing the hydrology of the Danube river basin, both for the reconstruction of the long-term averages and of the seasonal cycle. The reanalyses cannot in any sense be used as verification. We suggest that these results should be carefully considered in the perspective of auditing climate models and assessing their ability to simulate future climate changes.