Analysis of the streamflow extremes and long-term water balance in the Liguria region of Italy using a cloud-permitting grid spacing reanalysis dataset

Abstract

Abstract. The characterization of the hydro-meteorological extremes, in terms of both rainfall and streamflow, and the estimation of long-term water balance indicators are essential issues for flood alert and water management services. In recent years, simulations carried out with meteorological models are becoming available at increasing spatial and temporal resolutions (both historical reanalysis and near-real-time hindcast studies); thus, these meteorological datasets can be used as input for distributed hydrological models to drive a long-period hydrological reanalysis. In this work we adopted a high-resolution (4 km spaced grid, 3-hourly) meteorological reanalysis dataset that covers Europe as a whole for the period between 1979 and 2008. This reanalysis dataset was used together with a rainfall downscaling algorithm and a rainfall bias correction (BC) technique in order to feed a continuous and distributed hydrological model. The resulting modeling chain allowed us to produce long time series of distributed hydrological variables for the Liguria region (northwestern Italy), which has been impacted by severe hydro-meteorological events. The available rain gauges were compared with the rainfall estimated by the dataset and then used to perform a bias correction in order to match the observed climatology. An analysis of the annual maxima discharges derived by simulated streamflow time series was carried out by comparing the latter with the observations (where available) or a regional statistical analysis (elsewhere). Eventually, an investigation of the long-term water balance was performed by comparing simulated runoff ratios (RRs) with the available observations. The study highlights the limits and the potential of the considered methodological approach in order to undertake a hydrological analysis in study areas mainly featured by small basins, thus allowing us to overcome the limits of observations which refer to specific locations and in some cases are not fully reliable.