Rainfall data assimilation in a convection permitting model for improving meteo-hydrological forecasts over Liguria

During Autumn 2014, Liguria region in northern Italy was affected by a number of severe weather episodes associated with floods and damages, including a devastating event on 9 October. This coastal area, characterized by complex orography close to the coastline, is prone to heavy precipitation episodes associated with mesoscale convective systems or frontal passage and usually sustained by southerly moist low-level jet feeding the precipitation. The small dimensions of the catchments and their very short response time, make rainfall predictions necessary to drive rainfall-runoff models in order to provide streamflow forecasts. Therefore, improving quantitative precipitation forecasts (QPF) at short range is critical and to attain this aim a rainfall assimilation scheme, based on nudging of humidity profiles, has been applied to a convection-permitting meteorological model (MOLOCH). The meteorological model is then coupled with the operational hydrological forecasting chain and the whole system is implemented for a number of rainfall episodes, taking into account operational requirements. SAL object-oriented methodology has been used to verify QPF, but also discharge predictions have been evaluated in a probabilistic perspective using a proper rainfall downscaling technique and the reduction continuous rank probability score to compare the results with benchmark. Results show that although the positive impact of the assimilation lasts only few hours during the forecast, it can improve the hydrological prediction, especially in case of high flow. Moreover, assimilation is more effective when non-equilibrium convection dominates and thus an accurate prediction of the local triggering for the development of the precipitation system is required.