Data Assimilation in high resolution Numerical Weather Prediction models to improve forecast skill of extreme hydrometeorological events.

The complex orography typical of the Mediterranean area supports the formation, mainly during the fall season, of the so-called back-building Mesoscale Convective Systems (MCS) producing torrential rainfall often resulting into flash floods. These events are hardly predictable from a hydrometeorological standpoint and may cause significant amount of fatalities and socio-economic damages. Liguria region is characterized by small catchments with very short hydrological response time, and it has been proven to be very exposed to back-building MCSs occurrence. Indeed this region between 2011 and 2014 has been hit by three intense back-building MCSs causing a total death toll of 20 people and several hundred million of euros of damages. Building on the existing relationship between significant lightning activity and deep convection and precipitation, the first part of this work assesses the performance of the Lightning Potential Index, as a measure of the potential for charge generation and separation that leads to lightning occurrence in clouds, for the back-building Mesoscale Convective System which hit Genoa city (Italy) in 2014. An ensemble of Weather Research and Forecasting simulations at cloud-permitting grid spacing (1 km) with different microphysical parameterizations is performed and compared to the available observational radar and lightning data. The results allow gaining a deeper understanding of the role of lightning phenomena in the predictability of back-building Mesoscale Convective Systems often producing flash flood over western Mediterranean complex topography areas. Despite these positive and promising outcomes for the understanding highly-impacting MCS, the main forecasting issue, namely the uncertainty in the correct reproduction of the convective field (location, timing, and intensity) for this kind of events still remains open. Thus, the second part of the work assesses the predictive capability, for a set of back-building Liguria MCS episodes (including Genoa 2014), of a hydro-meteorological forecasting chain composed by a km-scale cloud resolving WRF model, including a 6 hour cycling 3DVAR assimilation of radar reflectivity and conventional ground sensors data, by the Rainfall Filtered Autoregressive Model (RainFARM) and the fully distributed hydrological model Continuum. A rich portfolio of WRF 3DVAR direct and indirect reflectivity operators, has been explored to drive the meteorological component of the proposed forecasting chain. The results confirm the importance of rapidly refreshing and data intensive 3DVAR for improving first quantitative precipitation forecast, and, subsequently flash-floods occurrence prediction in case of back-building MCSs events. The third part of this work devoted the improvement of severe hydrometeorological events prediction has been undertaken in the framework of the European Space Agency (ESA) STEAM (SaTellite Earth observation for Atmospheric Modelling) project aiming at investigating, new areas of synergy between high-resolution numerical atmosphere models and data from spaceborne remote sensing sensors, with focus on Copernicus Sentinels 1, 2 and 3 satellites and Global Positioning System stations. In this context, the Copernicus Sentinel satellites represent an important source of data, because they provide a set of high-resolution observations of physical variables (e.g. soil moisture, land/sea surface temperature, wind speed, columnar water vapor) to be used in NWP models runs operated at cloud resolving grid spacing . For this project two different use cases are analyzed: the Livorno flash flood of 9 Sept 2017, with a death tool of 9 people, and the Silvi Marina flood of 15 November 2017. Overall the results show an improvement of the forecast accuracy by assimilating the Sentinel-1 derived wind and soil moisture products as well as the Zenith Total Delay assimilation both from GPS stations and SAR Interferometry technique applied to Sentinel-1 data.