Short term weather forecasting for shallow for landslide prediction

The paper describes the activities developed within the work package 4.1 of the UE/FP7 SAFELAND Project. Meteorological hazards as severe rain and convective outbreaks can trigger rapid shallow landslides; usually these phenomena give rise to severe soil impacts as life and properties loss in the most vulnerable areas. Some expected effects of the climate changes: as the modification in the precipitation patterns (more extreme rain events), in the amount of snow cover and in the ground temperatures, can alter the soil stability and can trigger in the future more landslides and, in general, decrease the slope stability supporting the vulnerability growth. The first scope of this research activity is firs of all to define a warning system for shallow landslide prediction, at large and slope scale, based on the forecast precipitation (rain and snow. The numerical codes evaluating the risk of shallow landslides after intense rainfall event have a very high resolution (order or 10 – 100 meters), then, they generally run on circumscribed area, while meteorological model, also the most resolute run with about 3 km of horizontal resolution. This resolution gap requires the definition of tools to connect the different simulation models (rain and soil instability) and the optimization of the numerical codes (accurate, robust, and efficient). Different components contribute to the definition of the chain; it follows a brief description of them. The first simulation models of the “chain” are represented by the Numerical Weather Prediction (NWP). Two different kind of NWP models are used: the IFS global (by ECMWF), for global forecast at 20 km of horizontal resolution and two version of the regional model COSMO-LM model one with 7 km of horizontal resolution and forecast range up to 72 hours and a second one with 2.8 km of horizontal resolution and up to 24 hours of forecast time range. This 2 step nesting is useful to provide forecast at the higher possible resolution. In order to avoid inconsistencies, due to the big jump of resolution beetwen the atmospherical models and the soil stability analysis, downscaling techniques have also been used. A tool to obtain downscaling of meteorological data in GIS environment has been developed, this is able to take as input file with hourly precipitation forecast and to produce as output raster file used as input for the code performing stability analysis. The evaluation of the soil saturation, using the 2 different versions of COSMO-LM will provide a first and a second level of warning (the second one more precise but available some hours after the first one). This data will be used as input into the basin scale hydrological-stability or slope stability model which will take into account also the soil depth, geotechnical parameters, soil permeability and slope gradient. The second scope of the research is to test this new system producing a high number of test cases permitting to make an evaluation of the performances of the system and to capture advantages and deficiencies. For this reason the “hydrometeorological simulation chain” has been tested on the instrumented Cervinara site (from 6 to 10 February 2007, from 6 to 7 March 2007, from 3 to 4 April 2007) and on the Tuscan region (from 20 December 2009 to 6 January 2010, from 30 October 2003 to 1st November 2003) and on the Ischia isle (from 29 April 2006 to 1st May 2006). The test cases regard different meteorological situations: advection rain events, with a long time period, convective rainfall happening in a short time periods and also days in which the rainfall is not intense. The stability models work on regional and slope scale and on soils with different properties.