Abstract : 3P.3
Establishing a COSMO-LEPS model climatology to improve the warning of extreme events
The limited area ensemble prediction systems of the COSMO consortium COSMO-LEPS has proven to reliably forecast extreme weather events like e.g. the flooding event in August 2005 in Switzerland and surrounding countries.
In this study, an extreme forecast index (EFI) is derived for the COSMO-LEPS as a new warning product for extreme weather events. Contrary to probability forecasts that rely on an absolute threshold to rank a forecast as extreme, the EFI involves the climatology of the model used for the forecasts. This is done by comparing the probability distribution of the ensemble forecast to the underlying model climatology. An advantage of using the model climatology instead of the observed climatology is the insensitivity to a possible model bias. To achieve a model climatology we run re-forecasts with the COSMO-LEPS model setup for the period from 1971-2000.
The characteristic of the COSMO-LEPS model climatology is investigated and compared to the observed climatology. This unveils biases in the model and shows where particularly the usage of an EFI is advisable. Further on, it is shown that the sensitivity of the model climatology on the forecast lead-time is marginal. Finally, variant definitions of the COSMO-LEPS EFI are discussed and first results using the COSMO-LEPS EFI as a warning tool are shown.