Naslov
Dynamical downscaling of the ECMWF experimental seasonal forecasts

Autori
Patarčić, Mirta ; Branković, Čedo

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
Abstracts of the EMS Annual Meeting (ECAC 2008). Vol. 5 / - , 2008

Skup
European Conference on Applied Climatology

Mjesto i datum
Amsterdam, Nizozemska, 29.09.2008. - 03.10.2008

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
seasonal forecasts; dynamical downscaling; RegCM

Sažetak
The 50-km Regional Climate Model (RegCM) was used to dynamically downscale ECMWF experimental seasonal integrations from the ENSEMBLES project. Downscaling has been done for nine member ensembles for winter (JFM) and summer (JAS) seasons for 11-year period (1991-2001). RegCM domain covered the central and southern Europe and the Mediterranean. Global and regional model errors and anomaly correlation coefficients were calculated against ERA-40 and CRU data for temperature and precipitation fields. Accuracy measures for categorical events, such as bias, hit rate, false alarm ratio, threat score and Kuipers skill score, were calculated for 2m temperature and precipitation. RegCM, on average, improves T2m and precipitation forecasts over the mountains of central and southern Europe, in particular for higher thresholds examined (severe events). However, it must be emphasised that part of the improvement seen in the RegCM results was attained because model’ s systematic errors acted in the direction that favoured higher skill scores. For precipitation, the improvement of accuracy measures with RegCM is genuine because, for this parameter, the model systematic error is relatively small. This improvement is seen not only over the mountains of the integration domain, but also in other areas with relatively inconspicuous orography. If the global model results are not very successful, the boundary forcing may adversely affect dynamical downscaling and therefore no (major) improvement can be expected. In the case of good global forecasts, dynamical downscaling can bring an improvement only if a regional model is free of relatively large systematic biases.

Izvorni jezik
Engleski

Znanstvena područja
Geologija

POVEZANOST RADA