Improving NWP Model Cloud Forecasts Using Meteosat Second-Generation Imagery

Abstract

Abstract The cloud mask of the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) is a nowcasting Satellite Application Facility (SAF) that is used to improve initial cloudiness in the High-Resolution Limited-Area Model (HIRLAM). This cloud mask is based on images from the Meteorological Satellite (Meteosat) Second Generation (MSG) satellite. The quality of the SAF cloud mask appeared to be better than initial HIRLAM clouds in 84% of the cases. Forecasts have been performed for about a week in each of the four seasons during 2009 and 2010. Better initial clouds in HIRLAM always lead to better cloud predictions. Verification of forecasts showed that the positive impact is still present after 24 h in 59% of the cases. This is remarkable, because initial dynamics was kept unchanged. The magnitude of the positive impact on cloud predictions is more or less proportional to the initial cloud improvement, and it decreases with forecast length. Also, forecast 2-m temperatures are affected by initial clouds. The generally positive bias of the 2-m temperature errors becomes a few tenths of a degree larger during the night but it decreases a comparable amount during daylight, because MSG tends to increase the cloud amounts in HIRLAM. The standard deviation of the errors often improves slightly in the first part of the forecast, indicating that forecast temperatures correlate better with observations when MSG is used for initialization. For longer lead times, however, standard deviations deteriorate a few tenths of a degree in seven of the eight verification periods, which all had a length of about a week.