Skill and economic benefits of dynamical downscaling of ECMWF ENSEMBLE seasonal forecast over southern Africa with RegCM4
Int. J. Climatol., (Jun 1, 2015)DOI: 10.1002/joc.4375
Abstract
This study presents an assessment of the skill and economic benefits of a seasonal forecast carried out by nine-member ensembles of regional and global models over southern Africa for the period of 1991–2002. The regional ensemble seasonal hindcasts were conducted at a resolution of 25 km by driving RegCM4 with the corresponding ensembles of the European Centre for Medium-Range Weather Forecasts (ECMWF) seasonal hindcasts. Skill and cost/loss ratio analysis were carried out on the probabilistic hindcasts to assess how much added value a regional model would bring over the driving ECMWF forecasts. RegCM4 shows an added value over most areas of southeastern Africa where the skill of the global circulation model (GCM) is already positive. In terms of economic benefits, RegCM4 brings additional benefits for southeastern Africa either by benefiting more users with interest in larger cost/loss ratio or by providing a higher economic value for users with small cost/loss ratio. Additional RegCM4 simulations with the same resolution were performed using a near 'perfect' boundary conditions with 'continuous (climate)' and 'reinitialized' modes to test the impact of land surface initialization/ spin up time on the seasonal mean precipitation simulation. Evaluation of the 'perfect' boundary forced simulations reveals that the regional model not only generates finer spatial-scale features that are missing in the ECMWF GCM, in terms of the December to April (DJFMA) climatology, but also reduced the wet bias of the driving GCM. Land surface initialization is found to have less impact on the DJFMA seasonal mean rainfall for forecasts initialized on 1 November, suggesting that 1 month is enough to recover from the shock of poor soil moisture initialization.