Evaluation of the Added Value of Regional Ensemble Forecasts on Global Ensemble Forecasts

Yong Wang,Simona Tascu,Florian Weidle,Karin Schmeisser
DOI: https://doi.org/10.1175/waf-d-11-00102.1
2012-01-01
Weather and Forecasting
Abstract:The regional single-model-based Aire Limitee Adaptation Dynamique Developpement International-Limited Area Ensemble Forecasting (ALADIN-LAEF) ensemble prediction system (EPS) is evaluated and compared with the global ECMWF-EPS to investigate the added value of regional to global EPS models. ALADIN-LAEF consists of 16 perturbed members at 18-km horizontal resolution, while ECMWF-EPS includes 50 perturbed members at 50-km horizontal resolution. In ALADIN-LAEF, the atmospheric initial condition uncertainty is quantified by using blending, which combines large-scale uncertainty generated by the ECMWF-EPS singular-vector approach with small-scale perturbations resolved by the ALADIN breeding technique. The surface initial condition perturbations are generated by use of the noncycling surface breeding (NCSB) technique, and different physics schemes are employed for different forecast members to account for model uncertainties. The verification and comparison have been carried out for a 2-month period during summer 2007 over central Europe. The results show a quite favorable level of performance for ALADIN-LAEF compared to ECMWF-EPS for surface weather variables. ALADIN-LAEF adds more value to precipitation forecasts and has greater skill for 10-m wind and mean sea level pressure results than does ECMWF-EPS. For 2-m temperature, ALADIN-LAEF forecasts have larger spread, are statistically more consistent, but also have less skill than ECMWF-EPS due to the strong cold bias in the ALADIN forecasts. For the upper-air weather parameters, the forecast of ALADIN-LAEF has a larger spread, but the forecast skill of ALADIN-LAEF is from neutral to slightly inferior compared to ECMWF-EPS. It may be concluded that a regional single-model-based EPS with fewer ensemble members could provide more added value in terms of greater skill for near-surface weather variables than the global EPS with larger ensemble size, whereas it may have limitations when applied to upper-air weather variables.
What problem does this paper attempt to address?