No abstract

This paper describes 27 years of scientific and operational achievement of Regional Cooperation for Limited Area Modelling in Central Europe (RC LACE), which is supported by the national (hydro-) meteorological services of Austria, Croatia, the Czech Republic, Hungary, Romania, Slovakia, and Slovenia. The principal objectives of RC LACE are to 1) develop and operate the state-of-the-art limited-area model and data assimilation system in the member states and 2) conduct joint scientific and technical research to improve the quality of the forecasts. In the last 27 years, RC LACE has contributed to the limited-area Aire Limitée Adaptation Dynamique Développement International (ALADIN) system in the areas of preprocessing of observations, data assimilation, model dynamics, physical parameterizations, mesoscale and convection-permitting ensemble forecasting, and verification. It has developed strong collaborations with numerical weather prediction (NWP) consortia ALADIN, the High Resolution Limited Area Model (HIRLAM) group, and the European Centre for Medium-Range Weather Forecasts (ECMWF). RC LACE member states exchange their national observations in real time and operate a common system that provides member states with the preprocessed observations for data assimilation and verification. RC LACE runs operationally a common mesoscale ensemble system, ALADIN–Limited Area Ensemble Forecasting (ALADIN-LAEF), over all of Europe for early warning of severe weather. RC LACE has established an extensive regional scientific and technical collaboration in the field of operational NWP for weather research, forecasting, and applications. Its 27 years of experience have demonstrated the value of regional cooperation among small- and medium-sized countries for success in the development of a modern forecasting system, knowledge transfer, and capacity building.

The regional single-model-based Aire Limité e Adaptation Dynamique Dé veloppement InternationalLimited 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.

Abstract. During winter cold strong winds associated with snowfalls are not unusual for South and Southeastern Romania. The episode of 2-4 January 2008 was less usual due to its intensity and persistence. It happened after a long period (autumn 2006-autumn 2007) of mainly southerly circulations inducing warm weather, when the absolute record of the maximum temperature was registered. The important snowfalls and snowdrifts, leading to a consistent snow layer (up to 100 cm), produced serious transport and electricity supply perturbations.Since this atypical local weather event was not correctly represented by the operational numerical forecasts, several cross-comparison numerical simulations were performed to analyze the relative role of the coupler/coupling models and to compare two ways of process-scale uncertainties mitigation: optimizing the forecast range and performing ensemble forecast through the perturbation of the lateral boundary conditions. The results underline, for this case, the importance of physical parametrization package on the first place and secondary, the importance of the model horizontal resolution. The resolution increase is beneficial only in the local process representation; on larger scale it may either improve or decrease the accuracy effect, depending on the specified nudging between large-scale and small-scale information. The event capture is likely to be favored by two elements: a more appropriate time-scale of the event's physics and the quality of the transmitted large-scale information. Concerning the time scale, the statistics on skill as a function of forecast range are shown to be a useful tool in order to increase the accuracy of the numerical simulations. Ensembles forecasting versus resolution increase experiments indicate, for such atypical events, an interesting supply in the forecast accuracy through the ensemble method when applied to correct the minimum skill of the deterministic forecast.

⎯ The 11 km regional ensemble ALADIN-LAEF (Aire Limitée Adaption Dynamique Développment InterNational-Limited Area Ensemble Forecasting) is evaluated by comparison with the 5 km deterministic model ALARO (ALADIN and AROME combined model-Application of Research to Operations at Mesoscale), in order to investigate the advantages and disadvantages facing short-range ensemble and high-resolution forecasts. To make rational decisions about the benefits or challenges of both systems, the forecast skill was measured through probabilistic and deterministic approaches over a 2-month period from late spring to summer season of 2013. The verification uses observations from 1219 SYNOP stations and 1 km × 1 km precipitation analysis from INCA (Integrated Nowcasting through Comprehensive Analysis) nowcasting system. The evaluation was carried out for three essential meteorological variables: 2 m temperature, 10 m wind speed, and 6-hour cumulated precipitation. From the probabilistic point of view, the results show that ALADIN-LAEF outperforms ALARO-LAGGED ensemble system, being statistically more reliable. From the deterministic point of view, the high-resolution deterministic system simulates better the precipitation forecast structure with respect to the observations. Compared to the ensemble system, the deterministic system cannot provide guidance concerning the forecast uncertainties or probabilities, making the ensemble products a powerful tool for risk assessment and decision making.