Convection-permitting ensemble design strategies within a UFS style framework
Glen Romine
NCAR
5 Mar, 1:30 pm, in 2155
Abstract:
The NGGPS process aims to move NOAA's model development activities toward a unified forecast system (UFS) to dramatically simplify the production suite. Simultaneously, efforts are underway to develop a skillful and reliable ensemble prediction system at convection-permitting horizontal grid spacing. The first operational implementation of a convection-permitting ensemble prediction system, known as the High Resolution Ensemble Forecast version 2 (HREF), is a multi-model conglomerate of well-tuned deterministic forecasts. Thus, the HREF is not in the spirit of the NGGPS process, yet it also provides both skillful and reliable guidance for high-impact weather that exceeds the performance of formally developed, single dynamic core (SDC), uniform-physics ensemble prediction systems, as demonstrated by several research teams from NOAA labs, universities, and other communities. Among these development teams, a group from NCAR, GSL, and EMC are developing community tools to enable a formal approach to convection-permitting ensemble design. These tools and application methods aim to improve the mean predictive skill of the model, within a continuously cycled data assimilation system, as a pathway to improve the analysis and subsequent forecasts. Additional tools will monitor error growth rates in the convection-permitting ensemble, providing much needed guidance to understand how best to boost ensemble dispersion within SDC uniform-physics ensembles, which are notoriously under-dispersive. The talk will introduce these tools and methods, demonstrate their use in ensemble design, and encourage discussion on how these approaches might be applied in the development of NOAA's future Rapid Refresh Forecast System.