A014-04
Toward a continental-scale convective-allowing UFS model initialized with GOES-R lightning data
Monday, 7 December 2020: 05:58
Virtual
Amanda Back1, Stephen Weygandt2, Curtis Alexander2, Stan Benjamin3, Ming Hu4, Guoqing Ge5 and Eric James6, (1)Cooperative Institute for Research in the Atmosphere, Fort Collins, CO, United States, (2)NOAA Earth System Research Laboratory, Boulder, CO, United States, (3)NOAA Global Systems Laboratory, Boulder, CO, United States, (4)CIRES/University of Colorado and NOAA/ESRL/GSD, Boulder, CO, United States, (5)Earth System Research Laboratory, Boulder, CO, United States, (6)Cooperative Institute for Research in Environmental Sciences, Boulder, CO, United States
Abstract:
Coordinated work amongst NOAA groups is underway to develop a new UFS-based short-range weather prediction system, to be known as the Rapid Refresh Forecast System (RRFS). It will employ a limited-area version of the FV3 grid within a convective-allowing weather forecast model and assimilation system, potentially encompassing all of North America at about 3 km horizontal resolution. This entire domain, including large oceanic regions adjacent to the continent, must be initialized with information about ongoing convective storms. Whereas observations from ground-based sources diminish in quality and availability further from land, the GOES-R instruments and products, with fields of view covering most of the hemisphere, offer needed information throughout much of the domain. In particular, the Geostationary Lightning Mapper (GLM) supplies information about storm-scale behavior such as nascent and ongoing convection, complementing radar data available only near land.
The utility of GLM to initialize the planned RRFS is demonstrated through a set of experiments assimilating GLM into current models that share many features with the future RRFS. In the current-generation 13-km Rapid Refresh (RAP), experiments show the ability of GLM to initialize convection over a continent-scale domain including large oceanic regions, also demonstrating the accumulated impact of hourly cycled GLM assimilation. The RAP forecast fields then provide backgrounds for a 3-km nest over the Caribbean (HRRR-CAR). We study the effect of GLM ingest on HRRR-CAR forecasts, both from direct assimilation into the 3-km system and indirectly from the cycled RAP. Impacts of assimilation are measured against model runs that lack GLM assimilation, with emphasis on regions outside radar coverage. Experiments using current NOAA models (RAP/HRRR) are complemented by preliminary GLM assimilation experiments in our evolving prototype RRFS.
In all experiments, GLM data are ingested along with ground-based lightning and radar data where available, and configurations with and without GLM assimilation are contrasted both qualitatively and quantitatively. Experimental convective weather forecasts are compared to lightning flash density observations from GLM, demonstrating the potential of GLM for model assessment over data-scarce regions.
