Assimilation of lightning data in the Rapid Refresh and evaluation of lightning diagnostics from nested HRRR forecasts

Lightning data from the GLD 360 and NPN networks, along with BLM lightning data from Alaska, are being assimilated in the real-time experimental Rapid Refresh (RAP) system run at ESRL / GSD. The GLD 360 and Alaska data provide coverage over regions not observed by the WSR-88D radar network, allowing for the initialization of convective storm areas that likely would not otherwise be represented in the RAP initial fields. The hourly RAP lightning data assimilation algorithm involves first computing a lightning flash rate density within each RAP model grid square over a 20 minute period (from 15 min. before the hour to 5 min. after the hour). This flash rate density is then converted to a proxy column maximum reflectivity value using seasonally dependent empirical relationships derived from composited comparisons of national mosaic reflectivity data and NLDN lightning flash density data. This proxy column maximum reflectivity data is then converted to a vertical profile of reflectivity using seasonally dependent empirical relationship derived from the 3D national reflectivity mosaic data. These 3D proxy reflectivity data are then used within the digital filter-based reflectivity assimilation algorithm within the Rapid Refresh. Within this algorithm, 3D reflectivity is converted to a three-dimensionally varying latent heating based temperature tendency, which is applied in place of the heating from the explicit microphysics scheme and cumulus parameterization during the forward model integration portion of the digital filter.

In addition to assimilating lightning data within the Rapid Refresh, we are creating lightning threat diagnostic fields from the High Resolution Rapid Refresh (HRRR) that it run as nest within the RAP. These lightning threat diagnostic fields are computed using McCaul's formulation, which assesses graupel flux at -15 deg C (highlights lightning near the storm core) and vertically integrated ice content (highlights lightning throughout the storm including the anvil and stratiform precipitation region). These indicators are output individually in addition to a combined lightning threat indicator. We have begun monitoring and evaluating these predictors to assess their utility for anticipating lightning risk in forecast convective storms.

At the workshop, a comparison of real-time RAP forecast cycles with and without the lightning assimilation will be shown, with an emphasis on evaluating differences over oceanic regions and other regions outside of CONUS, where there is no radar data coverage. In addition we will report on our assessment of the utility of the lightning threat diagnostics from HRRR runs nested within the RAP.