Application of WRF-based forecasts of total lightning threat to the CONUS

Methods of forecasting total lightning threat using proxy fields of graupel flux and vertically integrated ice (VII) from explicit convection WRF simulations have been devised in recent years in support of planned geostationary satellite monitoring of lightning (McCaul et al. 2009). As the launch date for GOES-R approaches, these methods continue to evolve and receive testing in additional model-based forecasting venues. Studies have indicated that these methods perform well even on meshes coarser than the original 2 km, although the graupel flux component of threat tends to underestimate the amplitude of the VII threat, suggesting the need to normalize its amplitude to match that of the VII component prior to blending of the threats. To test the method on larger scales in anticipation of operational use, the modified lightning threat algorithm was installed in the NSSL WRF system and examined during the 2010 NSSL/SPC Spring Experiment. For use on the 4 km mesh and the nonhomogeneous horizontal grid covering the entire contiguous United States, algorithm output for the graupel flux threat was boosted such that its peak value matched the VII threat peak. Results were then normalized by the space-varying map factor, to yield fields having units of flash origins per 5 min per square km.

Perusal of algorithm output and comparison with a series of cases having ground-truth lightning flash rate data from available lightning mapping array (LMA) systems shows the method provides reasonable results. However, the NSSL WRF tests reveal the occasional occurrence of very intense storms, for which the peak value of the graupel-flux threat exceeds that for VII, and for which observations indicate the VII threat may be too small. These findings suggest that, in such storms, with flash rate densities sometimes exceeding 25 flashes/(5 min) per sq. km, nonlinearities in the calibration of the VII threat may be surfacing. We suspect that for very high flash rates, the VII-based threat may reach a physical limit that does not exist for the graupel flux threat. When such nonlinearities arise, it may be appropriate to allow the larger of the threat values to be the standard that the other threat is forced to match before blending.