Assimilation of Lightning Observations and GOES Infrared Imagery at Convection-Permitting Scales

This paper explores new methods for assimilating lightning observations into numerical models and how geostationary infrared imagery can aid this process and serve as a basis to remove spurious simulated convection. Our simulations use the Weather Research and Forecasting (WRF) model at convection-permitting scales (3-4 km horizontal grid spacing) to explore these improved methods. Existing lightning assimilation methods nudge humidity and/or temperature profiles where lightning is observed. We present a new method that instead nudges the vertical velocity field to produce a storm where lightning is observed. However, this method, like most other lightning assimilation methods, typically does not produce a simulated storm until approximately 30 min after the observed strike. To account for this time lag, we use GOES 11 micron imagery to track the thunderstorm backwards in time to find a more suitable time and location to begin forcing the convection, i.e., not merely using the first strike location. We also use cloud top heights derived from GOES 11 micron imagery to justify the suppression of spurious simulated storms to complement the use of lightning observations to initiate observed storms. In lieu of GOES-R data, Earth Networks Total Lightning Network (ENTLN) observations and GOES-13 and GOES-14 11 micron observations are used. We note the limitations of these existing datasets and how lightning observations from the GOES-R Geostationary Lightning Mapper (GLM) and infrared imagery from the Advanced Baseline Imager (ABI) may improve our methods. We verify the precipitation simulations against NCEP Stage IV precipitation observations by computing fractions skill score (FSS), a neighborhood method, and bias score. Simulated temperature, winds, humidity, and pressure also are verified against radiosonde and METAR observations.