The Robustness of Lightning Proxies for Numerical Weather Prediction and Data Assimilation

Abstract

Abstract Numerical weather models rely on data assimilation of observations with previous model forecasts to obtain an initial state most representative of the current atmosphere. While radar is a vital tool used in assimilation, limitations result in lacking data across mountains, oceans, and dead zones. The Geostationary Operational Environmental Satellite-R East and West satel- lites launched beginning in 2016 have introduced a high detection efficiency lightning detec- tion system via the Geostationary Lightning Mapper (GLM). Although GLM has its own lim- itations, it avoids the coverage limitations of radar providing observations in radar-insufficient regions. With lightning’s capability to indicate storm cores and provide insight into storm in- tensity, GLM’s viability to supplement radar in assimilation is being explored. Studies have examined empirical relationships between flash extent density (FED), a high-resolution prod- uct of the GLM, and hydrometeor-related variables simulated by convection-allowing models to propose parameterizations, but they still have shortcomings. We propose and hypothesize a non- linear function weighted on multiple proxies will perform more skillfully as a forward operator connecting models to FED. This study looks at a diverse set of case data from two convection- allowing numerical weather models, the High-Resolution Rapid Refresh (HRRR) and the Rapid Refresh Forecast System (RRFS), and fits several function forms to the data by minimizing squared error. Comparison between these forms to each other and to previous parameteriza- tions through forecast verification techniques revealed " " skill with the " " form being the most skilled. With the continued operational implementation of HRRR and RRFS, this study provides insight and suggestions for improvements in the empirical relationships that model FED and in the assimilation of lightning data.