Development of a self-consistent lightning NOx simulation in large-scale 3-D models

We seek to develop a self-consistent representation of lightning NOx (LNOx) simulation in a large-scale 3-D model. Lightning flash rates are parameterized functions of meteorological variables related to convection. We examine a suite of such variables and find that convective available potential energy and cloud top height give the best estimates compared to July 2010 observations from ground-based lightning observation networks. Previous models often use lightning NOx vertical profiles derived from cloud-resolving model simulations. An implicit assumption of such an approach is that the postconvection lightning NOx vertical distribution is the same for all deep convection, regardless of geographic location, time of year, or meteorological environment. Detailed observations of the lightning channel segment altitude distribution derived from the NASA Lightning Nitrogen Oxides Model can be used to obtain the LNOx emission profile. Coupling such a profile with model convective transport leads to a more self-consistent lightning distribution compared to using prescribed postconvection profiles. We find that convective redistribution appears to be a more important factor than preconvection LNOx profile selection, providing another reason for linking the strength of convective transport to LNOx distribution. Plain Language Summary Lightning is a major source of nitrogen oxides (NOx = NO + NO2) and can significantly affect the chemistry and atmospheric composition in the upper troposphere. We develop a self-consistent representation of lightning NOx (LNOx) simulation in a regional 3-D model. After testing a large suite of meteorological and microphysical variables related to convection, we find that the parameterization using convective available potential energy and cloud top height gives the best flash rate estimates compared to July 2010 observations from ground-based lightning observation networks over the United States. With the advancements of meteorological models, we show that observation-based preconvection LNOx emission profiles can be directly applied in 3-D model simulations. The redistribution of LNOx by simulated convective transport produces satisfactory results compared to available in situ observations. The effect of convective redistribution is found to be larger than the prescribed preconvection LNOx profile.