Lightning Simulations over the Boreal Zone: Skill assessment for different lightning parameterization schemes and model resolutions

The boreal latitudes have experienced more severe fire years during the last decades, often coinciding with anomalously high lightning occurrence. Several studies indicate that lightning might further increase with global warming, threatening the huge carbon pools above (forests) and below the ground (peatlands) by possibly shifting wildfire regimes. However, lightning predictions are typically very uncertain, because they either rely on empirical relationships based on the present climate, or they use coarse-scale climate scenario simulations in which the critical process of deep convection is parameterized and the detailed representation of land-atmosphere interactions is lacking. In this study, we evaluate lightning simulations of the NASA Unified-Weather Research and Forecasting (NU-WRF) model for four lightning parameterization schemes over a 550,000 km2 domain with the Great Slave lake in the north. Simulation experiments were run at a convection-parameterized (9 km) and a convection-permitting (3 km) spatial resolution. The evaluated lightning parameterization schemes were the pointwise product of convective available potential energy (CAPE) and convective precipitation rate, the Lightning Potential Index (LPI), the McCaul threat, and the Price and Rind parameterization (PR) scheme. The results of these indices were compared with observational data from the Canadian Lightning Detection Network (CLDN). The evaluation was done for six simulated lightning seasons (2015-2020, June-August each). The performance evaluation was done based on the spatial lightning flash density pattern, daily time series, average diurnal cycle, average seasonal climatology, and lightning frequency distribution. Our results showed that lightning simulated with NU-WRF output at convection-permitting resolution was consistently superior to lightning based on output at convection-parameterized resolution. Furthermore, the comparison of the different indices indicated that the overall performance of the LPI index and the CAPExP product were superior to estimates of the McCaul threat and the PR scheme. In the presentation, we will discuss performance differences and the implications for the application of the lightning parameterization schemes for future predictions of lightning in the boreal zone.