Graupel mixing ratio forecast from a cloud resolving numerical weather prediction model as a tool for lightning activity prediction

Abstract

Graupel mixing ratio over Bulgaria for the warm half year of 2021 based on the AROME-BG numerical weather prediction (NWP) model, is evaluated and connected lightning data detected by the ATDnet lightning location network. Lightning data and forecasted graupel mixing ratios were considered on resolutions of 5×5 km and 10×10 km with flash rate for one and three hours, as well on a daily base using upscaling neighborhood method. Two daily model runs are considered – at 06 and 18 UTC. Commonly used skill-scores in meteorological forecasts are used as evaluation metrics – probability of detection (POD), false alarm rate (F), proportion correct index (PC), and frequency bias index (FBI). Lightning probability forecast (based on graupel mixing ratio) is evaluated at diurnal, monthly, and spatial bases. Results show that graupel mixing ratio taken from the cloud resolving NWP model AROME-BG could be used as a tool to forecast lightning probability with a relatively high performance. Decreases of forecast spatial resolution and time frequency lead to improvement of forecast probability of detection (POD) and frequency bias index (FBI) and to a slight deterioration of its false alarm rate (F) and its percent correct (PC), and the impact of forecast time frequency is more pronounced.