A satellite lightning observation operator for storm-scale numerical weather prediction

Pauline Combarnous, Félix Erdmann, Olivier Caumont, Éric Defer, Maud Martet

Natural hazards and earth system sciences · 06 September 2022 · Météo-France (incl. CNRM)

In one sentence. Machine-learning regression maps AROME proxy variables to pseudo MTG lightning flash extent, giving a learned observation operator for storm-scale EnVar assimilation.

Abstract

Abstract. The Lightning Imager (LI) onboard the Meteosat Third Generation (MTG) satellites will provide total lightning observations continuously over Europe with a spatial resolution of a few kilometers. The objective of this study is to prepare the assimilation of the flash extent accumulation (FEA) measured by LI in the French storm-scale regional AROME NWP system within a new EnVar assimilation scheme, by developing a lightning observation operator. This study relies on pseudo LI FEA observations as LI on MTG is still to be launched in the end of 2022, meaning actual observations will be available by 2023. Since neither flashes nor the electric field are predicted by the AROME NWP system, the observation operator relies on proxy variables to link the flash observations to the prognostic variables of the NWP system. A total of 8 storm parameters were selected from a literature review to be used as proxies. Two different proxy types emerged from this literature review: microphysical and dynamical proxies. The proxies are calculated from 1 h AROME forecasts from the assimilation cycle. Machine learning regression models are used to relate observed FEA and the simulated proxies. The training of the observation operator is performed on a dataset of 44 days and 3 additional days are used for the validation. The data are processed as a climatology over the whole domain (i.e. France) and time period. The performances of each proxy are evaluated by computing Fraction Skill Scores (FSS) between observed FEA and proxy-based FEA. The present study suggests that microphysical proxies seem to be more suited than the dynamical ones to model satellite lightning observations with the AROME NWP system. The performances of multivariate regression models are also evaluated by combining several proxies after a feature selection based on a principal component analysis and a proxy correlation study but no proxies combination yielded better results than microphysical proxies alone. Results are also compared to those obtained with another lightning calibration, i.e., a relationship between lightning and proxies, from the literature but the simulated FEA amplitudes were systematically lower than the observed ones. Finally, different accumulation periods of the FEA had little influence, i.e., similar FSS, on the observation operator's ability to reproduce the observed FEA.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.