Nowcasting thunderstorm hazards using machine learning: the impact of data sources on performance

In order to aid feature selection in thunderstorm nowcasting, we present an analysis of the utility of various sources of data for machine-learning-based nowcasting of hazards related to thunderstorms. We considered ground-based radar data, satellite-based imagery and lightning observations, forecast data from numerical weather prediction (NWP) and the topography from a digital elevation model (DEM), ending up with 106 different predictive variables. We evaluated machine-learning models to nowcast storm track radar reflectivity (representing precipitation), lightning occurrence, and the 45 dBZ radar echo top height that can be used as an indicator of hail, producing predictions for lead times of up to 60 min. The study was carried out in an area in the Northeastern United States for which observations from the Geostationary Operational Environmental Satellite-16 are available and can be used as a proxy for the upcoming Meteosat Third Generation capabilities in Europe. The benefits of the data sources were evaluated using two complementary approaches: using feature importance reported by the machine learning model based on gradient-boosted trees, and by repeating the analysis using all possible combinations of the data sources. The two approaches sometimes yielded seemingly contradictory results, as the feature importance reported by the gradient-boosting algorithm sometimes disregards certain features that are still useful in the absence of more powerful predictors, while, at times, it overstates the importance of other features. We found that the radar data is the most important predictor overall. The satellite imagery is beneficial for all of the studied predictands, and therefore offers a viable alternative in regions where radar data are unavailable, such as over the oceans and in less-developed ares. The lightning data are very useful for nowcasting lightning but are of limited use for the other hazards. While the feature importance ranks NWP data as an important input, the omission of NWP data can be well compensated for by using information in the observational data over the nowcast period. Finally, we did not find evidence that the nowcast benefits from the DEM data.