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

Jussi Leinonen,Urs Germann,Ulrich Hamann,John R. Mecikalski

doi:10.5194/nhess-22-577-2022

Abstract

Abstract. 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.

Highlights

Thunderstorms regularly cause significant risk to human life and damage to property through lightning, heavy precipitation, 20 hail and strong winds
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 found that the persistence assumption is biased: the MAXZ at t > 0 is, on average, lower than at t = 0; this can be seen in most of the examples of Fig. 3. This reflectivity bias is caused by a combination of two sources: first, sampling bias which occurs because we select centers of intensive thunderstorms with MAXZ > 37 dBZ, and second, the thunderstorm track drifting off the actual center of the storm due to inaccuracies in the tracking procedure

Summary

Introduction

Thunderstorms regularly cause significant risk to human life and damage to property through lightning, heavy precipitation, 20 hail and strong winds. These hazards are highly localized and develop within time scales ranging from tens of minutes to a few hours, which makes them difficult to forecast precisely using numerical weather prediction (NWP) models. Issuing localized short-term warnings of impacts is better achieved with nowcasting, statistical prediction of near-term developments based on the latest available data. 25 Various tracking and nowcasting systems for thunderstorms have been developed over the years since the 1960s, usually primarily using radar but sometimes combining other information such as lightning detection and location data. Other algorithms are designed to utilize satellite data instead; prominent examples of these include GOES-R Convective Initiation (Mecikalski and Bedka, 2006; Mecikalski et al, 2015), the Rapid Developing Thunderstorm (RDT; Autonès and Claudon, 2012) algorithm of 35 the Nowcasting Satellite Application Facility (NWCSAF), and Cb-TRAM (Zinner et al, 2008)

Objectives

Methods

Results

Conclusion