Lightning jump in the NWC SAF RDT-CW product: application to MTG-LI, validation and comparison with Météorage

The main goal of the NWC SAF is to deliver an advanced, reliable system to support operational and research tasks in Nowcasting and Very Short Range Forecasting for meteorological users globally. This is achieved through the development and distribution of software packages like NWC/GEO for near real-time meteorological product generation. Météo-France is in charge of the two NWC SAF Convection products: RDT-CW (Rapid Developing Thunderstorms – Convection Warning) and CI (Convection Initiation). Both products primarily utilize geostationary satellite data across the entire disk and at the satellite frequencies.

The latest software release (v2025) is Meteosat Third Generation (MTG) compliant and fullfill Day-1 requirements. The v2025 already exhibits interesting improvements, in particular thanks to the Lightning Imager (LI) for the Yes/No convection decision of the RDT product. Indeed, RDT is designed to identify, track, and forecast convective cloud systems. To enhance severe weather nowcasting, RDT also computes several severity indicators within convective cells, including the detection of lightning jumps (LJ)—sudden increases in total lightning activity—which are recognized as reliable precursors of storm intensification and severe weather.

RDT can integrate both ground-based and satellite-based lightning observations, enabling the calculation of two independent LJ indicators. In this study, we evaluate the performance of lightning jumps derived from the LI and from Météorage, the French national ground-based lightning detection network. The evaluation is conducted using MTG satellite data and focuses on two main datasets: a European-wide analysis on August 13, 2024, using severe weather reports from the European Severe Weather Database (ESWD), and a set of case studies over France in 2025, enhanced with Maximum Estimated Size of Hail (MESH) data from the French operational radar network.

We present the methodology used to match lightning jumps with severe weather reports, as well as a novel approach for estimating the location and spatial extent of LJs within RDT cells. Performance scores are analyzed with respect to different LJ algorithms, lightning data sources, and day/night conditions. Results show that LJs derived from Météorage data generally achieve higher accuracy. Nevertheless, LJs detected by the LI provide promising signals particularly during nighttime hours.