Thunderstorm occurrence data based on environmental conditions

Note: please use the latest version v1.2, available from: https://zenodo.org/records/16892383

The Broadscale Thunderstorm Environment (BTE) dataset provided here covers all tropical and mid-latitude regions throughout the world on a 0.25-degree grid in longitude and latitude (from 69.25°N to 69.25°S), at 6-hourly time steps (0000, 0600, 1200 and 1800 UTC) from 1979 to 2023 (noting potential for data back to 1950 and updates as more recent years become available). It is intended to be used for broadscale analysis including of climatological features relating to thunderstorm occurrence, similar to how an earlier dataset version for the Australian region (doi.org/10.1007/s00382-020-05167-9) was used in various subsequent studies.

A feature of this dataset is that it was produced using spatially varying thresholds of a diagnostic, with the threshold set at each location such that the resultant data from applying the diagnostic have the same occurrence frequency as the number of observed thunderstorms (using lightning occurrence data from 2012 to 2023 as an observations-based proxy for thunderstorm occurrence at a given location and time). The use of spatially varying threshold values means that at each individual grid location the average occurrence frequency of thunderstorm environments in the BTE dataset is consistent with the average occurrence frequency of observed thunderstorms.

The diagnostic used here is based on data for wind and for convective available potential energy (CAPE) obtained directly from ERA5 reanalysis. The diagnostic is calculated based on the product of (CAPE) and vertical wind shear from about 0 to 6 km above ground level (S06), with lower limits applied for CAPE ≥ 10 J/kg and S06 ≥ 10 m/s (i.e., all values lower than those limits are set equal to 10). For a given location, a time step for which a diagnostic is above its threshold value is considered indicative of the potential for thunderstorm occurrence, with the BTE dataset having a value of 1 for that time step (or a zero value if not), while noting potential for hazards to also occur at other times around this. Details on this method and dataset are available from doi:10.3389/fclim.2025.1539873, noting further information on development steps from ISBN: 978-1-925738-65-0.

In addition to ERA5 reanalysis, this diagnostic method was also previously applied to an ensemble of 12 global climate models (GCMs) selected due to having suitable 6-hourly data available at the time of this method's development. Information on those data are also available in doi:10.3389/fclim.2025.1539873, with ensemble average data provided here for the historical time period 1979 to 2005 (as well as noting availability for a future period of data from 2081 to 2100 under a high emissions pathway for those same GCMs).

Funding: This research received support from the ARC Centre of Excellence for Climate Extremes (CLEX), as well as the Melbourne Energy Institute (MEI), through the University of Melbourne.