Compound Weather Systems Dataset based on combinations of cyclones, fronts and thunderstorms

This compound weather system dataset is based on combinations of cyclones, fronts and thunderstorm environments, defined by their simultaneous occurrence in time and space. Details are provided in this reference publication for this dataset: Dowdy et al. (2025) available from https://iopscience.iop.org/article/10.1088/1748-9326/ae2527 This dataset uses the compound weather system framework originally developed by Dowdy and Catto (2017, doi:10.1038/srep40359) based on ERA-Interim reanalysis. This updated dataset presented here is based on the more recent ERA5 reanalysis (Hersbach et al. 2020, doi:10.1002/qj.3803). The weather system occurrences are indicated using environmental diagnostics applied to the ERA5 reanalysis data. The dataset uses 6-hourly time periods centred on 0000, 0600, 1200 and 1800 UTC throughout the time period from 1979 to 2020, on a 0.25-degree grid globally in longitude and from 69.25°N to 69.25°S in latitude. For each grid cell and time steps, the data files represent the different weather system combinations using the following values: 1 for Cyclone Only 2 for Front Only 3 for Thunderstorm Only 4 for the combination of Cyclone and Front 5 for the combination of Cyclone and Thunderstorm 6 for the combination of Front and Thunderstorm 7 for the combination of Cyclone, Front and Thunderstorm 0 for Other when none of those weather systems occur Cyclones were identified using the method described in Dowdy and Catto (2017), based on an updated version of the method described in Wernli and Schwierz (2006, doi:10.1175/JAS3766.1). Those cyclone data were also supplemented with tropical cyclone occurrences using IBTrACS data (accessed January 2025; Knapp et al. 2010, doi:10.25921/82ty-9e16) as was done in Dowdy and Catto (2017), with a range of +/- 2.5 degrees in latitude and longitude to estimate surrounding grid cells considered for the tropical cyclone data (following the global median sizes such as reported by Chavas et al. (2016: doi:10.1175/JCLI-D-15-0731.1)). Front data were produced using the method described in Sansom and Catto (2024, doi:10.5194/gmd-17-6137-2024) based on first identifying the zero contour lines of the Laplacian of the 850 hPa gradient of wet-bulb potential temperature. Locations along those zero contour lines are then found that exceed a threshold of the thermal front parameter following Hewson (1998: doi:10.1017/S1350482798000553), with front locations defined for this dataset where the gradient of wet-bulb potential temperature in the adjacent baroclinic zone exceeds a threshold value (based on Eq. 11 of Hewson (1998)). Uncertainties are noted in the front data in some high altitude regions of the Himalayas and Andes due to the use of 850 hPa data, similar to the case in previous studies such as Catto and Dowdy (2021: doi:10.1016/j.wace.2021.100313), such that results should be interpreted accordingly in those regions. Thunderstorm data were obtained from https://zenodo.org/records/16892383 v1.2 using environmental parameters derived from ERA5 reanalysis. The reference publication for those data is available from https://doi.org/10.1007/s00382-026-08076-5. Those thunderstorm data are an update to an earlier dataset based on ERA-Interim reanalysis as described in Dowdy (2020: https://doi.org/10.1007/s00382-020-05167-9). The data are based on the product of convective available potential energy (CAPE) and vertical wind shear, with calibration at each grid location such that the occurrence frequency of thunderstorm environments is equal to the number of observations-based thunderstorms (i.e., a form of quantile-quantile bias correction for consistency with observations at each individual grid location). Contact: [e-mail]