Abstract
Thunderstorms cause natural hazards, including hail, floods, strong winds, and lightning. Simulating thunderstorms in climate models is challenging due to their small scale, the complexity of their physical drivers, and the need to parameterize subgrid processes. Thunderstorm activity can be inferred by identifying relevant historical environmental parameters, e.g. convective available potential energy, humidity, and wind shear, and building statistical models that use these parameters as proxies for thunderstorm occurrence. Climate model projections of the parameters can be used with the statistical models to assess future thunderstorm activity. In this context, a multi-decade dataset with lightning flash totals and 201 convective parameters has been compiled for North America, focusing on areas north of 40°N. Parameters from the European Centre for Medium-Range Weather Forecasts reanalysis version 5 are available at 3-hour intervals on a 0.25° grid. The same variables are calculated for HighResMIP climate model simulations at 6-hour intervals for the historical period coinciding with global warming of 1°C above preindustrial and future periods at 2°C, 3°C, and 4°C warming.