Dataset for "Convective Environments Over the Arabian Peninsula in Current and Future Climates: A Machine Learning Approach"

This dataset provides convection indices or predictors derived from ERA5 over the Arabian Peninsula, covering the period 2001 - 2024 at 1 degree x 1 degree and a 6-hour interval. Additionally, the 6-hourly accumulated precipitation from IMERG V07 is included. The dataset includes raw (not bias-corrected) convective indices estimated from the historical and SSP5-8.5 experiments from CMIP6. However, only the years corresponding to the current regional warming level (+1.22°C above the preindustrial level) and to two future levels (+2.22°C and +4.22°C) are provided. The indices were computed using the R package MeteoMate and are provided in HDF5 format. Bias correction was applied to the CMIP6 indices using the MBCn algorithm from the R package MBC prior to estimating convection probabilities. The repository includes trained random forest (RF), extreme gradient boosting (XGB), and deep learning (DL) models for detecting or estimating the probability of convective environments. Additional information can be found in the README file. The dataset is intended to foster research on convective environments, extreme precipitation, convection-permitting modelling and climate change impacts over the region.