A global dataset of spatiotemporal drought events from reanalysis and hydrological model data for 1980–2024

This dataset contains a global catalogue of spatiotemporally clustered drought events spanning 1980–2024, identified using a three-dimensional implementation of the DBSCAN algorithm. The catalogue focuses on persistent, spatially coherent large-scale drought events and is therefore intended primarily for regional-to-global drought analyses rather than local-scale drought monitoring. It consists of two complementary components for each of the four analyzed variables—precipitation (PRE), potential evapotranspiration (PET), runoff (Q) and soil moisture (SM): NetCDF files: Four global gridded files that store the 3D DBSCAN cluster identification for each variable at 0.5° spatial and daily temporal resolution (1980–2024) named:- PRE_ERA5_3Ddbscan_global_res0p5_1980_2024_1day.nc- PET_ERA5_3Ddbscan_global_res0p5_1980_2024_1day.nc- Q_mHM_3Ddbscan_global_res0p5_1980_2024_1day.nc- SM_mHM_3Ddbscan_global_res0p5_1980_2024_1day.ncwhere each file provides, for every grid cell and day, the integer cluster identifier corresponding to the drought event detected at that location and time. R-catalogue tables: Two accompanying R data files {.rds} per variable that summarize cluster characteristics in tabular form named:- summary_<var>.rds – one record per drought event containing event-level characteristics such as start and end date, duration, severity, centroid location, and maximum/minimum affected area.- stats_<var>.rds – daily time-step records describing the evolution of individual drought events through time, including event area, severity, centroid position, displacement, and propagation characteristics.Cluster identifiers are consistent across the NetCDF and RDS products, allowing users to directly link spatial cluster maps with corresponding event-level and time-step statistics. All spatial products are provided on a regular global 0.5° latitude–longitude grid (WGS84). In addition to the data products, the repository includes a custom R script (3DDBSCAN_v1.0.R) developed specifically for this study to enable efficient large-scale spatiotemporal clustering and event tracking. This script builds upon the original three-dimensional DBSCAN framework proposed by Cammalleri & Toreti (Cammalleri, C. & Toreti, A. (2023). A Generalized Density-Based Algorithm for the Spatiotemporal Tracking of Drought Events. Journal of Hydrometeorology, 24(3), 537–548.), with adaptations and extensions tailored to the requirements of global, multi-variable drought analysis. This dataset corresponds to a manuscript submitted to Scientific Data (Collection: Droughts): Stovicek et al. (2025): A global catalogue of spatiotemporal drought events derived from multivariate hydrological data. Future updates may be released as new repository versions, to extend the database, ensuring the dataset remains a living resource for climate, hydrology, and risk‑assessment research.