A global persistent heavy rainfall event dataset identified from hourly precipitation using a three-dimensional spatiotemporal connectivity algorithm

This database provides a global dataset of persistent heavy rainfall events (PHREs) derived from ERA5 hourly precipitation data for 1961-2025. The dataset was generated through heavy-precipitation grid-cell screening, three-dimensional 26-neighbour spatiotemporal connectivity analysis, and additional filtering based on event duration and precipitation intensity. It enables the complete evolution of PHREs to be tracked from initiation and development to dissipation. By accounting for both temporal continuity and spatial connectivity, the method captures heavy rainfall systems characterized by persistence, spatial organization, and substantial regional impacts. The database primarily contains two NetCDF4 data products. The event-level product includes 51,351 PHREs identified during 1961-2025, with each record representing an independent event. It provides information on the event ID, start and end times, duration, longitudinal and latitudinal extent, cumulative precipitation, time and location of maximum precipitation, maximum precipitation intensity, mean precipitation intensity, and affected area. Events can be efficiently filtered by year and month. The event-based gridded product stores hourly gridded precipitation data and the corresponding spatial locations for each event throughout its lifetime. It captures the movement, expansion, contraction, and intensity changes of the precipitation area within an event over time. The main variables include time, latitude, longitude, and hourly precipitation. Grid cells not belonging to the event are assigned missing values, so that only valid grid cells constituting the corresponding PHRE are retained. This data structure facilitates both the rapid retrieval of overall event characteristics and detailed analyses of the spatiotemporal evolution of individual events. The database also provides a variable dictionary, data documentation, event identification and processing scripts, and example Python code for reading the NetCDF4 products. The example code can be used to inspect the data structure and variable attributes, verify the number of event records, read event catalogues for specified years or months, extract valid precipitation grid cells for individual events, calculate event-based precipitation statistics, and visualize cumulative precipitation and its spatial distribution. The examples are intended to demonstrate basic data access and analysis procedures rather than to fully reproduce the figures presented in the accompanying paper. This dataset supports climatological studies of PHREs at global and regional scales, including analyses of the spatial distributions and long-term changes in event frequency, duration, affected area, cumulative precipitation, and precipitation intensity. It can also be used to investigate the seasonal variability, spatial migration, intensity categories, and regional differences of PHREs. In addition, the dataset provides a valuable data foundation for assessing changes in extreme precipitation, investigating flood risk, and examining the physical mechanisms responsible for persistent heavy rainfall.