WATTCH: World Archive of Tracked Tropical Cyclone Hindcasts

This Zenodo record provides access to the WATTCH dataset, a global collection of tropical cyclone (TC) track data derived from ECMWF IFS hindcast simulations, IBTrACS and ERA5 reanalysis. ECMWF produces ensemble hindcasts (reforecasts) using the same IFS model configuration as real-time forecasts but initialised on past dates. The hindcast system comprises 11 ensemble members (1 control plus 10 perturbations) run twice weekly at 9 km resolution with 6-hourly output to 15 days over a rolling 20-year period (to end of 2024). WATTCH contains tropical cyclone tracks and intensities from these IFS hindcast ensemble members (cycle 49r1), matched to observational records. Tropical cyclones are identified using the TRACK algorithm, which detects systems based on spatially filtered, vertically averaged relative vorticity (850–600 hPa) that persist for at least two days. Intensity is characterised by maximum 10-m wind speed and minimum mean sea level pressure. To establish a verification baseline, ERA5 reanalysis is tracked first using the same algorithm, then matched to IBTrACS best-track observations. This approach extends trajectories beyond the observational record, capturing early genesis and late decay stages often absent from IBTrACS. Finally, hindcast ensemble tracks are matched to these verified ERA5-IBTrACS events using a spatial separation threshold of 4°. This archive aims to identify plausible yet unprecedented TC tracks and intensities using the UNSEEN (UNprecedented Simulated Extremes using ENsembles) approach to support risk analysis and anticipatory action. Derived hazards relating to the TC tracks include Maximum Landfall Intensity (wind speed and surface pressure), Rapid Intensification (RI) and Extreme Rapid Intensification (ERI), RI and ERI close to landfall, TC stall duration and location. Related Publications The evaluation and methodology of the WATTCH dataset are featured in the following articles: WATTCH: World Archive of Tracked Tropical Cyclone Hindcasts for characterising UNSEEN hazards (Hooker et al., 2026, in prep). Hurricanes that haven’t happened, yet: Identifying unprecedented tropical cyclone scenarios (Heinrich et al., 2026, submitted to NHESS). Users are encouraged to check this Zenodo record for updated citations as these papers become available. A Digital Object Identifier (DOI) has been reserved for this project to ensure its persistent identification and citable access. The full dataset, documentation and associated metadata are provided in this record. Dataset Contents The WATTCH dataset consists of two primary CSV files for each hemisphere (Northern Hemisphere - NH, Southern Hemisphere - SH): `WATTCH_[NH/SH].csv`: Contains detailed 6-hourly track data points for individual tropical cyclones, including their geographical location, meteorological parameters, and derived hazard metrics (Rapid Intensification, Stalling, etc.). `WATTCH_landfalls_[NH/SH].csv`: Provides a summary of each detected landfall event, including entry/exit points, duration over land, and Maximum Landfall Intensity (MLI). Contact For more information, please contact [e-mail].