Knowledge graph and dataset of global storms (1990–2023)

 Storms of the eleven storm subtypes recorded in the Emergency Events Database (EM-DAT) from 1990 to 2023, in 3,492 storm-country records, with the hazard, exposure, vulnerability and cascade context that can shape their impacts.

 For each record with administrative units from Geo-Disasters, hazard and exposure measures suited to its subtype were read over the storm's units and days in Google Earth Engine: ERA5 and ERA5-Land, the Gridded Population of the World v4, the Global Human Settlement Layer, SRTM, ESA WorldCover, the low elevation coastal zone grid v3, MERRA-2, CHIRPS and Global Mangrove Watch. Storm surge comes from the Global Tide and Surge Model v3, and best tracks from IBTrACS, with wind fields modelled in CLIMADA. Displacement records of the Internal Displacement Monitoring Centre (IDMC) and appeals of the International Federation of Red Cross and Red Crescent Societies (IFRC) are matched to storms. Vulnerability comes from UNDP's Human Development Index and the World Bank's World Development Indicators and Worldwide Governance Indicators. Cascades link storms to disasters that struck the same area within 91 days.

 Every value has a status column and, when empty, a gap column that gives the reason, and every link states its method and confidence.

 Files:
 - Tables of comma-separated values, one per class of the graph, linked by the keys in schema.yaml
 - storms_public.ttl.gz: the Resource Description Framework knowledge graph (Turtle), linked to Wikidata, with its vocabulary, SHACL shapes and VoID description
 - data_dictionary.csv: every column, its unit and its meaning
 - query1.rq–query6.rq: the competency questions in SPARQL
 Storms of the eleven storm subtypes recorded in the Emergency Events Database (EM-DAT) from 1990 to 2023, in 3,492 storm-country records, with the hazard, exposure, vulnerability and cascade context that can shape their impacts.

 EM-DAT's values are not redistributed, as its licence forbids it. Records are keyed by EM-DAT's disaster number, and users can obtain the values from CRED at https://public.emdat.be. The processing code rebuilds the full dataset, with the SQLite database and the complete graph, from a user's own export.