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

Ridwan Bin Monjur , NISFARIZA MOHD NOOR MARIS, Siti Hafizah Ab Hamid, Nur Nasuha Mohd Daud

From 1990 to 2023, storms accounted for 45.3% of recorded disaster damage and
18.2% of disaster deaths. Yet impact records do not say how strong a storm was
where it struck, who was exposed or why a value is missing. This knowledge
graph and dataset cover storms of eleven subtypes recorded in the Emergency
Events Database from 1990 to 2023, in 1,595 country-years and 3,492 storm-
country records. For those records with administrative units from Geo-Disasters,
we extracted hazard (e.g., wind speed, precipitation), exposure (e.g., population
density) and vulnerability (e.g., Human Development Index) measures suited to
each subtype over the storm's affected administrative units and days from multiple
sources including national and government indicators, an atmospheric reanalysis
and the Google Earth Engine. Displacement records of the Internal Displacement
Monitoring Centre and appeals of the International Federation of Red Cross and
Red Crescent Societies are matched to storms. Cascades link storms to disasters
that struck the same area within 91 days. Every method, from the modelled wind
fields to the cascade links, follows a published study, with few minor stated de-
viations. The data are released as a Resource Description Framework knowledge
graph linked to Wikidata and as tables of comma-separated values. The tables load into a SQLite database with the primary and foreign keys given in the schema