Anticipatory Action - Cuba Hurricanes
Preface
This book holds the analysis behind OCHA’s anticipatory action (AA) framework for hurricanes in Cuba. Most of it is code and output. The text explains what each step does and why.
The trigger has a forecast part and an observational part. Both look at a storm’s maximum wind speed while it is in, or forecast to be in, Cuba’s Zona de Máxima Atención (ZMA), together with two-day rainfall over the country. The forecast part uses NHC forecasts up to 3 days ahead and CHIRPS-GEFS forecast rainfall. The observational part uses IBTrACS storm tracks and IMERG satellite rainfall.
For each part we varied the wind speed threshold, the rainfall aggregation (the national mean, or the 50th, 80th, 90th or 95th percentile across pixels) and the rainfall threshold. Every combination kept had to trigger in 7 of the 25 years from 2000 to 2024, a return period of 3.7 years. We then compared combinations by how much past impact they would have caught, using people affected from EM-DAT and CERF allocations.
The working group chose one of these combinations, option 1b, in July 2025. The Summary chapter lists the storms it would have triggered. The trigger was redesigned in 2026, and that work is not covered here.