⚠️ Work in progress. This site is in active development and is auto-generated from an internal knowledge base. Details may be incomplete, out of date, or inaccurate — treat figures and statuses as indicative, not authoritative.⚠️ Travail en cours. Ce site est en cours de développement et est généré automatiquement à partir d’une base de connaissances interne. Les détails peuvent être incomplets, obsolètes ou inexacts — considérez les chiffres et les statuts comme indicatifs, sans valeur officielle.

OCHA Anticipatory Action Frameworks

Knowledge Base

MadagascarTropical cyclone recently triggered

Anticipatory action framework mdg-storms

CountryMadagascar
HazardTropical cyclone
Monitoring periodNov–Apr (SW Indian Ocean cyclone season 1 Nov–30 Apr — stated in Summary/Method/Monitoring.)
Area of interestNational
Current version2024-12-13 · valid until 2026
Trigger windows
Readiness Wind speed forecast + distance to coast (Météo-Madagascar bulletin): ≥ 166 km/h (≈ 90 kt) forecast within 100 km of coast
Action Wind speed forecast + distance to coast (Météo-Madagascar bulletin): ≥ 166 km/h (≈ 90 kt) forecast within 100 km of coast
Pre-arranged financing$3.0M (CERF $3.0M)
Budget by sectorFood Security & Livelihoods $0.9M · WASH $0.5M · Health $0.4M · Shelter/NFI $0.3M · Protection $0.3M · Education $0.3M · Logistics $0.1M · CCCM $0.1M · Early Warning Messaging $90,000
Budget by agencyUNICEF $0.9M · WFP $0.8M · IOM $0.4M · WHO $0.3M · UNFPA $0.3M · FAO $0.3M
Single CERF $3M table by sector and agency (12 agency-sector rows: IOM=Shelter/NFI 300k+CCCM 120k; UNICEF=Early Warning 90k+Education 270k+WASH 450k+Protection 90k; UNFPA=Protection 210k+Health 90k; WHO=Health 330k; WFP=Food Security 600k+Logistics 150k; FAO=Food Security 300k).
Full breakdown as stated in the framework document (12 budget lines, $3,000,000)Ventilation complète telle qu’indiquée dans le document du cadre (12 lignes budgétaires, $3,000,000)
AgencySectorAmount
IOMShelter/NFI$300,000
IOMCCCM$120,000
UNICEFEarly Warning Messaging$90,000
UNICEFEducation$270,000
UNICEFWASH$450,000
UNICEFProtection$90,000
UNFPAProtection$210,000
UNFPAHealth$90,000
WHOHealth$330,000
WFPFood Security & Livelihoods$600,000
WFPLogistics$150,000
FAOFood Security & Livelihoods$300,000
Framework document2024-12-13 ↗
Code repositoryds-aa-mdg-storms ↗

Activation history

Real activations of this framework: the public trigger announcement and the CERF allocation that funded the response, with the people targeted and reached that CERF reports for the allocation (reported ~9 months after disbursement).Activations réelles de ce cadre : l’annonce publique de déclenchement et l’allocation du CERF qui a financé la réponse, avec les personnes ciblées et atteintes rapportées par le CERF pour l’allocation (rapporté ~9 mois après le décaissement).

DateWindowAnnouncementCERF allocationCERF approvedPeople targetedPeople reached
2026-02-09readiness + actionannouncementCERF-MDG-24-RR-1434 ↗$3.0M67,7730
Cyclone Gezani — Madagascar's FIRST anticipatory action for cyclones. Framework triggered 9 Feb 2026 when Météo-Madagascar forecasts met the Scenario 3 threshold (intense TC, ≥166 km/h, within 100 km of coast). BOTH windows fired — the readiness trigger (72–120 h) followed by the action trigger (≤72 h) — per the CHD team (@t-downing); public sources evidence only the 9 Feb activation. The full US$3M pre-arranged CERF envelope was allocated to six UN agencies (IOM, UNICEF, UNFPA, WHO, WFP, FAO), targeting >93,000 vulnerable people; of that, **~US$900,000 was actually disbursed ~1 day ahead of landfall** (the figure OCHA/WFP/UNICEF flash updates report as 'released ahead of landfall'). Gezani made landfall near Toamasina/Tamatave on 10 Feb 2026 (~17:30 UTC) as an intense TC — 10-minute sustained winds ~180–195 km/h with gusts ~230–250 km/h (WMO reports ~180 km/h sustained / 230 km/h gusts; Météo-France La Réunion peak 195 km/h). Confirmed via OCHA Gezani flash updates (Feb 2026), WFP, UNICEF + the Anticipation Hub.

Trigger statistics (current version)

Analysis period 1982–2026. Backtested years are when each trigger would have fired in the historical record — not real activations.Les années backtest sont celles où chaque déclencheur se serait déclenché dans l’historique — pas des activations réelles.

WindowReturn periodProbabilityBacktested activation years
Trigger3.5 years29%1984, 1994, 2000, 2002, 2004, 2007, 2008, 2009, 2014, 2017, 2019, 2022, 2026
Overall (any window)3.5 years29%