Risk index, anticipatory action triggers and early warning system for cyclones

Total: 136,500 places missing for 173,000 exposed people. Mananara Nord combines the highest index, the lowest shelter coverage and the lowest warning coverage (31%). Fénérive Est has as many places missing in absolute terms as Maroantsetra but a lower risk thanks to its capacities. The index is a prioritisation tool, not a measurement: it depends on the scores assigned and must be discussed with the authorities. Shelters will never be enough: shelters, evacuation to safe buildings and staying with relatives outside the exposed area must be combined.

Triggering: Maroantsetra (65%) and Mananara Nord (55%) reach the 50% action threshold. Fénérive Est (45%) only reaches the readiness threshold: teams on alert, shelters checked, no transfer. Soanierana Ivongo (20%): monitoring.

Households: Maroantsetra: 52,000 / 5 = 10,400 coastal households × 40% = 4,160. Mananara Nord: 7,600 × 50% = 3,800. Total: 7,960 households.

Cost: 7,960 × 50 = 398,000 USD. Balance: 52,000 USD, i.e. 1,733 roofing kits (51,990 USD), with priority to Mananara Nord (highest risk). A small reserve may also be kept for Fénérive Est in case the forecast worsens, which reduces the number of kits.

Risks: (1) 30% of households, i.e. about 2,390, have no mobile money account: plan cash payment through an agent, which is slower and should be launched first; (2) the transfer must arrive before the winds and the network may go down: send payments within the first 12 hours; (3) kits must be installed in time: only where they are prepositioned; (4) markets may run short of stock: alert traders.

1. Risk knowledge: participatory mapping by fokontany of flood-prone areas, shelters and people who need help to evacuate (disability, age, pregnancy), updated every year before the season.

2. Monitoring and forecasting: systematic relay of meteorological service bulletins at district level, local thresholds linked to the protocol.

3. Dissemination and communication: messages in Malagasy, simple and action-oriented; multiple channels (community radio, SMS, town criers, flags); daytime dissemination; specific warning for fishers (associations, landing sites) before any trip to sea.

4. Response capability: two keyholders per shelter and a duplicate with the local committee; equipped shelters (lighting, separate latrines, space for women and children); designated buddies for each person with a disability; annual evacuation drill before the season; review after each event.

Priorities: priority 1, understanding disaster risk (participatory mapping); priority 4, enhancing disaster preparedness for effective response and to "build back better" (warning, shelters, anticipatory action). Priority 2 (governance, local committees) may also be cited.

Targets: target G (access to multi-hazard early warning systems and disaster risk information); target A (reduce disaster mortality); target B (affected people) is also relevant.

Indicators: (1) % of coastal households in the four districts receiving the warning at least 24 hours before the winds arrive (baseline 31 to 72%, target 90%), disaggregated by sex and disability; (2) % of shelters open and equipped within 6 hours of the evacuation order (target 100%); (3) % of people identified as needing help who are evacuated (target 95%). Mortality data monitored with the authorities, without making it an attributable project indicator.

The protocol was validated in advance precisely so that the decision is not reopened under pressure: the threshold is reached, the action is triggered. A probability of 55% means sometimes acting without impact: that is the accepted cost of anticipation, and the actions chosen are "no regret" (the money remains useful to households, as do the reinforced roofs).

I listen to him and respond to his concern: calm and precise messages, coordinated with him; it is not our organisation that orders evacuation, which remains his responsibility. I inform him that the transfer is going ahead, I refer the matter to my management and to the national disaster risk management authority, a signatory to the protocol, and I document the exchange. After the event, the protocol review will examine the trigger, including in the case of a false alarm.