Comment les sites sont choisis▸
Phase 1 — loss and damage. Only communities in districts with documented damage from the 2026 cyclones, Fytia and Gezani, are eligible for Phase 1 — 11 districts in the sourced damage register (BNGRC, UN OCHA/ReliefWeb, IFRC/Malagasy Red Cross, UNICEF, ACAPS, ECHO), which covers 52 districts across 2023-26 in all and stays attached to every site as evidence. Loss-and-damage finance answers for harm that has occurred, so the first phase is evidenced by harm, not by modelling. Click any Phase 1 site to see the cyclones and the reported impact behind it.
Phases 2 and above — predicted exposure. Elles sont anticipatives : les endroits les plus susceptibles d’être frappés ensuite. Exposition cyclonique = nombre de cyclones tropicaux sévères distincts (≥64 kt) passés à moins de 100 km, 1980–2026, d’après IBTrACS (archive NOAA/OMM des meilleures trajectoires), la même règle sur tous les littoraux. Portée : à moins de 50 km du rivage le plus proche (tout littoral). Besoin = population (recensement HDX 2018). Score = population × exposition ; exposition ≥ 3 retenue ; centres urbains exclus.
Impact des couches, par phase▸
| Phase | Sys | No water¹ | Rem² | In³ | Water⁴ |
|---|
| Phase 1 | 81 | 31 | 9 | 9 | 25 |
| Phase 2 | 80 | 43 | 4 | 13 | 54 |
| Phase 3 | 80 | 54 | 3 | 16 | 72 |
| Total | 241 | 128 | 16 | 121 | 128 |
¹ No water = no improved source within 2 km. ² Rem = sites removed from that phase's tier (operator present, or MEAH rehabilitating there); only Phase 1's are drawn on the map. ³ In = sites pulled up to fill the gaps (the backfill); it exceeds Removed mid-programme because each phase also refills the slots it passed upward (e.g. Phase 2 loses 4 and sends 6 to Phase 1, so pulls 10 from below); the final 23 are brand-new. ⁴ Water = sites reshuffled if the water-need weighting is switched on.
Cartes d’évaluation des dégâts (Copernicus EMS)▸
The EU's Copernicus Emergency Management Service was activated for Cyclone Gezani (EMSR867, landfall near Toamasina 10 February 2026, 19 deaths in Toamasina) and Cyclone Fytia (EMSR863, 30 January 2026), both requested by DG ECHO. It graded damage from very-high-resolution imagery, building by building, across 14 areas: 60,015 buildings affected. Toamasina alone: 32,930 of 236,112 assessed. Toamasina South West: 8,730 of 24,298. Toamasina North West: 4,528 of 11,397. Click a diamond for the breakdown by building class, the sensor and acquisition date, and the proposed sites within 40 km.
This is the hardest damage evidence in the pack. It is a count, not a description, made by an EU institution, and the products are CC BY 4.0, © European Union, so they can be reproduced in the proposal with credit and without asking permission. Grey diamonds are areas where flood extent was delineated but buildings were not graded. Earlier activations exist for Cheneso (EMSR645), Batsirai (EMSR564), Belna (EMSR415) and Ava (EMSR264) on the legacy service.
Imagerie des dégâts cycloniques▸
Published satellite imagery of the storms that struck the areas this programme proposes to work in, one marker per image, placed on the town its own caption names: Toamasina and Brickaville (Gezani, 10 Feb 2026, Category 3 at 200 km/h), the Sava coast (Gamane, 27 Mar 2024), the north-west and the Betsiboka (Cheneso, Jan 2023), and Mananjary and Manakara (Freddy, Feb 2023; Batsirai and Emnati, Feb 2022). Each popup carries the image, the storm, the date, the full credit line, a link to the source, and the proposed sites within 90 km — 51 of them around Toamasina and Brickaville alone.
All are NASA Earth Observatory images, which are public domain and may be republished with credit; that is why they are used rather than agency press photographs, which are not free to reuse. Where no image was published for a storm, the marker carries the impact record and its source instead of a picture. Note what these show: the storm and the flooding it left, at regional scale. They are not photographs of a damaged water point.
Points d’eau enquêtés : en service / hors service▸
A field survey that recorded, at GPS level, the state each water point was actually found in: 3,381 points, surveyed 2017-2026 and mostly in 2026. Red = not working, 1,236 in all (1,051 not functional, 79 partially functional and needing repair, 106 no longer there). The same survey found 2,145 points working; those are not drawn, to keep the map about what has failed. Click any point for its name, type, fokontany and survey date.
This is the strongest functional evidence we hold, because it is observed rather than inferred. Its limit is coverage. This is not a national survey: it covers Atsinanana 2,552, Analamanga 788, Ambatosolo 21, Boeny 20. Where it has not been run there are no dots, and that means not surveyed, never "all working" — Sava, Analanjirofo and Vatovavy-Fitovinany, which hold most of the programme, have no functional data at all. For that reason it is used as evidence and never as a ranking input.
Liste de réhabilitation du MEAH▸
The Ministry of Water's post-cyclone rehabilitation list, 38 locations still seeking financing (Atsinanana 8, Boeny 12, Melaky 15, Analamanga 3). Teal squares. Where a MEAH location resolves to the same fokontany as a ranked candidate, that candidate is excluded from the programme outright and the ranking closes up, so no clash is drawn on the map: 14 candidates were removed this way, 0 remain in conflict. A further set share a commune with our sites without a direct clash. Most of the MEAH list falls outside our coastal programme entirely, Melaky and inland Analamanga especially, so the two plans are largely complementary. Note the MEAH sheet gives no coordinates: locations are matched on administrative names, and each popup states whether it resolved to a fokontany or only to a commune or district centroid.
Switching the layer on re-phases the programme. The 38 MEAH locations are treated as delivered under Phase 1, so Phase 1 becomes 38 MEAH + the 43 highest-ranked sites (81 in all) and every ranked site below them shifts down one place, cascading through the later phases. Phase 1 is sized at 81 for exactly this reason. Population for a MEAH location is the 2018 census figure for its fokontany, or the commune average where it only resolved to a commune.
MEAH priority regions. The shaded, dashed areas are MEAH's five priority regions: Atsinanana, Analanjirofo, Boeny, Analamanga and Melaky. 150 of our 241 systems (62%) fall inside them, 499,957 people; 91 do not (398,335 people). The largest divergence is Sava, 59 systems and the highest severe-cyclone exposure on the coast, which is not among MEAH's five. In the other direction, Analamanga is a MEAH priority but is inland, so no coastal CRSDW site falls in it. Switching this layer on also restricts the programme to those regions: every site outside them drops out and the phases refill, in rank order, from the next candidates inside them. The phase sizes do not change, so the same 241 systems are delivered in a different place.
Trajectoires d’atterrissage des cyclones▸
The 38 cyclones that struck Madagascar at severe strength (Category-1 or stronger, the same bar used to select sites), 1980-2026, from the IBTrACS best-track archive (NOAA/WMO). Each line is the storm's observed approach and stops at landfall; it thickens as the storm intensifies, and the arrowhead marks the landfall point and direction. The category shown is the storm's peak Saffir-Simpson intensity on its approach. Colour = category: Cat 1 · Cat 2 · Cat 3 · Cat 4 · Cat 5. Future tracks are not shown: no agency forecasts individual cyclone paths years ahead, so the historical pattern of landfalls is the best guide to where future cyclones are likely to strike.