About the storm alerts

These alerts help humanitarian organisations monitor tropical cyclones and their potential impact on populations. All forecast data originates from the US National Hurricane Center (NOAA NHC). Population exposure estimates are derived from NHC forecasts by WFP ADAM, GDACS, and the OCHA Centre for Humanitarian Data. Alerts are sent automatically when a storm is forecast to affect a subscribed country.

When will I receive an email?

NHC issues tropical cyclone advisories four times a day at 03:00, 09:00, 15:00, and 21:00 UTC. The pipeline processes each advisory shortly after it is released and sends an email if at least one storm has non-zero forecasted wind exposure in any monitored country.

You will only receive an email for the country lists you are subscribed to. If a storm affects Cuba and Haiti but not Jamaica, only the Cuba and Haiti lists (and any applicable bundled lists) will receive that alert.

Final update emails

When a storm that previously had forecasted exposure in a country no longer does, one final email is sent for that country. It is marked (final) in the summary table. The figures in a final update reflect only observed exposure — they will not change unless the storm track reverses and approaches the country again, in which case a new alert would be issued.

A final update is only sent when the storm actually brushed the country — that is, when some population experienced observed exposure. For a near miss, where a country was only ever in the forecast and the storm never arrived, the alerts for that country simply stop without a final-update email.

Already passed notice

If a storm has already moved through a country in an earlier advisory but is still active elsewhere, that country appears in a small footnote at the top of the email labelled "Countries already passed." This is a reference note, not a separate email — it points you to earlier alerts for the final exposure figures for those countries.

Which subscription list is right for me?

List You receive an email when…
Storm Alerts – LAC (recommended) An alert includes at least one Caribbean, Central American, or South/North American country (excluding USA and Canada). The recommended default: nearly all humanitarian-relevant NHC activity falls in this region.
Storm Alerts – All Countries Any storm alert is issued, anywhere in the world.
Storm Alerts – [Country] An alert specifically includes that country.

You can subscribe to multiple lists. Listmonk deduplicates deliveries — being on several matching lists will not result in duplicate emails for the same advisory.

Understanding the email

Summary table

Every email opens with a summary table. Each row covers one wind speed threshold for one affected country:

ColumnWhat it shows
StormStorm name and season year.
CountryAffected country. Rows marked (final) are final-update advisories.
Similar stormsThe 3 historical storms (since 2002) with the most similar population-exposure profile across all wind thresholds.
WindWind speed threshold: 34 kt, 50 kt, or 64 kt (see below).
Exposure [% pop.]Estimated population within the wind radius — the maximum across available data sources (CHD, ADAM, GDACS) — with the percentage of that country's total population in parentheses.
Return periodHow historically rare this level of exposure is (see below).

Wind speed thresholds

ThresholdDescription
34 kt (~63 km/h, ~39 mph)Tropical storm force winds — the minimum monitoring threshold.
50 kt (~93 km/h, ~58 mph)Strong tropical cyclone force winds.
64 kt (~119 km/h, ~74 mph)Hurricane / typhoon force winds (Category 1 minimum).

Return period

The return period (RP) estimates how rare the current exposure level is compared to all storms since 2002. A 10-year RP means that roughly one storm per decade has caused this level of exposure in that country at that wind threshold. The formula is:

RP = (number of seasons since 2002 + 1) ÷ (number of past storms with equal or greater exposure + 1)

Return periods above 3 years are colour-coded in the summary table:

>3-year RP >5-year RP >10-year RP

Return periods use the CHD estimate only. Unlike the headline exposure figure (which takes the maximum across sources), the return period is always computed from the OCHA CHD estimate. CHD has the longest and most consistent historical record — every storm since 2002 is processed the same way — so it gives a like-for-like comparison across storms and countries. (Colour bands are keyed to the rounded displayed value, e.g. a 4.6-year RP shows as "5-year" and is coloured accordingly.)

Forecast landfall

When the forecast track's centre crosses a country, the storm section opens with a landfall callout: where (nearest admin-1 unit), when (US Eastern time, with lead time), and the storm's approximate intensity at that point. This is computed from the NHC forecast track densified the same way the wind swaths are built (30-minute monotone-cubic interpolation), so it sits on the same curved path as the map — a country brushed by the wind field but never crossed by the centre shows exposure but no landfall entry.

Map

Below the summary table, each storm has one map: the observed and forecast track, the observed and forecasted wind swaths (34 / 50 / 64 kt), and the population exposed shaded by admin-1 unit (darker = more people exposed; small island territories that would be invisible at map scale are shown as a colour-coded ring). The shading uses the same consolidated 34 kt figures as the attached spreadsheet.

Charts

Each affected country then gets a chart per wind threshold at which at least one source reports exposure (no chart when none does), comparing the three source estimates — CHD, ADAM and GDACS as labelled dots on a shared scale. The probabilistic distributions and the wind-speed-probability map live on the online full-detail layout, shown for a set of historical storms on the example alerts page. The charts read as follows; the x-axis is population exposed.

Comparing the estimates to history. The historical ticks (and the return period) use the CHD-only estimate, because CHD is the only source with a consistent record back to 2002. The summary table's headline, however, is the maximum across sources, so it is intentionally at least as high as any single estimate. For a like-for-like historical read, use the CHD dot and the return period.

Maps on the full-detail pages

Basemap tiles © CARTO, data © OpenStreetMap contributors. If the tile service is unreachable when an alert is generated, the maps fall back to a plain boundary background — the storm layers are unaffected.

Data sources

All exposure estimates are derived from NOAA NHC track and wind-speed-probability forecasts, processed independently by three organisations:

LabelSource
ADAM WFP ADAM — the World Food Programme's Anticipatory Action Data Management platform.
GDACS GDACS — the Global Disaster Alert and Coordination System (European Commission / UN).
CHD OCHA Centre for Humanitarian Data — derived from NHC track forecasts and the probabilistic WSP model.

How the estimates are combined and timed

For each country and wind threshold, the headline exposure figure is the maximum across whichever of the three sources have data for that advisory — a deliberately conservative choice. Each source's individual estimate is shown separately in the strip charts so you can see the spread, and the email lists which sources contributed (and notes any that were unavailable).

The sources publish on slightly different schedules. Each is paired to the NHC advisory by matching its issue time to the advisory time, or to the closest issuance up to three hours earlier — so a single email reflects one consistent advisory cycle even when a source lags slightly. The probabilistic WSP product, in particular, is published on a 6-hourly cycle and is matched this way to the advisory.

If a source is delayed beyond that window (more than three hours late for a given advisory), it is simply omitted from that email rather than shown with stale figures — the consolidated number degrades gracefully to whichever sources are available (in the limit, CHD only). The intro paragraph always lists which sources contributed and flags any that were not available.

Subnational (admin 1) figures in the attached spreadsheet

Each storm's attached spreadsheet (Excel workbook) includes subnational (admin 1) rows in addition to the country (admin 0) rows, distinguished by an admin_level column (with pcode and adm1_name identifying each unit). The three sources report subnational areas differently, so they are harmonized onto a common set of FieldMaps admin-1 units before being combined.

As at the country level, each admin-1 figure is the maximum across the available sources for that unit. The set of sources is held consistent across all admin-1 units within a given storm and country, so the units are directly comparable (rather than one area combining two sources and its neighbour only one). Taking the highest credible estimate per unit follows the convergence-of-evidence logic of the inter-agency Joint Intersectoral Analysis Framework (JIAF). Because the maximum is taken independently for each unit (and different sources may give the highest estimate in different places), the admin-1 figures will not necessarily sum to the country total.

Questions and feedback

These alerts are experimental. For questions or feedback, contact the OCHA Centre for Humanitarian Data at ocha-datascience@un.org.