Correction, 8 October 2026
This study was published on 16 September 2026 under the headline "no weather station stands near Cumbuco". That headline was too broad and has been retracted and replaced. On 7 October 2026 we verified, from the operator's own public API, that SiMCosta station 307 at Pecém — a meteorological station operated by FURG — reports wind speed and direction at approximately one-minute cadence 12.9 km from Cumbuco, closer than the Fortaleza airport station this programme already uses.
The finding in this article is unchanged and still stands: it concerns INMET's national automatic-weather-station archive, in which 124 of 639 stations recorded no valid reading in the first eight months of 2026, 7 of Ceará's 14 were completely dark, and no INMET station falls within 60 km of Cumbuco. The title now says so explicitly.
The distinction matters, and it is the one we should have drawn at publication: the defensible claim is that no fit-for-purpose, independently verified nearshore observation has yet been established for the Cumbuco beach decision problem — not that instruments are absent. Station 307's anemometer height, exposure and siting are not established, no quality-control package covers the recent period, and SiMCosta states that the data it serves carry no quality-control filters. Its fitness for any particular decision therefore remains unestablished, and we are now measuring exactly that.
What the public archive contains
Brazil's national meteorological institute, INMET, publishes an hourly public archive of automatic weather-station records. The version covering the first eight months of 2026 lists 639 stations. This finding treats a hard fact plainly: for a given hour, a station either has a usable wind, pressure or temperature value, or it does not — the archive's own documentation marks the gap with the value 9999, the word Null, or a blank field, and this program counted which stations never once left that state.
124 of the 639 stations — one in five — recorded not a single valid hourly reading across the entire eight-month window. That is not a station with gaps. For accountability purposes, it is a station that does not exist for 2026.
Restricted to states with at least eight stations, so one dead unit cannot swing a percentage, six states have half or more of their network completely dark for the year:
- Amazonas: 11 of 19 stations completely dark (57.9%).
- Maranhão: 8 of 15 (53.3%).
- Ceará: 7 of 14 (50%).
- Rio Grande do Norte: 4 of 8 (50%).
- Santa Catarina: 12 of 24 (50%).
- Piauí: 9 of 21 (42.9%).
Zero INMET stations within reach of the kite coast
Cumbuco, one of Ceará's busiest kite and kitesurf coasts and the site of two of this program's case studies, sits inside none of the archive's station footprints. Ranking all 639 stations by distance from the beach, not one falls within 60 km.
The nearest, Guaramiranga, is 72.6 km away and itself only 28.9% complete for the period — a mountain-town station, not a coastal one. Of the next five closest, ranging from 96 km to 180 km away, three — Itapipoca, Acaraú and Sobral — recorded zero readings for the entire eight months, and a fourth, Quixadá, is 15.7% complete. Only the farthest of the six, Morada Nova at 169 km, is effectively whole, at 99.9%. The one station a forecaster could actually trust to check the coast against is built the least like the coast, and sits the farthest from it.
A sensor that never stops “reporting”
A dead station is the easy failure to catch — it simply stops sending numbers. A harder one turned up in this program's own review of station A219, Turiaçu, on the Maranhão coast: across 14–17 August 2026, its wind-direction sensor logged the exact value of 236° in 93 of 96 hourly readings, with the three exceptions at 241°, 245° and 253°. Natural wind does not sit that still. The station's wind-speed sensor, meanwhile, kept varying normally across the same 96 hours — averages of 2.9–3.1 m/s, gusts to 11.5 m/s — which is exactly what makes a stuck wind vane easy to miss: the station keeps “reporting”; only the direction is fiction.
This program flagged the fault to the institute's technical coordination team in early September 2026. The team confirmed in writing that the report was received and is under review — the kind of response this finding is written to make easier to ask for, not harder.
Why this is a finding, not a complaint
639 stations is a large, genuinely useful public network, and most of it works. Two different failure modes sit inside the 124 that do not: a station that has gone fully silent, easy to spot in a single query, and a station that keeps transmitting a wrong number, which this review only caught by checking whether a value that should vary ever actually did. A network this size will always have some of both; the value of counting them in public is that the count can be checked, station by station, by anyone who downloads the same archive.
The finding that matters for the coast is the simplest one, and none of the above changes it: not one INMET station stands within 60 km of Cumbuco. The question underneath is the one this programme also asks about the last two kilometres before the beach: whether a fit-for-purpose, independently verified observation has been established for a beach decision. Along this stretch of the Ceará coast the national network does not provide one, and the nearest live station outside it, 12.9 km from Cumbuco, has not yet been shown fit for that decision (see the correction above). Closing the gap does not wait on a network's budget cycle — it is the same small, tracked releases this program has already proposed to measure how kite equipment drifts, recovered the same session, logged like a station reading.
This is research. Estimates are models, not guarantees; a model is never an observation, and a machine estimate is never “confirmed” — only a human reviewing it in the field makes it verified. Nothing on this page is an operational-capability claim, a safety promise, or a prediction of a specific event at a specific place.
References & data
- 1.INMET — Instituto Nacional de Meteorologia, historical data archive — Public hourly records for Brazil's 639 automatic stations, January–August 2026 (the archive's own 2026 historical bundle), plus the public station catalog used to place each station and compute its distance from Cumbuco. The completeness and distance figures in this article are AYON Ecosystem's own count against that archive, not a statistic INMET itself publishes.
- 2.INMET — station A219 (Turiaçu) sensor fault, correspondence on file — AYON Ecosystem reported the wind-direction anomaly to INMET's External Data Conformity Coordination (COCDP/CGDM) on 2 September 2026; the coordination confirmed in writing on 9 September 2026 that the report was received and is under technical review. No station operator's name is published; the correspondence is on file.
- 3.AYON Ecosystem research archive — National completeness audit (per-station valid/invalid hour counts, January–August 2026) and the Cumbuco-distance ranking, both computed from the INMET archive above; full station-level results on file.