← All Posts Caribbean Dust Season 2026 · AI Health Forecasting

The Dust Is Back Over Barbados. AI Could Turn 60 Years of Data Into a Health Warning System.

Adrian Dunkley, the AI Boss August 11, 2026 13 min read

Barbados woke up on August 11, 2026 to a familiar warning. The Barbados Meteorological Services forecast for the day, carried by the Caribbean Broadcasting Corporation, called for a plume of thick dust haze that would cut visibility across the island and the surrounding marine area, on top of partly cloudy skies and scattered afternoon showers. A day earlier, the Caribbean Institute for Meteorology and Hydrology's Dust and Air Quality Forecasting Centre in Bridgetown had already flagged the same plume in its own model run, tracking dust concentration, aerosol optical depth, and fine particulate levels across the Eastern Caribbean. Neither warning was a surprise. Saharan dust has crossed the Atlantic and settled over Caribbean islands every year for as long as anyone has measured it, and CIMH has measured it for sixty years running from a single monitoring station on Barbados's easternmost point.

What has changed is not whether the dust comes. It is how often, how early, and how directly it now reaches the region's lungs, water tanks, and hospitals, and whether the Caribbean's forecasting effort, still built mostly for meteorologists rather than for the parent deciding whether to send a child with asthma to school, can keep up. Barbados's Tuesday advisory is one data point in a pattern that has run through the whole of 2026. Jamaica went through a heavier and longer episode in May. The same plumes have touched the Bahamas, the Cayman Islands, Puerto Rico, the Windward Islands, and reached as far as Colombia's Caribbean archipelago. The region has excellent physics behind its dust forecasts and almost nothing connecting that physics to the person who needs to act on it. That gap is where AI belongs.

A Warning Barbados Has Learned to Read

Dr Andrea Sealy, Regional Chair of the WMO Sand and Dust Storm Warning Advisory and Assessment System and based at CIMH, has spent the past two years watching the season's shape change. "Coming into April and May, you're getting a lot of dust, whereas before you might have gotten the dust later in May, maybe going into June," she said in an interview earlier this year. The shift matters because it stretches the exposure window for outdoor workers, athletes in training, and anyone with a respiratory condition across more of the calendar than the historical pattern would suggest. Sealy also credited CIMH's own model with real predictive value. "Our model at CIMH was able to predict that dust episode and does a pretty decent job predicting others," she said.

That model has a genuine pedigree behind it. CIMH, formed in 1999 from the merger of the Caribbean Meteorological Institute and the Caribbean Operational Hydrological Institute, has hosted the Americas Regional Centre of the WMO Sand and Dust Storm Warning Advisory and Assessment System from Bridgetown since it was established in 2012. Its ground truth comes from Ragged Point, on the island's eastern tip, where a University of Miami monitoring station has recorded continuous atmospheric dust measurements for sixty years, the longest such record anywhere on Earth. Few regions in the world can check a forecast model against six decades of the same instrument at the same spot. The Caribbean can, and largely has not used that advantage the way it could.

What Happened in Jamaica in May Is the Preview

In late May 2026, Jamaica's National Environment and Planning Agency recorded PM2.5 readings at its Spanish Town Road station in the Six Miles area of 33 and 32 micrograms per cubic metre on May 27 and 28. Two weeks earlier, on May 14 and 15, the same station had recorded 36 micrograms per cubic metre, above the daily ambient air quality standard of 35. At the Cross Roads station, readings of 20 and 24 on the same dates were lower, but NEPA still described conditions there as having deteriorated "from good to moderate," adding that "persons who are sensitive to air pollution will begin to experience some effects."

The Ministry of Health and Wellness responded with an advisory covering asthma, chronic obstructive pulmonary disease, respiratory infections, allergies, and skin and eye irritation, and it flagged a risk most people would not think to associate with dust: contaminated rainwater. Households that harvest water in open catchment tanks were told to keep tanks covered and water properly treated. One Kingston general practitioner interviewed by the Gleaner reported that six of ten patients seen in a single day presented with sinus symptoms, a runny nose, post-nasal drip, and throat irritation. Health Minister Dr Christopher Tufton offered a telling qualifier at the time: there was "no information it has led to any hospitalisation." That statement is accurate and also revealing. Jamaica's system, like every other territory's in the region, finds out about a dust episode's health toll after clinics start filling up, not before.

The Core Argument

The Caribbean is not short on dust data. CIMH runs a credible five-day physics forecast, and Barbados holds the world's longest dust observation record to check it against. What the region lacks is the automated bridge from that forecast to the household with a child on an asthma inhaler, the school deciding whether to hold outdoor sport, and the clinic that could staff up a day ahead instead of finding out from a waiting room full of patients. That bridge is an AI problem, not a meteorology problem.

A Physics Model With No Bridge to the Public

CIMH's Dust and Air Quality Forecasting Centre publishes daily WRF-based forecasts of dust concentration, aerosol optical depth, and PM2.5 and PM10 levels across its member states, including Antigua and Barbuda, Barbados, Dominica, Jamaica, and Trinidad and Tobago. The forecast run dated August 10, 2026, the one that anticipated Barbados's Tuesday haze, is exactly the kind of product a national meteorological service needs. It is not, by the centre's own account, meant to reach the public directly. CIMH's site states plainly that its products are "provided to assist in the development of forecasts by trained meteorologists" and offers no warranty on their accuracy or reliability for other use.

That caveat exists for good reason; raw model output needs expert interpretation. But it also describes the exact point where the region's dust forecasting chain breaks down before it ever reaches a household. A forecast has to pass from CIMH to a national meteorological office, then to an environmental agency like NEPA, then to a health ministry, then to a press release, then to a news broadcast, before it reaches the person deciding whether their child with asthma should stay indoors. Each step adds delay and loses precision. Jamaica's May episode shows the result: a health ministry reacting to sinus complaints reported by a single physician rather than a forecast that told them, four days ahead, which parishes would cross the PM2.5 threshold and roughly how many additional respiratory visits to expect.

Where AI Actually Fits

None of this calls for replacing CIMH's physics model. WRF-based chemical transport modelling is the right foundation, and CIMH has spent over a decade building institutional expertise around it. AI's role is to sit on top of that foundation and close three specific gaps. The first is accuracy at the local scale. Machine learning bias correction, trained against the Ragged Point sixty-year record and Jamaica's own NEPA sensors, is standard practice at air quality agencies from the United States to China, where it routinely sharpens a regional chemical transport model's output down to neighbourhood level. Applied here, it would let CIMH's forecast distinguish Six Miles from Cross Roads with real confidence, rather than treating an entire parish as one number.

The second gap is coverage. Jamaica, the best-monitored territory in this account, has ground sensors at only a handful of sites. Satellite and ground-sensor fusion, blending NASA and NOAA aerosol retrievals with the sparse station data that does exist, can estimate PM2.5 in the wide spaces between monitors where most Caribbean people actually live. The third gap is translation. A model that links forecast dust concentration to expected respiratory clinic demand, built on the pattern Jamaica's May episode already demonstrated, gives a health ministry days of lead time to staff clinics and stock inhalers, instead of a minister explaining after the fact that no hospitalisation data exists yet.

What a Caribbean Dust and Health Alert System Would Do

Picture the pipeline running automatically rather than through a chain of press releases. CIMH's dust, AOD, PM2.5, and PM10 forecast feed updates each morning. An AI layer cross-references it against ground sensor readings and each territory's air quality standard, then flags the districts where forecast concentration is set to cross a health-relevant threshold in the next 24 to 72 hours. From there, the system routes itself. Patients registered with a health ministry as having asthma or COPD receive a personalised SMS or app alert timed to arrive before the plume, not after clinic waiting rooms fill up. Schools and sports federations, given the risk to athletes in training that Barbados Today has already flagged this year, get advance notice ahead of outdoor practice and exam days. Water utilities and rainwater-harvesting households get a prompt to check catchment covers, echoing the exact guidance Jamaica's Ministry of Health issued manually in May. Marine operators and airport authorities get the same visibility data CIMH already produces, delivered automatically instead of relayed by hand.

Every one of these connections already has a plausible owner. Health ministries maintain patient registries. Mobile carriers already run emergency alert channels. NEPA and its counterparts across the region already issue advisories. What is missing is the automated link between CIMH's forecast and each of those existing systems, and that link is a data integration and machine learning problem the region is well positioned to solve, not a research problem waiting on new science.

The Money the Region Is Not Counting

Reactive dust response has a cost that rarely shows up on a single line item. Clinic visits that could have been prevented with a few days' warning still get paid for, just later and less efficiently. A dust-linked marine advisory that grounds small craft on short notice costs fishing operators a working day they had no way to plan around. Outdoor tourism excursions cancelled at the last minute cost more in refunds and reputation than one planned around a four-day forecast would. None of these costs are large individually. Added up across a season that now runs from April into early June, and recurs several times within that window, they are not small either.

There is an opportunity sitting alongside that cost. Barbados already hosts the Americas Regional Centre of the WMO Sand and Dust Storm Warning Advisory and Assessment System, and Ragged Point holds the longest continuous dust record in the world. Very few dust-affected regions, including large parts of West Africa, the Gulf states, and South Asia, can claim either asset. An AI health-alerting layer built and proven on top of CIMH's existing infrastructure would not just serve the Caribbean. It would be a genuine export product, something Barbados and the wider region could license, advise on, or operate for other governments managing the same seasonal problem. A small island institution with sixty years of unmatched data has a stronger claim to global leadership in this specific field than almost anywhere else on Earth. Right now, that claim is sitting unused.

What Should Happen Before the Next Plume

Four concrete steps would move this from an argument to a working system. First, CARICOM and the Caribbean Development Bank should fund an AI bias-correction and downscaling layer on top of CIMH's existing WRF-Chem output, calibrated against the Ragged Point record, so the region's own sixty years of data finally gets used to sharpen its own forecast. Second, CIMH, NEPA, and equivalent agencies across member states should build an automated alert pipeline running from the daily forecast straight into national health ministries, mobile carriers, schools, and sports authorities, rather than leaving that translation to a press release cycle. Third, ground-level PM2.5 and PM10 sensor coverage needs to expand well beyond the handful of stations Jamaica currently runs, since even the best-monitored territory in this account has only sparse data to work with. Fourth, water utilities should be looped into the same alert pipeline given the rainwater contamination risk the Jamaican Ministry of Health has already had to flag manually.

None of this requires new science. CIMH's model already works well enough that its own regional chair credits it with predicting the episodes it has been asked to predict. What is missing is the layer that turns a forecast built for meteorologists into a warning a parent, a farmer, a fishing captain, or a school principal can act on without waiting for a ministry to catch up. Barbados's advisory on August 11 will not be the last one this season. CIMH's own five-day model already knows roughly when the next plume arrives. The only question is whether the region builds the bridge from that knowledge to the people who need it before or after the next round of sinus complaints fills up a Kingston waiting room.

Frequently Asked Questions

What caused the Barbados dust haze advisory on August 11, 2026?

Barbados Meteorological Services, reported through the Caribbean Broadcasting Corporation, forecast a plume of thick dust haze that would cut visibility across the island and the surrounding marine area on August 11, 2026, alongside partly cloudy skies and scattered afternoon showers. The dust originates as the Saharan Air Layer, mineral dust lifted from the Sahara and Sahel that crosses the Atlantic on the trade winds and settles over the Caribbean basin. A day earlier, CIMH's Dust and Air Quality Forecasting Centre in Bridgetown had already tracked the same plume in its own model run covering dust concentration, aerosol optical depth, and fine particulate levels across the Eastern Caribbean.

Is 2026 a worse Saharan dust year than usual for the Caribbean?

Not necessarily worse in total volume, but the timing has shifted. Dr Andrea Sealy, Regional Chair of the WMO Sand and Dust Storm Warning Advisory and Assessment System and based at CIMH in Barbados, has said that severe dust episodes now regularly arrive in April and May, where in previous years the heaviest dust would not reach the region until late May or into June. That earlier start, combined with pulses arriving roughly every five to seven days through the season, means Caribbean territories face more months of elevated dust exposure than the historical pattern suggests, even without a change in the total amount of dust crossing the Atlantic each year.

Which Caribbean and nearby countries have been affected by Saharan dust in 2026?

Dust plumes tracked in 2026 have reached Barbados, Jamaica, the Cayman Islands, the Bahamas, Puerto Rico, the Windward Islands, and Key West and South Florida in the United States. The same system has extended as far as Colombia's Caribbean archipelago of San Andrés, Providencia, and Santa Catalina, along with the Colombian departments of Magdalena, Bolívar, Sucre, and Córdoba. CIMH's forecasting centre in Bridgetown covers dust prediction for member states including Antigua and Barbuda, Barbados, Dominica, Jamaica, and Trinidad and Tobago.

What health risks does Saharan dust pose in the Caribbean?

Jamaica's Ministry of Health and Wellness has warned that Saharan dust raises the risk of asthma attacks, chronic obstructive pulmonary disease flare-ups, respiratory infections, allergies, and skin and eye irritation, and can also contaminate water stored in household catchment tanks. During a heavy episode in Kingston in May 2026, one general practitioner interviewed by the Jamaica Gleaner reported that six of ten patients seen in a single day presented with sinus symptoms, including a runny nose, post-nasal drip, and throat irritation. Health Minister Dr Christopher Tufton said at the time that there was no information linking the episode to hospitalisations, a statement that reflects reactive monitoring rather than a system built to anticipate the health impact before it lands.

What is CIMH's Dust and Air Quality Forecasting Centre?

The Dust and Air Quality Forecasting Centre, run by the Caribbean Institute for Meteorology and Hydrology from Bridgetown, Barbados, produces five-day forecasts of dust concentration, aerosol optical depth, and PM2.5 and PM10 particulate levels using the WRF weather modelling system. CIMH has hosted the Americas Regional Centre of the WMO Sand and Dust Storm Warning Advisory and Assessment System since it was established in 2012. The centre's own disclaimer states that its products are provided to assist trained meteorologists in building forecasts, and it makes no warranty about their accuracy or reliability for direct public use, which is the gap this article addresses.

Why does Ragged Point, Barbados matter to global dust science?

A University of Miami dust-monitoring station at Ragged Point, on the easternmost tip of Barbados, has recorded continuous atmospheric dust measurements for sixty years, making it the longest-running dust observation record anywhere in the world. That record allows CIMH's model to be checked against decades of real ground truth rather than a handful of recent seasons, and it is the calibration base any future AI layer built on top of the region's dust forecasting would need.

How could AI actually improve on the Caribbean's existing dust forecast?

AI would not replace CIMH's WRF-based physics model, which already produces credible five-day regional forecasts. Machine learning bias correction, trained against the Ragged Point sixty-year record and Jamaica's NEPA ground sensors, can sharpen the model's local accuracy island by island. Satellite and ground-sensor fusion can fill in hyperlocal PM2.5 estimates between the small number of stations currently reporting, since even Jamaica, the best-monitored territory in the region, runs sensors at only a handful of sites. A third layer can link forecast concentration directly to expected respiratory clinic demand, giving health systems days of lead time instead of finding out about a spike in sinus complaints after it has already happened.

What would a Caribbean AI dust and health alert system look like in practice?

It would pull CIMH's dust, PM2.5, and PM10 forecast feed automatically, cross-reference it against ground sensor readings and each territory's air quality standard, and trigger alerts the moment a district's forecast concentration crosses a health-relevant threshold. Registered patients with asthma or COPD would receive a personalised SMS or app alert through their national health ministry. Schools and sports federations would get advance notice ahead of outdoor training and exam days. Water utilities and rainwater-harvesting households would get a prompt to check catchment covers and treatment. Marine operators and airport authorities would get the same visibility data CIMH already produces, routed automatically rather than left for a meteorologist to relay manually.

What should someone with asthma or COPD do during a dust advisory right now?

Jamaica's Ministry of Health and Wellness advises staying indoors as much as possible during a dust episode, keeping windows closed and air conditioning set to recirculate, and wearing a well-fitted mask if going outdoors is unavoidable. Households that harvest rainwater should keep storage tanks covered and ensure water is properly treated before use, since dust deposition can affect stored water quality. Anyone with worsening respiratory symptoms, chest tightness, or breathing difficulty during a dust advisory should seek medical attention early rather than waiting for symptoms to become severe.

Could the Caribbean turn its dust forecasting capability into something it exports rather than only a problem it manages?

Yes, and the region already holds two assets that most dust-affected parts of the world do not. Barbados hosts the Americas Regional Centre of the WMO Sand and Dust Storm Warning Advisory and Assessment System, and Ragged Point holds the world's longest continuous dust observation record. An AI health-alerting layer built and proven on top of that foundation would be a genuine export product, one the Caribbean could license or advise on for other dust-exposed regions including the Sahel, the Gulf states, and South Asia, turning a recurring seasonal vulnerability into a capability the region sells rather than only suffers through every year.

Get the Weekly Post

New thinking on Caribbean AI, governance, and technology, delivered every week.