Dust is the weather here
Across most of this map, more flights are cancelled, more roads closed, more schools shut and more hospital admissions caused by airborne dust than by rain. It is the region's dominant hazard, its dominant air-quality problem, and the single largest loss factor for its enormous and growing solar fleet. It is also the field that no free global weather model carries at all.
Why the global models are blind to it
ECMWF's IFS, DWD's ICON and NCEP's GFS — the three models behind almost every forecast you will read for this region — are dry-aerosol-free. They model the atmosphere; they do not model what is suspended in it. A Shamal that reduces Kuwait to 300 metres of visibility appears in these models as a windy day.
That leaves three ways to see dust, and this site uses all of them.
1 · Visibility, as a proxy
NCEP's GFS is the only one of the four models here that publishes a surface visibility field, and it is the reason GFS earns a place on this site despite being the least skilful of the four. Its visibility is derived from the model's own hydrometeors and boundary-layer state rather than from dust, so it will find fog and it will find rain — but over the Iraqi and Kuwaiti desert with a 60 km/h northwesterly, a visibility collapse is a dust signal, and it is the earliest one a free model gives you.
2 · Meteosat, from directly overhead
EUMETSAT's Indian Ocean satellite sits at 45.5°E — almost exactly over Riyadh. Its dust RGB composite, built from the 8.7, 10.8 and 12.0 µm infrared channels, renders airborne dust as a distinctive magenta against a blue-green background, day and night, every fifteen minutes. It is the best real-time picture of a dust outbreak that exists for this region and it is free through EUMETView's WMS. No other part of the world has its dedicated geostationary satellite parked this conveniently over its dust source.
3 · Dedicated dust models
The WMO's Barcelona Dust Regional Centre runs the Sand and Dust Storm Warning Advisory node for North Africa, the Middle East and Europe, and publishes a ten-model ensemble plus a multi-model median on open THREDDS over exactly this domain — MONARCH, ICON-ART, SILAM, NASA GEOS, MetOffice-UM, MOCAGE, LOTOS-EUROS, WRF-Chem, NOA and EMA-RegCM4. There is no comparable free product for any other region on earth. Copernicus CAMS adds a global dust optical depth and PM10 forecast to four days.
Reading the visibility map
Where the dust comes from
Three sources dominate this frame, and they behave differently enough that knowing which one is active tells you what to expect.
The Tigris–Euphrates floodplain and the Syrian desert. The most productive dust source in the region and the origin of the classic Shamal. Decades of drainage of the Mesopotamian marshes, upstream damming and drought have expanded the erodible surface enormously. A northwesterly behind a cold front picks dust up over Syria and Iraq and carries it the entire length of the Gulf — Basra, Kuwait, Dammam, Bahrain, Doha, then the UAE and the Strait of Hormuz, usually over 24 to 48 hours. This is the event that closes Gulf airports.
The Rub' al Khali and the Nejd. Local emission across the Arabian interior, usually shallower and more diurnal than the Iraqi source — it builds through the afternoon as the boundary layer deepens and the low-level jet mixes down, and eases overnight.
Convective outflows — haboobs. A thunderstorm's cold pool spreading across dry desert lifts a wall of dust hundreds of metres high that arrives with almost no warning and drops visibility to nothing in minutes. Sudan and Iraq get the most dramatic ones, but they occur across the whole domain, and they are also the phenomenon that a parameterised-convection model such as any on this site fundamentally cannot forecast. That is the strongest argument for running a convection-permitting model over this region.
Why we don't run our own dust model
Adding online mineral dust to a mesoscale run — WRF-Chem with GOCART, or ICON-ART — multiplies its compute cost by roughly two to four. It is affordable. It is also, for this region specifically, probably the wrong thing to spend on: the Barcelona centre already publishes ten dust models and their median over precisely this domain, for free, and an eleventh of the same kind adds very little. The compute is better spent on resolving the convection that generates haboobs in the first place, which nobody publishes freely for this region at all.