Long readsThe smoke

From late February the hills burn — crop residue, forest undergrowth, land clearance, on both sides of the border — and the smoke settles into exactly the valleys this road runs through. It lifts with the first rains. This page carries four years of daily figures for every town on the circuit, and says plainly where those figures are weaker than they look.

Two sources were fetched, and they are printed together because they disagree.

The long record is Copernicus CAMS through the Open-Meteo air quality service: hourly PM2.5 at ten points around the loop from 1 August 2022 to the day of the build. That is 15,050 point-days. Reduced to a daily mean and then to a month, the shape is not subtle. Take Pai: a July mean of 3.0 micrograms per cubic metre, a December mean of 9.9, a March mean of 26.1, an April mean of 29.8. Thirty of the thirty April days measured sat above 35.4 µg/m³, the figure the United States uses as its 24-hour standard. In July, across four years, not one day did.

The same shape repeats at every point on the circuit, and one thing about it is worth saying out loud: Chiang Mai is worse than the loop. Chiang Mai's March mean is 37.8 µg/m³ against Mae Hong Son's 29.1 and Pai's 26.1. The city you leave from is the dirtiest air on the route.

Where the long record fails. In 15,050 point-days the model never once produced a daily mean above 99.4 µg/m³. Ground stations in northern Thailand have recorded far higher. A model cell is tens of kilometres across; a Mae Hong Son valley on a windless March morning is not, and the averaging that makes a continental model tractable is exactly the averaging that removes the worst hour of the worst day. Inference — read the monthly shape from this data and do not read the ceiling from it.

The ground record, such as it is. Thailand's Pollution Control Department runs a national network and publishes the current hour. Its historical endpoint returned 404 on the fetch date, so what this site can show from it is now, not then. And there is a second problem, which is a finding in itself: Mae Hong Son province — 12,765 square kilometres, the largest by area in the north — has one station. It is 58t, in Mae Hong Son town. Pai has none. Soppong has none. Khun Yuam, Mae La Noi and Mae Sariang have none. Nobody official is measuring the air in the town most riders sleep in on night one.

The practical version. February is not the problem up-country: Pai's February mean is 15.0 µg/m³ and no February day in four years cleared 35.4 there. March and April are. May is over — Pai drops to 10.0 and stays down. If the dates are fixed and they land in March, the ride is still rideable; what changes is that a viewpoint is a white wall, a long descent is done with stinging eyes, and anyone with asthma is having a bad week. A fitted FFP2 or N95 under the helmet is the mitigation people use; a surgical mask and a buff are not filtering PM2.5.

Inference — the scheduling advice this data supports reverses the usual one. The cool season is sold as the season, and November to February is genuinely lovely. But the cleanest air on this circuit by a wide margin is the rains — Pai at 3.0 µg/m³ in July against 26.1 in March — and the hills are green, the waterfalls are running and the rooms are cheap. What you trade is dry tarmac.