A community sensor in Kerobokan logged four hundred and sixty-nine days of Bali's breath before it fell silent on 15 March — and on roughly two in three of them, the twenty-four-hour reading was above the line the World Health Organization now sets for a single day's exposure.
The only long-running public record of Bali's air is a single community instrument on the back of a house in Kerobokan. It is a GAIA A12 laser-scatter sensor, corrected by the AQICN network, and since November 2024 it has posted a daily median to a page almost no one reads. Across its 469 days of operation, the mean reading reached 21.9 µg/m³ of PM2.5 — more than four times the World Health Organization's annual guideline of 5 µg/m³.
On roughly two of every three days of that record, the daily median was above the 15 µg/m³ ceiling the WHO set for a single day's exposure. On the fifteenth of March 2026, without announcement, the sensor fell silent.
The WHO's 2021 global air-quality guidelines lowered the twenty-four-hour PM2.5 threshold from the 25 µg/m³ it had set in 2005 to 15. The tightening reflected a body of epidemiological evidence that had, in the intervening decade, continued to tie moderate exposure to cardiovascular and respiratory harm. Indonesia's national ambient standard, codified in Peraturan Pemerintah 22 of 2021, remained at 55 µg/m³. Under domestic law, most of Bali's bad-air days are still formally normal.
There are two difficulties with the Kerobokan record. The first is the instrument itself. A laser-scatter sensor — however carefully corrected against a reference-grade monitor — is at the absolute end an indicator rather than a regulatory finding. Its numbers carry a slack of perhaps a few micrograms per cubic metre in either direction.
The second difficulty is the harder one. No one has built a reference-grade record against which the community figure could be checked. The island has one such government instrument — in Denpasar — and through 2025 it reported sporadically, and then not at all. (See Dispatch II.)
Indonesia's population-weighted annual PM2.5 in 2024, according to IQAir's World Air Quality Report, was 35.5 µg/m³, ranking the country fifteenth in the world. Bali's annual mean sits lower than Jakarta's. It is not the worst air in the country. But the 5 µg/m³ guideline was written at that level for a reason — and on an island that sells its climate as the product, two-thirds of days above 15 is a number that deserves to be recorded. The Kerobokan sensor was, at the time of writing, the only one keeping count.
"The sensor did not log pollution events, in any dramatic sense. What it logged was a baseline."Dispatch I · field note
What does two in three days mean, lived? It means that a child born in Kerobokan on the day the sensor began breathing the Bali average has, by her first birthday, received roughly twice the WHO-recommended annual dose. It means that on any given week of her life, four of its days will meet the short-term ceiling an international body considers a line of public-health consequence — and three of those four will exceed it.
Record is the quiet thing that is absent. When the Denpasar government sensor reports, it publishes an hourly AQI and a twenty-four-hour PM2.5 reading; when it does not, there is nothing on the national ISPU dashboard for the island. And on the one private long-record available, the daily guideline has been routinely exceeded for a year and a half.
A laser-scatter sensor, again, is not a regulatory instrument. The Kerobokan data is an indicator. Its 68 % figure is a Dispatch analysis of AQICN's public daily feed, and the raw CSV sits alongside this page in the Appendix so that anyone with a spreadsheet can reproduce it. The sensor is not reference-grade. That is the whole problem. For the island of Bali, in April of 2026, this is the best long-running public record that exists.
The instrument that should have been standing beside it — the government station in Denpasar, designed and maintained for exactly this purpose — has not posted to the ISPU portal in months. That is the subject of the next dispatch.