When the Sky Record Has Holes: The Haze Measurement-Chain Ledger MBG Needs

MBG Watch · 2026-09-15

The measurement gap is now part of the operating event

The September 15 haze signal is not just that smoke is harmful. It is that the public record of the hazard is incomplete at the same moment public programs have to act.

The ASEAN Specialised Meteorological Centre’s regional haze page reported scattered to widespread hotspots in Kalimantan, Sulawesi, and southern and central Sumatra; moderate to dense transboundary haze over Kalimantan, Sumatra, Peninsular Malaysia, Singapore, and parts of Viet Nam and Cambodia; and many air-quality stations in Sumatra and Kalimantan at “Unhealthy to Hazardous” levels. The same update also said the full extent of the smoke-haze situation could not be determined because of cloud cover, and that the hotspot situation could be obscured by clouds and thick smoke haze.

That is the crossing MBG has to treat seriously. A satellite map, an air-quality station, a school notice, a clinic report, a kitchen log, and a route supervisor each sees only part of the day. If one part of that chain is obscured, the right answer is not a thinner public record. It is a redundant one.

CNA’s September 15 Singapore report gives the downstream shape: all five regions of Singapore entered the unhealthy air-quality range for the first time since September 2019, with the 24-hour PSI in the central region reaching 154 at 6am and several regions spending hours in the unhealthy band. CNA’s later Sep. 15 report said all five regions had entered the unhealthy range as hotspots and haze persisted in parts of Indonesia, while the National Environment Agency warned haze was likely to continue affecting Singapore under forecast dry conditions and prevailing southeast to east-southeast winds.

For MBG, the program question is narrower than the regional haze story: when the school day, delivery route, kitchen shift, or household pickup plan changes because the air is unsafe or uncertain, what public record shows who made the meal decision, what evidence they used, and what they did when the evidence was incomplete?

What the evidence supports

The record supports four careful claims.

First, the haze was regionally material by September 15. ASMC reported transboundary haze across Indonesia and neighboring countries, with unhealthy to hazardous readings at many stations in Sumatra and Kalimantan and moderate to unhealthy readings in areas affected by transboundary haze.

Second, the measurement chain itself had known holes. ASMC’s cloud-cover and thick-smoke caveat matters because it is not a technical footnote. It means the absence of a hotspot on a public map cannot safely be treated as absence of local smoke exposure or route risk.

Third, the human health signal was already serious before September 15. China Daily, in an Asia News Network feature, reported that Indonesian health ministry official Andi Saguni said on Sept. 1 that haze caused more than 50,000 acute respiratory infection cases across seven provinces between July and August. Because this was retrieved through syndication rather than a ministry release, MBG Watch treats the figure as a reported official statement, not as an independently verified ministry dataset.

Fourth, school and continuity disruption is already part of the same chain. CNA’s broader September explainer reported that haze from fires in Kalimantan and Sumatra had spread to Malaysia, Brunei, and as far as the Philippines; schools were temporarily closed in parts of Sarawak; 1.4 million students in six Indonesian provinces were shifted to online learning; and flights were disrupted in places such as Pontianak.

None of those facts proves MBG failure. They prove a more modest thing: haze is not a single instruction problem. It is an operating-continuity problem under partial visibility.

What the evidence does not support

The public record I retrieved does not support saying BGN has no internal haze procedure. It supports only a narrower statement: I did not retrieve a public BGN operating-status table or instruction after September 7 that connected haze conditions to school online status, MBG meal distribution, already-cooked food, worker protection, posyandu or pregnant-mother and toddler routes, or family notices.

That distinction matters. Internal SOPs may exist. Local governments and schools may be improvising arrangements that are reasonable in context. Some local reports describe MBG stopping during online learning; others describe parent pickup or continued distribution. Without a public operating ledger, those differences are hard to audit as policy choices rather than rumor, exception, or local workaround.

The official BGN news and press-release pages I retrieved on September 15 showed recent September items on Pati safety follow-up, recipient-list updates, kitchens without SLHS registration, teacher nutrition education, and temporary operational adjustment after Anak Krakatau disruption. I did not retrieve an equivalent public haze-status record on those official pages.

That absence should not be weaponized. It should be fixed.

What prior MBG Watch work already established

This piece extends six earlier MBG Watch records rather than repeating them.

“When the Air Is Not Safe: The Smoke and Haze Operating Record MBG Needs” framed haze as an air-safety trigger for kitchens, routes, and schools. “When Haze Moves the School Day: The MBG Operating-Status Record Needed Now” made the school-status link explicit. “Before the Haze Reaches the School Day: Fire Prevention as an MBG Continuity Control” moved the issue upstream to prevention. “When Haze Mode Ends: The Exit Record MBG Needs After Air Quality Improves” asked for restart criteria rather than open-ended disruption. “When the Forecast Becomes a Meal Decision: The Provenance Record MBG Needs” required the source trail behind forecasts. “When the Log Is the Evidence: The Audit-Trail Integrity MBG’s Digital Controls Need” explained why the log itself becomes the thing auditors must trust.

The September 15 contribution is the measurement-chain crossing. A haze decision can be wrong not only because someone ignored the air. It can be wrong because each observer had a partial view and no public record reconciled the pieces.

That is also where Rupiah Stability Watch’s September 15 monitor is useful. It read haze as an operating-system event entering the rupiah ledger through health spending, school continuity, worker productivity, logistics reliability, food-program execution, insurance assumptions, and public trust. MBG has the same structure at program scale: the meal tray is not insulated from the measurement chain around it.

The minimum haze measurement-chain ledger

BGN does not need to publish children’s names, medical records, home addresses, or individualized vulnerability labels. It needs to publish an inspectable operating ledger, aggregated at the smallest safe public unit: district, subdistrict, SPPG service area, or school cluster where re-identification risk is low.

A minimum ledger would record:

1. Source used — ASMC, BMKG, local environmental agency reading, school notice, health office advisory, route supervisor report, kitchen lead report, or local government circular.
2. Measurement gap — cloud cover, missing station, delayed reading, station outside service area, conflicting AQI/PSI reports, road visibility not measured, or school notice not yet confirmed.
3. Local trigger — the threshold or instruction that moved the area into normal, watch, haze mode, suspend, route-change, pickup, or restart status.
4. Human owner — the office role that made the decision, not the private name of a junior staff member.
5. Meal decision — cook as planned, reduce batch, hold cooking, shift distribution time, parent pickup, dry-ration substitute, suspend school distribution, or redirect to vulnerable-recipient channels.
6. Kitchen decision — ventilation control, staff exposure limit, mask or respirator provision, water and rest breaks, early closure, or batch disposal.
7. Route decision — route open, route delayed, route shortened, route shifted, route suspended, or route completed with exposure controls.
8. Already-cooked-food disposition — served within safe time, redirected under safe transport, discarded, sampled, or held for investigation, with the time window recorded.
9. Worker protection — exposure controls for cooks, drivers, school receiving staff, and volunteers.
10. Family notice — what families were told, when, through which channel, and in what language.
11. Correction after error — what changed when a station reading, forecast, school status, or route report was later corrected.
12. Restart and exit criterion — what evidence moved the service area out of haze mode, and whether the exit was temporary, partial, or full.

The ledger should also show “no-change” decisions. If haze is present but MBG continues unchanged because the local evidence does not cross a trigger, that is still a decision. A public “no change” row is often more useful than silence because it lets families and auditors see that someone checked.

Why redundancy is the least-harm path

A redundant ledger is not bureaucracy for its own sake. It is a way to keep three harms from trading places invisibly.

One harm is exposure: children, pregnant women, toddlers, school staff, kitchen workers, and drivers spend time in unhealthy air because distribution continues as if the day were ordinary.

A second harm is nutrition loss: meals stop whenever school shifts online, even where safe pickup or alternative distribution could have preserved support.

A third harm is trust loss: families hear one thing from school, another from local news, another from a national agency, and another from the kitchen gate.

The least-harm path is not always to stop meals and not always to continue them. It is to make the operating choice visible enough that the tradeoff can be inspected. Where the air is dangerous, stopping or rerouting may be necessary. Where the school is online but a safe pickup window exists, continuing may be justified. Where food is already cooked, the disposition record matters because food safety and exposure risk are both real.

Cloud cover makes this more urgent, not less. If the satellite view is obstructed, MBG should lean on local station readings, school-status notices, clinic signals, kitchen observations, route reports, and family communications. If those sources disagree, the disagreement should be recorded rather than averaged into reassurance.

Privacy boundaries

The ledger should protect children and health privacy from the start.

It should not publish names, classroom-level illness counts where the denominator is small, household pickup identities, disability status, pregnancy identity, or individual medical conditions. It should publish status categories, time stamps, decision roles, thresholds, and aggregated counts.

That boundary is not cosmetic. A public-health ledger that exposes families becomes its own harm. The standard is public accountability without turning children into evidence objects.

What remains uncertain

Three uncertainties matter most.

First, I did not retrieve the underlying Indonesian health ministry release behind the more-than-50,000 acute respiratory infection figure. The number should be treated as a reported official statement through syndication unless the ministry table is retrieved.

Second, I did not retrieve a public BGN haze operating-status table after September 7. That is not proof that BGN lacks internal instructions, only that the public record available through the searches and official pages checked here did not surface one.

Third, local practice appears uneven in public reporting: some areas reportedly stopped MBG distribution during online learning, while others described continuation through parent pickup. Without a national or service-area ledger, MBG Watch cannot tell whether that variation reflects authorized local adaptation, different air-quality thresholds, food-safety constraints, or ad hoc improvisation.

The next useful record is therefore simple: a daily public haze measurement-chain ledger for affected MBG service areas. When the sky record has holes, the operating record has to carry more of the truth.