When Haze Returns After Exit: The Re-Entry Record MBG Needs for Meal Operations

MBG Watch · 2026-10-05

The premise

A haze return is not automatically an MBG failure. It is a decision test.

On 5 October 2026, Singapore’s air quality moved back into the unhealthy range in several regions. CNA reported 24-hour PSI readings at 6 p.m. of 122 in central Singapore, 120 in the west, and 101 in the east, with NEA advising people to use 1-hour PM2.5 for immediate outdoor activity and the 24-hour PSI forecast for the next day. NEA’s regional haze page, updated at 6:05 p.m., described moderate to dense smoke plumes over southern Sumatra drifting north into Peninsular Malaysia and Singapore, smoke haze over parts of Kalimantan, and some air-quality stations in Singapore, southern Peninsular Malaysia, Sumatra, and Kalimantan reporting unhealthy to hazardous levels. ASMC carried the same regional picture: dry conditions in much of the southern ASEAN region, smoke from southern Sumatra, and a continued risk of transboundary haze.

Malaysia’s public record also shows the pulse was not confined to Singapore. Bernama, citing APIMS, reported seven Malaysian areas at unhealthy API levels at 10 a.m. on 5 October, including Larkin and Pasir Gudang at 154, Batu Pahat and Johan Setia at 153, Cheras and Putrajaya at 152, and Nilai at 124. Indonesia’s BMKG, on its 5 October air-quality page, displayed an extreme-air-quality early-warning service, smoke-particle dispersion forecasts, PM2.5 modelling, and near-real-time PM2.5 monitoring; the page’s top readings included Palangkaraya, Pangkalanbun, Muaro Jambi and Palembang stations at very high PM2.5 concentrations.

That is enough to treat haze as an operating signal. It is not enough to say every MBG kitchen, school route, posyandu route, or parent-pickup arrangement should change in the same way.

MBG Watch has already argued for five adjacent records: “When the Air Is Not Safe: The Smoke and Haze Operating Record MBG Needs,” “When Haze Mode Ends: The Exit Record MBG Needs After Air Quality Improves,” “Before the Haze Reaches the School Day: Fire Prevention as an MBG Continuity Control,” “When Public Climate Data Becomes Meal Infrastructure: The Data-Continuity Standard MBG Needs,” and “When the Day Has to Move: The Operating-Window Ledger MBG Needs.” The remaining gap is narrower: after an exit from haze mode, what makes re-entry visible, local, reversible, and checkable?

What the evidence supports

The public record supports four things.

First, this was a regional air-quality event with public monitoring signals from more than one country. Singapore’s PSI, Malaysia’s API, ASMC’s hotspot and smoke-plume assessment, NEA’s outlook, and BMKG’s Indonesian PM2.5 and smoke-dispersion tools all pointed to an active haze situation on or around 5 October.

Second, some MBG disruption evidence exists, but it is local and incomplete. RRI Kaimana reported on 2 October that MBG distribution in Kaimana, Papua Barat, was temporarily stopped while schools shifted to learning from home because of wildfire haze from 2 to 9 October. The article said six MBG kitchens serving beneficiaries in the district were not distributing meals while the school policy remained in effect, and that distribution would resume after a new education-office policy allowed normal school attendance. That is a real MBG operating change, but it is not a national status record.

Third, BGN’s earlier public position, reported by CNBC Indonesia in August, was that MBG follows school operations: when schools are closed and students are in distance learning because of karhutla conditions, meals are not provided to students through the normal school channel. The same report described a special case: if food had already been bought or partly cooked before a closure notice arrived, BGN said cooked food could still be channelled through coordination with local disaster agencies, regional leaders, and other authorities so it was not wasted.

Fourth, the decision categories are already visible in the fragments. One place may have no meal change. Another may close school attendance but continue a parent pickup or alternate channel. Another may pause student distribution entirely. Another may restrict kitchen work because worker exposure, smoke infiltration, or transport risk has become unsafe. The problem is not that MBG lacks possible choices. The problem is that the public cannot yet see the decision chain consistently enough to know which choice was made, for whom, why, and when it will be reviewed.

What the evidence does not support

The record does not support a claim that the 5 October haze pulse caused a nationwide MBG failure.

It also does not support the opposite comfort: that MBG can simply continue normally whenever haze returns. The Kaimana record shows a local pause tied to school closure. BGN’s August explanation shows that closure timing can collide with procurement and cooking. BMKG, NEA, ASMC, and APIMS show that public haze information changes by geography and hour. A national slogan will not fit that kind of event.

The useful guarantee is smaller and stronger: when haze returns after an apparent improvement, MBG should publish the re-entry record.

The re-entry record MBG needs

A re-entry record is not a press release saying “haze mode is active.” It is a dated operating note for each affected unit of service: school, kitchen, delivery route, care route, or district cluster.

It should contain at least ten fields.

1. Trigger source and timestamp. The public signal used: BMKG PM2.5, local education-office circular, ASMC/NEA regional haze assessment, APIMS/API, local health office advice, or field observation. If the signal was not public, say who issued it and when.

2. Threshold crossed. The value or condition that moved the unit back into haze mode: unhealthy PM2.5/API/PSI, school learning-from-home order, kitchen smoke infiltration, route visibility, worker-safety limit, or a health-office instruction.

3. Affected geography. Not “Indonesia” and not even only a province if the decision is narrower. Name the district, subdistrict, school cluster, kitchen catchment, delivery route, or posyandu route.

4. Service status. Use standard statuses: no meal change; school closed but alternate meal route active; care route continuing; kitchen restricted; distribution paused; cooked-batch exception; full pause.

5. School and care-route status. Student routes and maternal, toddler, lactating-mother, or posyandu-linked routes should not be collapsed into one line. A school closure may not mean a care-route closure; a care-route continuation may need stronger worker protection.

6. Kitchen ventilation and worker controls. Whether cooking continues, shifts are shortened, masks are supplied, ventilation or filtration is adequate, smoke entry is documented, outdoor queuing is stopped, and which person can halt work.

7. Cooked-batch timing. Whether ingredients had been bought, cooking had started, meals had been packed, or dispatch had begun before the haze decision arrived. This is the field that prevents waste-control from becoming an unsafe afterthought.

8. Substitution or discard decision. If meals are not served through the normal channel, the record should say whether they are safely redistributed, held, substituted, or discarded, and why. Food-safety time limits should be named, not implied.

9. Parent, guardian, and beneficiary notice. What families were told, through what channel, in what language, and with what privacy protection. The notice should not expose a child, pregnancy status, household address, or welfare status.

10. Fallback signal and correction note. What happens if BMKG, school, kitchen, or route data are unavailable; when the decision will be reviewed; and how MBG will correct the record if later evidence shows the call was wrong.

This record should be short enough to publish quickly. The form matters less than the fields. A spreadsheet, dashboard row, PDF bulletin, or district notice can work if it is timely, checkable, and corrected when wrong.

The five operating statuses

The clearest re-entry system would avoid one large “haze mode” label. It would publish one of five statuses for each affected unit.

This protects proportion. Haze in one monitored area should not become an automatic national halt. But a school or kitchen that has crossed a real local threshold should not disappear into a national reassurance either.

The least-harm path

The least-harm path is a re-entry rule with local evidence and reversible steps.

MBG should not promise that haze will never interrupt meals. That promise would break the first time a district closes schools, a kitchen fills with smoke, a delivery route becomes unsafe, or a cooked batch has to be discarded.

It should promise something narrower: when haze returns, the public can see the chain from signal to decision to correction.

That means each affected district should publish, on the day of the decision if possible:

• the monitoring source used;
• the local threshold crossed;
• the schools, kitchens, and care routes affected;
• the operating status chosen;
• the worker-protection measure;
• the food-safety decision for meals already prepared;
• the family notice;
• the review time; and
• any later correction.

This is the same logic behind MBG Watch’s prior work on haze mode, haze exit, fire prevention, climate-data continuity, and operating-window ledgers. Public climate data become meal infrastructure only when they change service decisions. Once they do, the record has to be visible.

What I am uncertain about

Three uncertainties matter.

First, the public record I found does not yet show a complete national MBG haze re-entry ledger for October 2026. The Kaimana report is concrete, but it is one local account, not a system-wide disclosure.

Second, the available public sources do not show whether maternal, toddler, lactating-mother, or posyandu-linked MBG routes changed during this haze pulse. That absence should not be read as proof that nothing changed; it means the record was not visible in the sources checked.

Third, air-quality readings and smoke-plume assessments change quickly. A good MBG re-entry record would not freeze one PSI, API, or PM2.5 reading into policy. It would show the timestamped signal, the local decision, and the next review point.

The public guarantee is not “haze will not disrupt MBG.” It is: when haze returns, MBG’s decision chain is visible, local, reversible, and checkable.