Southeast Asian haze is recurrent regional air pollution produced principally by vegetation and peat fires, with smoke transported across national borders. Singapore is periodically affected because it lies near major fire-prone landscapes in Sumatra and Borneo and within seasonal atmospheric circulation systems that can carry their emissions toward the island. Although the wider phenomenon of Southeast Asian haze also affects Malaysia, Indonesia, Brunei, Thailand, and other countries, its consequences for Singapore are especially visible in public air-quality reporting, disruption of outdoor activities, and regional environmental diplomacy.
Severe episodes arise from an interaction between human land use and environmental conditions. Land clearing, peat drainage, and ignition provide the sources; dry weather makes vegetation and organic soils more flammable; and winds determine where smoke travels. El Niño can intensify regional dryness, but neither El Niño nor unusually high temperatures alone explain a particular haze event. Fire location, rainfall distribution, peat condition, and atmospheric transport are all important.
A haze season lasting several months does not necessarily mean continuously unhealthy air throughout that period. Individual smoke incursions may last hours or days, separated by cleaner intervals, while fires continue elsewhere. Its eventual duration depends on whether fires are extinguished or persist, whether winds continue to connect the source regions to Singapore, and whether sustained rainfall restores wet conditions.
Fire sources, land clearing, and peat combustion
The most consequential sources for Singapore have generally been fires in the Indonesian islands of Sumatra and Borneo, although their relative importance varies between episodes. Fires in southern and central Sumatra are particularly relevant because winds can transport their emissions across the Strait of Malacca and surrounding seas toward Singapore. Local fires and other combustion sources also affect air quality, but severe, widespread haze episodes usually require a substantial regional smoke contribution.
The expression slash-and-burn, sometimes called burning off or burning for land clearance, covers several different practices. Fire can remove vegetation and agricultural residues relatively cheaply, helping prepare land for planting. It occurs in smallholder farming as well as in landscapes associated with commercial agriculture and plantation development. Oil-palm and pulpwood production are relevant land-use contexts, but smoke cannot automatically be attributed to a particular commodity, landowner, or company merely because a fire occurs nearby. Ignition, land control, and responsibility require separate evidence.
The distinction between a vegetation fire and a peatland fire is central. Peat is accumulated, partly decomposed organic material formed under persistently wet conditions. In tropical peat-swamp forests, a high groundwater level limits the availability of oxygen below the surface and helps preserve this material. Drainage canals constructed for agriculture, forestry, or settlement lower the water table and expose peat to drying and oxidation. Logging and other forms of forest degradation can further alter the vegetation and local moisture balance.
Once sufficiently dry peat ignites, combustion can continue underground as smouldering combustion, rather than as a conspicuous wall of flame. The fire spreads slowly through porous organic soil, sometimes beyond the visible surface burn. Buried hotspots may survive light rain or incomplete firefighting and later ignite vegetation again. This persistence helps explain why smoke production can continue long after an initial land-clearing fire.
Peat combustion produces large quantities of smoke because much of it proceeds under conditions of incomplete combustion. Emissions include fine particles, carbon monoxide, organic gases, and carbon dioxide. An area of burning peat can therefore have a very different pollution impact from an equal area of rapidly burning grassland. Fire counts or burned-area measurements alone do not fully describe the resulting exposure.
The underlying vulnerability is consequently not simply the presence of tropical vegetation. It is the combination of ignition with a modified, combustible landscape. An intact, waterlogged peat swamp and a drained plantation on peat may experience the same regional dry spell but have markedly different probabilities of sustained burning.
Atmospheric transport, monsoons, and El Niño
Smoke becomes a Singapore haze episode only when emissions intersect with a suitable atmospheric transport pathway. Prevailing winds, local circulations, atmospheric stability, and rainfall determine whether smoke disperses, remains near its source, or reaches densely populated areas. A severe fire season in Indonesia does not necessarily produce equally severe pollution in Singapore, and a comparatively localized cluster of fires can produce a sharp Singapore episode if the winds are unfavorable.
Singapore’s regional wind regime is shaped by the monsoon cycle. The southwest monsoon generally occurs from June to September, when surface winds around Singapore are often southerly or southeasterly, although actual directions vary. The season overlaps with relatively dry conditions in important fire-prone parts of southern ASEAN. The northeast monsoon generally extends from December to early March and often brings wetter conditions to Singapore and the surrounding region. Inter-monsoon periods commonly have lighter, more variable winds and substantial thunderstorm activity.
El Niño is the warm phase of a coupled tropical Pacific ocean–atmosphere fluctuation. Changes in the Walker circulation can reduce rainfall over parts of the Maritime Continent, particularly during seasons when those areas are already relatively dry. Longer rain-free periods lower fuel moisture and groundwater levels, extending the period during which fires can spread. The strong El Niño events of 1997–1998 and 2015–2016 coincided with major regional fire disasters.
Nevertheless, El Niño is a risk factor rather than an ignition mechanism or a deterministic haze forecast. Its rainfall effects vary by location, season, and event. The severe Singapore episode of June 2013 did not require a strong El Niño. Conversely, an El Niño year need not reproduce the same Singapore pollution levels as an earlier event if fire prevention, source locations, rainfall, or winds differ.
The Indian Ocean Dipole can also influence regional dryness. During its positive phase, relatively cool waters near Sumatra and Java are associated with reduced convection and rainfall over parts of Indonesia. Shorter-lived atmospheric disturbances, including the Madden–Julian Oscillation, can temporarily enhance or suppress rainfall and alter winds, creating substantial variability within a single season.
Once emitted, smoke evolves chemically and physically. Some gases form additional particles, while existing particles absorb water in humid air and scatter light more effectively. Stable atmospheric layers and weak mixing can retain pollution near the surface. Rain removes particles through scavenging, but rainfall over Singapore does not necessarily extinguish a peat fire hundreds of kilometres away. The conditions that clear local air and those that terminate regional burning are therefore related but distinct.
Major episodes and the lessons of their chronology
Singapore experienced transboundary smoke before the late twentieth century, but the 1997 episode became a defining regional environmental disaster. Fires spread across large areas of Indonesia during exceptionally dry conditions associated with a strong El Niño. Smoke affected multiple countries for extended periods. The episode drew attention to the importance of peat combustion and demonstrated that land-use decisions in one jurisdiction could create major economic and health consequences elsewhere.
The 1997–1998 fires also changed scientific understanding of the relationship between tropical land management and the atmosphere. Research showed that emissions could not be understood solely by measuring above-ground forest loss: burning organic soils released carbon accumulated over much longer periods. Regional haze subsequently became a recurring subject of atmospheric research, public health planning, and ASEAN negotiations rather than an isolated visibility problem.
In October 2006, another important haze episode affected Singapore and neighboring countries. It reinforced the recurring association among seasonal dryness, fires in degraded landscapes, and favorable smoke-transport pathways. The recurrence also illustrated that a regional agreement or a temporary reduction in fire activity did not by itself remove the underlying susceptibility of drained peat and fire-dependent land-clearing practices.
June 2013 produced an especially intense but comparatively concentrated Singapore episode. On 21 June, the country’s three-hour Pollutant Standards Index reached 401, then a record reading. This figure is frequently quoted, but it belongs to the reporting system in use at that time and should not be directly equated with a later 24-hour index value. The event demonstrated that Singapore could experience extreme short-term smoke exposure without a simultaneous strong El Niño.
The 2015 episode was more prolonged and coincided with severe regional drought during a strong El Niño. Singapore experienced repeated poor-air-quality periods from the later southwest monsoon into the inter-monsoon season. Primary and secondary schools were closed on 25 September because of the haze. The geographic scale and persistence of the Indonesian fires made 2015 a major case study in emissions estimation, regional exposure, economic disruption, and peatland management.
Regional burning and haze recurred in 2019, while Singapore also experienced a shorter deterioration in air quality in October 2023. These episodes differed in severity and duration from 2015. Such variation matters: the term “haze year” does not describe a uniform event. Comparisons require attention to peak concentrations, number of polluted days, geographic coverage, the index used, and the amount of time people actually spent breathing affected air.
Measuring haze and attributing its sources
Singapore’s National Environment Agency measures air pollutants through a ground-based monitoring network and publishes air-quality information for different parts of the island. The principal public index is the Pollutant Standards Index (PSI). Its six component pollutants are particulate matter with aerodynamic diameters of approximately 10 micrometres or less, fine particulate matter of approximately 2.5 micrometres or less, sulfur dioxide, nitrogen dioxide, ozone, and carbon monoxide.
Each pollutant concentration is converted into a sub-index using defined concentration breakpoints. The reported PSI is the highest relevant pollutant sub-index rather than the sum or average of all six. During smoke haze, fine particulate matter, conventionally written as , is commonly the controlling pollutant. The PSI categories are 0–50, good; 51–100, moderate; 101–200, unhealthy; 201–300, very unhealthy; and above 300, hazardous.
Singapore began incorporating into its revised PSI framework in 2014. Current 24-hour PSI reporting and separately reported one-hour concentrations describe different averaging periods. A sudden smoke plume may produce a substantial one-hour increase before a rolling 24-hour average fully reflects it. Similarly, the 24-hour index may remain elevated after short-term concentrations have fallen. This averaging effect explains some apparent discrepancies between changing outdoor conditions and published index values.
Visibility and odor are useful observations but are not substitutes for particle measurements. High humidity can make smoke look denser because some particles take up water. Odorous compounds can be detectable at concentrations that do not correspond directly to the fine-particle burden, while harmful particles can remain present when the smoke smell is weak. Different national air-quality indices can also assign different numerical values or labels to similar pollutant concentrations.
Regional surveillance combines ground measurements with remote sensing. The ASEAN Specialised Meteorological Centre, hosted by Singapore, uses satellite observations, meteorological information, and regional analysis to monitor fires and smoke. Thermal detections identify potential hotspots, while visible and other satellite imagery help reveal smoke plumes. Clouds, overpass timing, small fires, and underground combustion limit what satellites can detect. A hotspot is not a precise measurement of burned area, emissions, or legal responsibility.
Source attribution therefore draws on several lines of evidence: fire detections, land-cover information, wind fields, atmospheric transport models, and chemical measurements. Each has measurement uncertainty. A plume traced to a broad source region is stronger evidence of transboundary transport than of responsibility by a specific operator. Scientific identification of a smoke source and legal proof about who caused a fire are distinct tasks.
Why episodes persist and what determines their end
There is no fixed duration for Singapore haze. Three timescales are especially important: the lifespan of fires in source regions, the persistence of weather conditions that transport smoke to Singapore, and the residence and removal of pollution near the island. These timescales can overlap without moving together. A local improvement can occur while fires continue, and a temporary return of smoke can follow several clear days.
An individual plume passage may last hours or a few days as wind direction changes. A broader episode may recur over several weeks if active fires remain within frequently connected source regions. An entire seasonal risk period may extend across several months, especially when drought prolongs burning into a time when wetter conditions would normally develop. These are different meanings of “lasting for months,” and they should not be treated as equivalent.
Peat-fire persistence is a major reason that one shower may not end an episode. Surface wetting can suppress vegetation flames while heat remains deeper in the peat. Effective natural extinguishment requires enough sustained moisture to cool burning material and restore wetter subsurface conditions. The amount and distribution of rainfall matter more than the occurrence of an isolated storm. Severely drained areas can also lose water rapidly rather than retaining it across the landscape.
A change in wind direction may clear Singapore’s air relatively quickly by moving the plume elsewhere. That is a local interruption, not necessarily the end of the source event. Conversely, smoke remaining in the regional atmosphere can continue to reach Singapore after fire activity has begun to decline. Atmospheric mixing, rainfall along the transport route, and the altitude of the plume all influence the lag between reduced emissions and improved surface air quality.
Forecasting consequently has different levels of confidence at different horizons. Short-range assessments can examine active fires, observed smoke, and predicted winds. Seasonal outlooks estimate whether rainfall and temperature are more likely to favor fire development. They cannot specify an exact future date when Singapore will become haze-free, because human ignitions, fire suppression, and small-scale rainfall remain uncertain.
The seasonal transition toward wetter conditions often reduces regional haze risk, but a calendar date is not a physical cutoff. Delayed rains, persistent dry spells, or continuing peat fires can extend the risk. Conversely, favorable rainfall and successful prevention can keep air relatively clear during a season that initially appeared hazardous. A credible duration assessment must therefore be updated from actual fire, rainfall, wind, and air-quality observations rather than inferred from El Niño status alone.
Human exposure and established health effects
The main health concern during vegetation-fire haze is inhalation of fine particles and associated pollutants. Particle size influences how deeply material penetrates the respiratory tract. A substantial fraction of can reach the small airways and alveolar region. The biological effects depend on concentration, exposure duration, particle composition, and individual susceptibility, not simply on whether smoke is visible.
Established effects of particulate air pollution include irritation of the eyes and airways, cough, and worsening of existing respiratory conditions. Fine-particle exposure is also associated with cardiovascular effects through pathways involving inflammation, oxidative stress, altered vascular function, and changes in autonomic regulation. Smoke is a complex mixture rather than a single toxic substance, and its composition changes with fuel type, combustion conditions, and atmospheric aging.
Children, older adults, and people with existing heart or lung disease are important susceptible groups in official health assessments. Pregnancy is also relevant in the broader evidence on particulate-air-pollution exposure. These categories describe differences in population vulnerability; they do not imply that every exposed individual develops illness or that apparently healthy adults are unaffected. Outdoor workers and others whose activities require sustained time outside can experience greater exposure than people in the same neighborhood who remain in cleaner indoor environments.
Singapore studies have examined relationships between haze or particulate concentrations and outcomes such as respiratory healthcare use and cardiovascular events. Interpreting such research requires attention to temperature, seasonal infections, background pollution, lag periods, and other possible confounding factors. The distinction between an association and a demonstrated causal mechanism is particularly important when attributing a specific person’s illness to a particular haze day.
Long-term particulate-pollution evidence is extensive, including assessments by the World Health Organization, but a short regional smoke episode is not identical to years of continuous urban pollution exposure. Researchers sometimes estimate the mortality burden of a fire season by combining modeled concentrations, population distributions, and concentration–response relationships. These estimates are statistical population burdens, not direct counts of deaths individually certified as caused by haze. They depend on model assumptions and should not be presented as observed clinical totals.
Exposure also varies substantially indoors. Outdoor particles enter buildings through ventilation systems, doors, windows, and leaks; their indoor concentration depends on air exchange, filtration, and deposition. Public-health assessments therefore distinguish ambient concentration from personal exposure. Singapore’s haze response includes official health communications, healthcare preparedness, and institutional arrangements for schools and workplaces. Their operation reflects the fact that the same ambient pollution can impose different practical and physiological burdens on different groups.
Ecological damage and climate interactions
The largest direct ecological losses occur in the burned landscapes rather than in Singapore. Fire can kill trees, simplify vegetation structure, and destroy habitat used by forest-dependent animals. Repeated burning may prevent forest recovery and favor more open, fire-prone vegetation. In peat-swamp landscapes, these changes affect not only above-ground biodiversity but also the hydrological and soil conditions on which the original ecosystem depended.
Peat loss can be particularly persistent. Drainage exposes peat to decomposition, while fire removes accumulated organic material directly. Both processes contribute to land subsidence. In low-lying coastal settings, declining ground elevation can increase drainage difficulties and vulnerability to flooding, especially alongside sea-level rise. Restoring tree cover without correcting the altered water regime does not necessarily restore the original peat-swamp system.
Smoke also modifies the regional radiation environment. As an atmospheric aerosol, it scatters and absorbs incoming sunlight. Dense haze reduces the total solar energy reaching the surface, while changing the proportion arriving as diffuse rather than direct light. These changes can affect plant photosynthesis, surface heating, visibility, and cloud processes. Ecological responses are not uniform: a modest increase in diffuse light may improve light distribution within some canopies, whereas heavy smoke can strongly reduce total available radiation.
For Singapore, the most readily demonstrated environmental effects are degraded air quality, reduced visibility, and altered light conditions during severe episodes. Ecosystem-specific impacts are harder to isolate from weather, urban pressures, and other pollution sources. It is therefore inappropriate to assume that the island’s ecological damage is equivalent to that at a burning peat-swamp source. Reduced outdoor visibility is directly observable; a permanent change in local species diversity requires different and longer-term evidence.
The fires also release greenhouse gases, notably carbon dioxide, from vegetation and soil carbon stocks. Peat carbon may have accumulated over millennia, so its rapid combustion is not balanced by rapid regrowth. Net climate effects additionally depend on subsequent vegetation recovery, continued peat oxidation, and changes in land use. Large fire seasons can make a substantial contribution to a country’s annual emissions.
The relationship with climate change operates in both directions. Fires contribute to warming, while warmer conditions can increase evaporative demand and exacerbate fuel drying under suitable rainfall conditions. However, individual haze events cannot be attributed to climate change merely because they occur during hot weather. Regional rainfall variability, land drainage, ignitions, and management remain decisive. Event attribution and long-term projections require analysis of these interacting drivers rather than a single-cause explanation.
Economic costs and disruption in Singapore
Haze imposes costs through healthcare use, disrupted work and schooling, reduced outdoor activity, and changes in travel and consumption. Some costs appear in financial accounts, such as purchases of filtration equipment or medical treatment. Others, including discomfort, lost recreation, and the value of reduced visibility, are not recorded directly in market transactions. An economic estimate therefore depends strongly on its scope.
Singapore’s Ministry of the Environment and Water Resources estimated the economic cost of the 2015 haze at approximately S$700 million. This figure is a national estimate for a particular event, not a standard cost that applies to every haze season. It should also be distinguished from estimates of Indonesia’s losses, which include extensive damage in the source country. The World Bank estimated Indonesia’s 2015 fire losses at about US$16.1 billion, equivalent to approximately 1.9 percent of its gross domestic product that year.
Outdoor sectors can be especially exposed. Construction, port-related activity, landscaping, transport operations, and outdoor events face operational consequences when visibility or air quality deteriorates. The size of the effect depends on whether work is delayed, moved, reorganized, or completed with additional controls. A reduction in output at one site is not automatically a permanent loss to the entire economy if the work is later recovered.
Schools and households face related but distinct costs. A school closure changes childcare arrangements and can affect parents’ working time, while indoor activity restrictions alter educational and recreational routines even when schools remain open. Household spending on air cleaners and replacement filters transfers money to suppliers but also represents resources used to manage pollution that would otherwise not have been required. Economic accounting must distinguish this defensive expenditure from an improvement in underlying welfare.
Tourism and outdoor leisure may be affected by unattractive visibility, canceled activities, and changes in destination perceptions. Aviation and shipping can face visibility-related constraints, although actual disruption depends on conditions, operating systems, and safety procedures. It is inaccurate to assume that a high PSI automatically closes an airport or stops all maritime operations.
The burden is unevenly distributed. Indoor office work, outdoor manual labor, a family with a medically vulnerable member, and a business dependent on outdoor dining do not experience the same costs. Access to well-filtered buildings and flexibility in work arrangements also differ. These distributional effects make haze more than an aggregate question of gross domestic product: the same national pollution episode can produce markedly different losses in health, income, time, and quality of life across households and occupations.
Regional agreements, enforcement, and landscape restoration
Transboundary haze presents a governance problem because emissions and exposure occur in different jurisdictions. Singapore can monitor its air and regulate domestic activities, but it cannot directly manage every fire-prone landscape overseas. Effective regional control therefore combines domestic land management in source countries, international cooperation, and measures addressing commercial activities connected to burning.
The ASEAN Agreement on Transboundary Haze Pollution was signed in 2002 and entered into force in 2003. It establishes a framework for prevention, monitoring, information exchange, preparedness, and cooperation in responding to land and forest fires. Indonesia ratified the agreement in 2014. Its operation takes place within ASEAN’s intergovernmental structure, where implementation relies substantially on national authorities and cooperation rather than a single regional enforcement agency.
Singapore enacted the Transboundary Haze Pollution Act in 2014. The law provides mechanisms addressing conduct by entities that causes or contributes to haze pollution in Singapore, including conduct outside its territory. Its extraterritorial scope does not eliminate practical investigative difficulties. Establishing responsibility can require access to land records, corporate relationships, operational documents, witnesses, and evidence connecting particular fires with pollution in Singapore.
A satellite hotspot within a concession boundary is therefore relevant but not necessarily conclusive. Boundaries may be disputed, land may be occupied or used by other parties, and responsibility may involve questions of control, authorization, prevention, or negligence. Conversely, the absence of a clearly detected hotspot does not prove that no burning occurred. Enforcement depends on combining remote observations with reliable legal and operational evidence.
Prevention addresses both ignition and the conditions that allow fires to persist. Measures include fire patrols, early detection, locally organized suppression capacity, alternatives to land-clearing fires, and improved management of drained peat. Ecological restoration in peat landscapes often involves canal blocking or other forms of rewetting, together with vegetation recovery. Rewetting must account for the connected hydrological landscape: water-table management on one parcel can be undermined by drainage elsewhere.
Supply-chain initiatives provide another route. The Roundtable on Sustainable Palm Oil and other certification or procurement systems include requirements relevant to burning, land conversion, and peat management. Certification, corporate pledges, and satellite monitoring nevertheless serve different functions from public enforcement and do not automatically establish that every supplier or parcel is free from fire risk.
Sustained reductions in haze depend on whether fire-prone land becomes less combustible and whether actors have workable alternatives to burning. The effects of restoration and prevention are evaluated through groundwater levels, fire occurrence, verified burned area, emissions, and air quality over multiple seasons. A single wet year can suppress fires without demonstrating that the underlying land-management problem has been resolved.