top of page


Cassian Pirard PhD   

-----    [2024 UPDATE] -----

     This webpage is a short summary of the scientific research on the burning season in Chiang Mai and Northern Thailand. It is covering most aspects related to this seasonal event (health, public data, chemical & physical properties, meteorology & climate, sources of pollution, prevention, policies & actions, etc.)

Some specific short answers can be found in the Frequently Asked Questions page.

     This document is based on 'Comprehensive Review of the Annual Haze Episode in Northern Thailand (Pirard & Charoenpanwutikul, 2023)'. For readers who wish to know more about the burning season, the review covers many aspects of the annual haze episode in details.

     The main purpose of this page is to circumvent all the misinformation and misunderstanding regarding this annual meteorological phenomenon by giving direct access (through the reference list in the linked article above) to scientific sources of information.

Go directly to:


Description of Air Pollution (physical)

Description of Air Pollution (chemical)

Atmospheric Effects

Air Quality Index & Visibility

Sources of Air Pollution

Health Effects

Environmental Effects


Forecasts & Future Trends

Public Perception

Public Actions


     The burning season (haze episode, smokey season, PM2.5 event, etc.) is an annual weather phenomenon in tropical Asia occurring at the driest time of the year, causing unhealthy levels of air pollution.


     The burning of forests & crops charges the atmosphere with microdust that reach such concentrations that it affects daily life, particularly causing breathing difficulties for sensitive individuals.

     The burning season extend from South of Bangkok to Southern China, between mid-January and mid-April and peaks in March. The start of the burning season depends how and where it is defined and can vary by 2-3 weeks. The intensity of the burning season greatly varies from year to year and can sometimes prolong to the month of May.

     The main cause of the burning season is a combination of two factors: biomass burning (forest, crops) and meteorological conditions. A large amount of smoke is produced through burning and accumulated in valleys and river basins of Northern Thailand. It is not an urban pollution as it has a regional impact affecting remote valleys far from main cities.

Fig.1. AQI 8 years.jpg

Figure 1: Record of air pollution (PM2.5) in Chiang Mai between 2016 and 2023 as AQI color code. Green and Yellow are days with relatively low air pollution. Orange, Red and Purple are days with high air pollution with unhealthy to hazardous levels. The figure clearly shows February, March and April  as the months with abundant red-purple colours. It also shows that some years (ex: 2022) are mild while other (ex: 2019, 2023) are bad.



Air pollution in Northern Thailand is haze (not a smog) made of

micro-ashes from combustion

Size of microdust is divided in 3 categories PM10, PM2.5 and PM0.1

for below 0.01mm, 0.0025 and 0.0001 mm size

The amount of PM10 and PM2.5 decuples between

the rainy season and burning season

The worst month is March and air pollution is at its highest in the early monring

Microscopic properties of air pollution

     The air pollution in Northern Thailand is a thick haze. It is a dry air pollution made of small particulates with relatively insignificant amount of gases. By comparison, a smog contains  water and is often the result of precipitated toxic gases mixed with water droplets (secondary aerosols). Significant smog is very rare in Thailand and non-existent in the North. In December , it is quite common to confuse haze with fog or mist (a decrease of visibility due to water droplets) as humidity can be very high in mornings (see Visibility).

     The particulate matter that forms most of the air pollution is generally described as 'PM' followed by a number indicating the maximum particulate size. In Chiang Mai, for the general public, the air pollution is simplified in 3 categories:

PM10: Coarse particulates with a diameter of 10 microns or less (<0.01 mm, so <1/10th of the thickness of a hair). It is historically the first fraction to be monitored since detection is easier.

PM2.5: Fine particulates and the most commonly reported fraction in public announcement and monitoring devices as it represents a significant part of air pollution. These particulates are smaller than 2.5 microns (<0.0025 mm).

PM0.1: Nanoparticulates and the smallest non-gaseous fraction of air pollution. These particulates are smaller than 0.0001 mm and are difficult to collect and analyze. Despite being a potentially significant health concern, this fraction is rarely reported due to detection difficulties.

Gases: It includes ozone (O3), sulfur dioxide (SO2), nitrogen oxides (NOx), CO2 and carbon monoxide (CO). These are present in minor quantities compared to particulate matter.

     The type of particulate matter produced in a wildfire depends on several parameters such as the type of fire, the intensity of the different burning stages (ignition, flaming, smoldering), the fuel condition (live, dead, wet), its distribution (dense/light, flat/sloped) and type (grasses, leaves, branches), humidity, wind, available oxygen, displacement after emission, etc. As a result, the exact particulate size distribution (and chemical composition) changes over time and distance regarding to the emission source.

Fig.2. Particle_mass_size_number_distribution.jpg

Figure 2: Distribution of particulate matter according to size. The dotted green curve is the mass distribution and dominated by coarse particulates. The blue curve represents the number of particulates and the smaller they are, the more abundant. The red curve is the surface area distribution and is particularly important for health studies as most bioavailable toxic chemicals are on the outside of dust particulates.

     Most changes affecting physical properties of smoke are solely of academic interest and minor compared with other types of air pollution (ex: fossil fuel combustion, traffic, industrial, intense firestorms, etc.) On average, around 20% (in mass) of particulate matter is PM10, 42% is PM2.5 and 23% is PM0.1 with the remainder above 10 microns and mostly found in direct proximity of fires as they eventually settle down. The PM2.5/PM10 ratio often used to characterise air pollution is quite constant in Northern Thailand as dust gets homogenised. The only exception are fires occurring within a few kilometers of measurements (ex: Samoeng, 2019; Doi Suthep-Pui 2020) leading to coarser smoke and higher concentration of PM10 relative to PM2.5. It has an effect on the resulting AQI that can be determined on PM10 instead of PM2.5 (see Air Quality Index).

     During a haze episode, the concentration of PM2.5 and PM10 increases by 10 times from 20-30 μg/m3 monthly average in the rainy season to background levels as high as 200 μg/m3 with heavy haze events peaking at up to 600 μg/m3 hourly averages and obviously even higher values in direct proximity of fires. PM0.1 behaves slightly differently and the increase is around three-fold between rainy and dry seasons. The high background of PM0.1 indicates that in Chiang Mai, around 1/3 of PM0.1 has a source associated with permanent urban activities and unrelated to the burning season.

Fig.3. Samoeng_30_March_9am.png

Figure 3: Air pollution situation in 2019 (end of March) when large wildfires occurred in Samoeng valley. The proximity of fires pushed PM10 to be the dominant pollutant while PM2.5 is generally the factor defining air quality. Most AQI scale are limited to 500 and the value 989 (PM10) and 796 (PM2.5) are unnapproved extrapolation of AQI usage.

Variations through time

     The hourly concentration of air pollution is strongly controlled by atmospheric characteristics but the actual maxima and minima can depend on the location. In Chiang Mai, particulate matter concentration increases slowly overnight to reach a maximum in the early morning between 6 and 10 am. Minimum concentrations are reached at 3 pm and coincide with the hottest part of the day. It is not unusual for PM concentration to double or triple throughout the day.

Fig.4. 24h variation.jpg

Figure 4: Average variations of PM2.5 (red) over 24 hours show stable levels at night, a strong increase in the early morning and the lowest values in the second half of the afternoon. Ozone (green), a minor pollutant, shows a high doubling of concentration at sunrise (vertical line) followed by a slow increase until 3 pm. These curves.are weather controlled as both PM2.5 and O3 show a relationship with sunrise and temperature (yellow).

     Weekly changes can have 3-4 fold variations and it is not uncommon to go from a 100 μg/m3 background to reach 300 μg/m3 daily average a few days later. These variations are mostly related to emission sources and some weather patterns.

     Monthly variations are related to human activities and weather patterns. The concentration in particulate matter starts to increase dramatically over the whole region in the second half of January ±2 weeks and is often considered as the beginning of the burning season. Some authors make it start earlier in December but there is a general agreement that intense haze, when visibility is constantly reduced and health symptoms are significant, covers the entirety of February, March and April with only a few rare (if any) days with acceptable levels of pollution.

     The yearly average of PM2.5 and PM10 is 30 to 60 μg/m3 and these values are heavily controlled by haze episodes, which despite representing less than 15% of the year above 50 μg/m3, can have concentrations in hundreds of μg/m3. The very low rainy season background contrasting with pollution peaks in the burning season is partly responsible for the different psychological response of northern Thailand residents compared to cities with similar or worse annual pollution (ex: Dubai, Athens, Milano).

     The wide inter-annual variations, essentially dictated by the intensity of biomass burning, are dependent on the global 'climate' and particularly the state of the El Niño - Southern Oscillation (ENSO). El Niño and Neutral ENSO brings relatively high level of air pollution due to optimal meteorological conditions for burning and accumulation of aerosols while La Niña is characterized by more humidity and wind patterns limiting burning and aerosol confinement in valleys.

             Long term trends is a minor scientific topic of discussion, yet, it is omnipresent in the media and always described as a worsening situation. The truth is that it is unclear. Serious research on this subject show trends from a slight increase of 0.3 μg/m3 per year to a slight decrease of 0.8 μg/m3 per year. Considering the variability of monthly averages during the haze season from less than 100 μg/m3 (2011) to >350 μg/m3 (2007), a 10% increase or decrease would take decades to be visible in monthly or annual averages and it would take even longer to be felt as a difference in daily averages. The only conservative conclusion is that the haze season varies too much from year to year to identify a significant trend that could be experienced by Northern Thailand residents.

Fig.5 Annual Variations.jpg

Figure 5: Example of weekly variations of PM10 between 2010 and 2017 showing the extreme variability that exist between good years *ex: 2011) and bad years (ex: 2010)



Air pollution is made of smoke and is mostly (~60%) carbon

Air pollution contains a large number of chemical compounds

useful to identify the sources of emissions

Harmful chemicals are present but concentrations are not very high

compared to urban pollution

Carbonaceous components

     Carbon is the main component of particulate matter. It forms 50-60% of it and is divided into two main groups: elemental and organic carbon. Elemental carbon (EC) is a form of graphite-like black carbon that is very abundant in polluted cities but not as much in Northern Thailand haze. It can either be a residue of combustion (char-EC) or direct precipitation from combustion gases (soot-EC). These two components have slightly different compositions and absorb/scatter sunlight differently and therefore have different role in global warming. Organic carbon (OC) includes complex molecules combined with nitrogen, oxygen and hydrogen. It is the result of incomplete combustion of biomass burning that would normally be almost totally decomposed in CO2 and H2O. Aside from the chemical classification of OC components, an optical classification of OC also exists to explain sunlight absorbing properties of haze.

     The OC/EC ratio is indicative of emission sources. While fossil fuel burning has a very high EC content (OC/EC ~1), wood burning has OC/EC between 0.18 and 0.33 and undefined biomass burning between 0.1 and 0.2. The OC/EC range in Chiang Mai air pollution is 0.12 to 0.29 with small variations due to burning conditions and sources throughout the burning season.

Fig.6. Chemistry.jpg

Figure 6: Summary of chemical composition of particulate matter. The central pie graph shows the main component (carbonaceous, metals and anionic groups). Side pie charts are specific chemicals (inorganic compounds, carboxylic acids, PAH, dioxins, sugar alcohols) found in particulate matter with their respective concentrations (in %). Fractions show in orange to red are carcinogenic compounds

Polycyclic Aromatic Hydrocarbons

     Among organic carbon are the PAHs, some benzene-like compounds, which are of particular concern due to their carcinogenic properties. Concentrations in Chiang Mai are ~10±3 ng/m3 *<1 to 26). To put these concentrations in perspective, Bangkok PAHs vary between 36 and 55 ng/m3 and Beijing 30 to 279 ng/m3 which shows that their presence in not a major public health  concern but is constantly monitored. The variety of PAHs gives some information on sources and show that during the burning season, 55% are produced by traffic & fossil fuels, 15% are produced by agriculture and 30% by forest burning.


     An interesting group of organic molecules that are not a health concern. They form around 1% of particulate matter and are a bit the equivalent of caramel (dehydrated common sugar) for plants as a result of incomplete combustion of cellulose, starch, etc. The proportions between different species of anhydrosugars give some information on biomass emission sources and and can show the contribution of different forests, rice, grass, sugarcane, etc. and fire characteristics to the production of smoke. Some ratio such as high potassium/levoglucosan are indicative of a significant contribution of agriculture since potassium is associated with fertilizers.

Fig.7. Anhydrosugars.jpg

Figure 7: Example of how detailed chemistry can help understanding the sources of haze in Chiang Mai and elsewhere. Levoglucosan, mannosan and galactosan are three common anhydrosugars. The burning of different plant species will produce different ratio that can directly be compared with the composition of haze in Northern Thailand.

Carboxylic acids, sugar alcohols and others


     Other common organic chemicals in particulate matter are carboxylic acids (oxalates and acetates represent 2/3 of them) that form 1% of particulate matter. Like anhydrosugar, their relative proportion helps understanding sources and behaviour of haze. Carboxylic acids are not a health concern, nor a sugar alcohols that form a bit less than 1% of pollution. Other concerning organic molecules such as benzene, toluene, acetaldehyde, xylenes, acrylonitrile, etc., common in urban pollution, are not present in significant or detectable amounts during Northern Thailand haze episodes. The same applies to dioxins, a strongly carcinogenic substance, which has very low concentrations due to the absence of large industrial chlorinated combustion process in Chiang Mai.

Isotopic composition


     Finally, among all organic components, isotopic studies can also shed some light on pollution sources.  Carbon-14 is a type of carbon that can only be found in non-fossil organic matter. That includes all living organisms and recently dead (dead leaves, dry grasses & wood, etc.), agro-industrial products and biofuels. It is completely absent from all fossil fuels, most plastics, etc. Abundance of carbon-14 in haze show that 3 to 8% of total emitted carbon is from fossil fuels in urban areas and down to 0.1% in rural areas. Other isotopes of carbon and nitrogen show that C3-plants (trees, most grasses, most crops) are the main contributors to haze while C4-plants (ex: corn, sugarcane) are at best, a minor factor in pollution emission.

Fig.8 Isotopes.jpg

Figure 8: Discriminative diagram based on carbon and nitrogen stable isotopes. Haze composition is shown as red dots and clearly sit in the C3-plant group. At best, traffic and C4-plants (corn, sugarcane) can only be minor contributors in the air pollution of Chiang Mai province.

Other components

     Other major components in particulate matter are 13-15% of sulfates, 4% of nitrates and 1% of chlorides combined, among other things with ammonium, potassium, calcium, sodium, magnesium, etc.

     Sulfates are mostly precipitates from gaseous components (SO2) produced by fossil fuel combustion and urban sources. Photochemical reactions combines sulfates and ammonium and prevent the formation of sulfuric acid, limiting the pH of the first rains following the burning season between 4 and 7. Ammonium itself comes from urban sources and biomass burning as it is used as a fertilizer in crops. Nitrates also have two main sources as traffic and biomass burning where it is strongly correlated with potassium due to its use as a fertilizer. The origin of chlorine is less defined and could be long distance maritime sources, solid waste burning, fertilizers, herbicides.

     Cations include potassium that strongly correlates with particulate matter and form up to 2% of air pollution. The correlation with levoglucosan and nitrates points towards the use of fertilizers on crops while forests and grassland fires release potassium in lesser amount. Sodium forms 0.5% of air pollution and has an unclear source that could be marine, biomass burning, solid waste incineration or a crustal component. Calcium forms 1% and is associated moslty with soil as well as urban dust (concrete, roads). Aluminosilicates, iron, manganese and magnesium are soil or sub-soil related and best explained by wind-borne dust that can be mobilised by dry weather conditions and fires.

     Trace elements include some more crustal elements such as titanium but also transition metals like chromium, vanadium, zinc, copper, cobalt, nickel and heavy metals (lead, cadmium, tin, mercury). Without exception, all these elements are below WHO threshold for aerial toxicity in Northern Thailand. All concentrations for these metals are well below typical urban and industrial emissions that would be found in large cities. Various explanations have been given to explain their sources such as tire wear, brake pads, roads & road painting, traffic exhaust, sugarcane  pulp, garbage burning, etc.


     Gaseous components include sulfur dioxide, nitrogen oxides, ammonia, carbon dioxide & monoxide and ozone. With the exception of short-lived localised emissions occurring at random times during and outside the burning season, these pollutants are always present at very low levels. Despite the increase during the burning season, gas pollutants do not pose a significant health issue. Ozone concentrations have a systematic daily variation that mirror the behaviour of particulate matter as it reaches a high peak in the mid-afternoon (when PM is at its lowest) due to photochemical reactions (see figure 4).



Air pollution is enhanced by dry, low wind and hot weather

Colder surface temperature and warmer in altitude (thermal inversion)

prevents polluted air to escape from valleys

Shrunk layer where air is mixed prevents dillution over the whole atmopshere

El Nino creates conditions for bad burning season

La Nina is more humid with favorable weather conditions

Climate Conditions

     Biomass burning alone would not be able to create the high levels of air pollution seen over the whole region if it wasn't for a series of other factors that help accumulation of pollution to create thick haze. Many areas of the world suffers much larger annual burning events yet do not reach excessively high level of haze.

     The typical seasonal pattern starts to show above background PM concentration in December followed by a more drastic increase around mid-January to early February, reaching its peak in March with a gradual to sharp drop by the end of April or early May. The initial rise is linked to agricultural fires followed by forest fires in February-March. This time of the year coincides with the increasing dominance of thermal lows over Northern Thailand, producing conditions of hot, dry, stagnant air with clear skies, light winds and a low dew point. These conditions are ideal for wildfires but more importantly, they also help mobilizing the smoke produced by these fires and accumulating it in valleys and basins below 1500 meters. So while regional PM levels are strongly correlated with the number of fires (themselves correlated with seasonal temperature and humidity), it appears that medium ground temperature, low humidity, low wind speed, stable atmosphere with a strong temperature inversion are the factors that explain concentration to very high levels at the bottom of valleys.

Inter-annual variations

     The intensity of these meteorological parameters varies from year to year and are strongly related to the state of the climate pattern El Niño - Southern Oscillation (ENSO). During El Niño or Neutral ENSO, the dry season is prolonged and sometimes extreme, creating conditions ideal for forest fires. These conditions also produce a dominant air flow with a clear NW-SW trend that accentuate the haze situation in Northern Thailand by bringing air pollution from regions experiencing heavy biomass burning. The worse years of the last two decades (2007, 2010, 2019, 2023) are all associated with this situation.  La Niña produces a higher amount of precipitation and atmospheric humidity, limiting fires but also induces more erratic air flow with no specific source for long-distance aerosols. The best years in recent times (2003, 2011, 2022) had a clear La Niña weather system.

Fig.9 ENSO.jpg

Figure 9: A rather complicated plot of monthly PM10 values between January 2010 and December 2023. The colour code (red-orange for El Nino, yellow for neutral, greenish to green for La Nina) shows that the average haze during the burning season can be at least twice higher during El Nino and Neutral ENSO than during La Nina.

Weather Conditions

     The mixing layer of the atmosphere is a zone where air is carried by air flow vertically, in effect, mixing the whole lower atmosphere. While night-time mixing layers are always near the ground, the daytime thickness of the mixing layer is usually 3±0.5 km thick. During the burning season, this layer is reduced to almost nothing well below 1000 meters during daytime. It means that pollutants, instead of being diluted over several thousands meters of air, are concentrated in a collapsed layer. With the exception of some high mountains such as Doi Inthanon or Doi Ang Khang that penetrates out of this layer, the whole region lies within it and the situation is particularly problematic in valleys and basins where the high air pressure and topography produce a thermal inversion layer trapping pollutants close to the ground.

Fig.10. Inversion.jpg

Figure 10:  Atmospheric profile for temperature during the months of the burning season with a qualitative curve of air pollution distribution and the maximum height of mixing layer (horizontal lines). The daytime mixing layer is at its minimum in March and the highest thermal inversion is also observed for that month, up to 10ºC increase with altitude.

     The temperature inversion reaches its maximum during early mornings of March when the ground temperature can be 10ºC lower than a 1000 meters above. Lower temperature means higher air density, strongly limiting vertical movements and hills surrounding valleys limit any horizontal flow. As the ground heats up during the day, the thermal inversion eventually dissipates in the afternoon which explain why air pollution is generally a bit better at that time of the day.

     Another related factor is the humidity saturation and the difference between air temperature and dew point. In the weather conditions of the dry season, with a thermal inversion, there is very little chance to reach conditions where humidity could precipitate into rainfall and capture particulate matter in the process.

Daily Variations

     The change in atmospheric conditions throughout the day has a direct impact on air pollutants and 10-fold daily variations (70 to 600 μg/m3), dependent on emission sources, are recorded. Particulate matter highest concentrations occurs at 9 am in January-February to shift to earlier time (6-7 am) by mid-April. Lowest concentrations are met in mid afternoon (3 pm)

     As the sun rises, higher atmospheric layers are warmed up while ground in valleys is still in the shade, producing the strong thermal inversion trapping pollutants that couldn't escape during the night due to the suppression of the mixing layer. By 8 to 10 am, the atmosphere is warming up, initiating a slight decrease in pollutants until mid-afternoon. Ozone (and sulfates) shows an opposite trend, increasing after sunrise when the stable nocturnal layer break up and reaches a peak at 3 pm due to photochemical reactions. At night, ozone is decomposed or deposited and reacting with surfaces.

Fig.11. Haze cartoon.jpg

Figure 11:  Schematic representation of the state of the atmosphere during the burning season (top) and outside (bottom). A fire producing smoke is dispersed over a considerable atmospheric column in the bottom case while burning season conditions trap this smoke in valleys due to low mixing layer and thermal inversion


The Air Quality Index (AQI) is a scale to estimate

the health impact of air pollution

In Northern Thailand, the AQI is only determined by PM2.5

Most of the year is 'healthy' to 'moderate' and

becomes 'unhealthy' to 'hazardous' during the burning season

Air pollution decrease visibility down to a 10-20km with mild pollution and sometimes below 1km during intense haze events

Air Quality Index

     The Air Quality Index (AQI) is a quick estimation tool designed by environmental authorities in different countries to assess the effects of air pollution on human health and inform the public with a simple scale. The scale is not simply proportional to the concentration of pollutants as the index is assessed on the health effects that each measured pollutant has on the human body. As a result, the correlation between AQI and pollutants is not continuous and change when certain threshold are passed. It also has no cumulative effect. Only the highest pollutant has an impact on the AQI,regardless of the concentration of the other pollutants.

     Components that enter into the AQI determination are fine particulate (PM2.5), coarse particulate (PM10), ozone (O3), carbon monoxide (CO), Nitrogen oxides (NOx), Sulfur dioxide (SO2) and in some scales (i.e. India, Australia), other pollutants such as ammonia (NH3) and lead (Pb) are also measured. The different pollutants are computed into a piecewise linear function with various breakpoint (5 to 10) providing a result ranging from clean air to very hazardous air pollution. For particulate matter, the AQI considers an average of PM-related health effects established by government agencies and reassessed every few years based on recent research  and changes in air pollution in the region where the AQI is applied.

AQI Visibility
Fig.12. AQI-O3-PM.jpg

Figure 12: Equivalence between particulate matter concentration (PM2.5 and PM10) and ozone and the resulting AQI. The different break points are visible in the change of slope for different pollutants.

     In Chiang Mai, the AQI is determined by particulate matter as it is by far the most dominant pollutant. Most of the time, PM2.5 concentrations dictates the AQI but in some rare cases, when nearby forest fires are present, PM10 can take over temporarily. Other pollutants are almost never a main concern and their independent AQI value is moderate during the burning season while PM is high in the unhealthy to hazardous range, resulting in an AQI permanently solely defined by PM.

     Since AQI is a non-linear index based on health factors, different scales are used with different values and breaking points. The US AQI is very similar to the Chinese AQICN and scales used in Asia also have a similar range of values (0 to 500) but with slightly different definitions. Other scales used in Europe, UK, Australia, India, etc. can have different indices (0 to 10, 0 to 100) but health warnings for similar pollution levels remain essentially the same. In recent years, an homogeneisation of world data is available on the World Air Quality Index Project ( where all data is recaculated on the US AQI to avoid confusion.

     Some misunderstanding and inaccuracies can still occur. The close proximity between actual concentrations and AQI for particulate matter can be confusing; the mixing up of average values over different time periods is common place. There is also an inherent issue with AQI that assume all dust air pollutions are the same and calibration is mostly made on urban pollution which is considerably different to biomass burning of Northern Thailand

     A Thai AQI scale exists as a local index to estimate the health risk of air pollution. It differs from the typical US AQI or AQICN by having a higher tolerance for low and medium level of pollution (up to 400%) while providing higher AQI numbers in the unhealthy levels of air pollution (15 to 25%). The index is poorly defined and it is unclear in which condition the Thai AQI scale is used

Fig.13. USAQI-THAQI.jpg

Figure 13: Comparison between the US AQI and the Thailand AQI scale for PM2.5 and PM10. Divergences are shown relative to equivalence shown by the black dotted line.

     In recent years, in absolute terms, Thailand has lowered its warning level from 50 to 37.5 μg/m3 (AQI 135 to AQI 100 on the US scale; AQI 100 to AQI 50 on Thai scale) for daily averages. These values are worth comparing to WHO guidelines for yearly average pollution which are 15 μg/m3 for PM10 and 5 μg/m3 for PM2.5 with an upper daily limit of 15 μg/m3. Both PM categories have very unrealistic goals to be reached in South-East Asia since the natural background is relatively high to 10-20 μg/m3 in the rainy season.

AQI variations


     Outside the dry season, local variations can be extreme (up to 100x background) for a single pollutant. When sources are investigated and found, it is the result of a proximal emission polluting near a monitoring station but defective monitors have also been noticed.

Less extreme variations (<10x background) are likely explained by a local source (burning, house fire, industrial or urban activity) or a specific environmental feature (winds, topography, etc.) that can cause a short-lasting anomaly. Particulate matter size distribution and chemical composition can also change within these anomalies.

     Minor variations are the result of natural variability in air pollution, not so different from cloud distribution in the sky, but at ground level, affecting the distribution of haze. The position of detectors in the urban environment also creates variability when these are placed near traffic, buildings, different elevation, semi-closed spaces, etc.). Detectors also have inherent concenpt issues that create inconsistency in available data. All commercially available AQI-meter are poorly calibrated instruments with low-accuracy and low robustness also affected by time-drift. The precision of these instruments is 10-20%. It creates significant scattering in the data recorded but has no impact on the everyday interpretation of results and health decisions that would follow (i.e. when an AQI is stated as 'hazardous' (>300), it does not really matter if your device is measuring an AQI of 380 or 420). Public sources providing AQI maps of a region where no AQI-meter is present are based on satellite measurement of atmospheric transparency.

             Another source of variation and misunderstanding lies in the use of different scales and measurement characteristics. The scientific approach exclusively express air pollution in mass concentration per volume (ex: microgrrams per cubic meter), avoiding the AQI that is non-linear, based on health factors that can be modified and not consistent between indices. However, for AQI and concentration measurements, the time factor also has to be considered. All data (instantaneous, hourly, daily, monthly, yearly) cannot be compared for different time period due to the associated variability. It is not uncommon to see some media and public sources displaying instantaneous or hourly averages as daily averages, which can present the air pollution as a lot worse than it actually is.


     Along with health effects, low atmospheric visibility is the other main consequence of air pollution experienced by all residents of Northern Thailand. Since Doi Suthep is visible from most parts of Chiang Mai metropolitan area when the weather allows it, it is often used as the first indication of the presence of air pollution. As particulate matter absorb and scatter sunlight, there is a strong correlation between pollution and how far you can see

Fig.14. Doi Suthep pictures.jpg

Figure 14: View of Doi Suthep from Dr Artima Medical Clinic in Mae Hia, (7.5 km from the ridge line). a. In good atmospheric conditions (Low AQI, medium humidity). b. In mildly polluted conditions (AQI ~150) and medium humidity. c. In mildly polluted conditions (AQI ~150) but back lit by sunlight. d. In mildly polluted conditions (AQI ~150) but high humidity (~95%).

     Atmospheric transparency is dependent on a few parameters but the most important ones are particulate matter content and humidity. Carbonaceous matter emitted by biomass fires has strong absorption and scattering properties in the visible spectrum. From the ground, the sun appears redder than normal when light is passing through smoke plumes due to differences between blue and red wavelengths. It also results in thick haze having an orange tinge by through transmission of light but bluish in reflection.

     The reduction of visibility is directly related to these optical properties and has been described in Chiang Mai for decades. When the AQI<50, the visibility is mostly unaffected except for long distances. Between AQI75 and 200, the visibility decreases from 30 to 10 km, progressively hiding topographical features on the horizon. for AQI above 200, the visibility steadily decreases and is eventually observable within the city where light poles, bridges, buildings start to disappear in the distance.

Fig.15. Doi Suthep AQI.jpg

Figure 15: Approximate visibility threshold for the ridge line of Doi Suthep (grey) as a function of AQI PM2.5. This map is only relevant between 10 am and 3 pm for relative humidity below 60%.

     Although this correlation seems easy to apply, there is a strong tendency to overestimate air pollution in the pre-haze season based on visibility. This is due to relative humidity that start to have very strong effect on atmospheric transparency for values above 80% and could then be called fog (or mist) rather than haze. The most distinctive observation between fog and haze in Northern Thailand is the colour; while haze tends towards orange-yellow colours (sunlight through or above) or bluish (sunlight behind the observer), fog is consistently grey but a mixture of the two is possible. Fog can also dissolve quite suddenly.

     In the early stages of the burning season (December-January), it is not uncommon to have a mild early morning fog which makes visibility not compatible with measured levels of air pollution. A 95% relative humidity in clean air over Doi Suthep gives an atmospheric transparency equivalent to an AQI of 300. Although purely observational, the misinterpretation of fog for haze is enough to cause psychosomatic symptoms for some individuals. For most of the burning season (February-April) , the humidity lies between 40 and 60% and the effect on visibility is minor.

Fig.16. AQI_VIS_H2O.jpg

Figure 16: Correlation between visibility (in km) and air pollution given in PM2.5 AQI. Different curves are provided for different relative humidity levels.

     Contrast is also very important. Doi Suthep is always visible at dusk. Air pollution tends to be lower at the end of the day, but more importantly, the sun is right behind the mountain seen from the city centre, creating a strong contrast. In the morning, looking at Doi Suthep from the city, the sunlight is scattering and reflecting significantly on air pollution, preventing any contrast between the ridge line and the sky. The effect is a bit similar to car headlights in a heavy fog. Hard to see an object in front of you but if a car comes towards you with headlights, the object is more visible.

     Altitude of observation has a minor effect and only becomes significant if elevation is more than 1500 m at which point air pollution start to decrease. Other pollutants (gases) concentrations are low enough to have no significant effect on visibility.

     Other effects related to atmospheric transparency is the decrease in radiative flux at ground level by up to 100 W/m2 or even higher. As a result, solar panels efficiency in conditions with an AQI of 150 have an output reduced by 6 to 11% depending on the type of photovoltaic device used.



The source of air pollution in Northern Thailand

is biomass burning (forests, grassland, crops)

In Chiang Mai during the burning season, forest fires

produce ~90% of the smoke

In Chiang Mai, in December-January and May,

ricefields is the dominant source of smoke

In some areas of Northern Thailand, agricultural burning

can represent 50% of smoke production

Cross-country pollution only contributes to background levels

and not intense haze

     Historical data and more recent scientific reports have identified the source of air pollution in Northern Thailand as biomass burning (forest & grassland fires, agricultural residues) and excludes any major contribution from other sources such as ground dust, traffic, industrial and agro-industrial emissions.

     Based on emission of particulate matter, biomass burning produces 85 to 97% of air pollution in Northern Thailand. Detailed chemical studies and ground & satellite land surveys provide a very precise information of air pollution sources. From these data sets, identification of sources such as the type of forest that burn, the type of plant species (trees & crops), the type of fuel (leaves, stems, seeds, sheath, stubbles, etc.), the type of fire (flaming, smoldering, etc.), the contribution of each of these elements to the air pollution and to specific chemicals, variations between districts and over a few days; all of it can be reasonably determined using proper analysis that is too long and complicated to expose here.

Fig.17. Punsompong et al 2021_Land cover.png

Figure 17: Simplified land use map of Thailand showing the surface occupied by forest in the north and the presence of significant agricultural areas only in the eastern part of uppermost Northern Thailand.

Forest fires

     In Chiang Mai province, around 80% is covered by forest with 5% of land allocated to rice paddies and another 5% to other crops. Of these 80%, 70% are considered by the Forest Fire Control Division as subject to fire, with dipterocarp  & deciduous forests carrying the highest risk and also representing the largest surface (86% of forests). At the national level, forest-related haze is estimated to represent 46% but 2/3 of forest fires occur in uppermost northern Thailand and more particularly in Chiang Mai, Tak and Mae Hon Song provinces. In Chiang Mai province, it is estimated that forest fires represent 80 to 92% of air pollution emission during the haze episode. The correlation between forest fires and haze is clearly established for decades with zero fires from June to November, a handful of events in December and May, hundreds in January, February and April and thousands of fire events in March.

     The causes of forest fires are monitored through surveys for the past two decades and are mostly unchanged. 40 to 75% of fires are started to help the collection of non-timber products (mushrooms, bamboos, herbs, honey), 10 to 25% for hunting activities, 6 to 20% for agriculture (slash & burn, land clearing, animal farming), 0.2 to 2% for illegal logging, 1 to 10% for criminal or accidental causes and a significant percentage of unclear causes. In recent years, surveys show that the collection of plant and animal products from forests represents 73 to 82% of identified causes of fires.

Fig.18. Source contribution.jpg

Figure 18: Source contribution estimation to the air pollution in Chiang Mai province. The haze episode (February to April) is dominated by forest fires while the pre-haze period has higher contributions from agriculture burning

Agricultural fires

     This emission source includes the burning of farming residues, clearing of cultivated fields and some pre-harvest crop preparation. Wild grassland is typically included in forest burning and is dominant in some provinces such as Tak. Agricultural burning is applied to rice, sugarcane, cassava, corn, soybean and potato and is practiced at any time of the year. However, locally and depending on the crops, this can be highly seasonal and in the uppermost Northern Thailand, 60% of rice burning occurs during December-Janurary while only 3% happen in February to April. A second post-harvest burning in April to June represent another 30%. 

     It is important to note that agricultural burning, while a major emission source in Isaan and Central provinces, is significant but minor pollution in the north. However, substantial variations occurs between provinces and while heavy haze is strongly associated with forest fires in Chiang Mai; Eastern provinces (Chiang Rai, Nan) have very significant agricultural contributions that represent up to 60% of air pollution emissions.

     The reasons behind agricultural burning are mostly traditional and similar over all South-East Asia, Southern China and India. Farmers believe that it makes tilling easier, control insects, diseases and weeds, release nutrients and increase yeild. These benefits have been scientifically demonstrated but there is equally scientific evidence that burning lead to loss of organic soil, degradation, compaction, erosion and a lower soil fertility with a loss of nutrient and medium-term lower crop production with higher need for fertilizers. Regardless of agricultural and pedological advantages and drawbacks, the main reason for open-burning is economical or technical as it is the cheapest option to clear a field, the most efficient in regard of contract farming (with a short crop rotation) and in some mountainous areas, the only suitable option to clear land.

     In the North, it is estimated that 6 to 11% of rice straw is burned but some regions have up to 50% burning. Corn burning is frequently blamed as a significant source of air pollution but scientific evidence shows that while 25 to 75% of corn residues are burned, the contribution to air pollution in Chiang Mai province is almost insignificant. Sugarcane pre-harvest burning is a very important cause of air pollution in Thailand but with the exception of some fields in Tak, Phrae and Uttaradit provinces, sugarcane is a rare crop in uppermost Northern Thailand.


     There are no large industrial complexes in Northern Thailand. Most factories in the Chiang Mai-Lamphun area are small-scale companies with only a handful of metal-producing, chemical and non-metal (ceramic, concrete, pottery) factories that would produce very small amounts of pollutants. One local exception is the Mae Moh lignite-fired power plant in Lampang that is a potential minor source of particulate matter. No relationship has been found between busy commercial and industrial areas in the North and the contribution is estimated to be <0.1% during the burning season.

Traffic & Urban sources

     In large urban centres, traffic can have a significant influence on air pollution and its composition. Isotopic studies show that fossil fuels form up to ~40% of particulate matter during the rainy season in Chiang Mai city but it is down to a couple of percent during the burning season (and 0.1% in rural areas). However, for some harmful chemicals such as PAHs, traffic can represent a very significant contribution. Urban activities also contribute to some ground-like dust production (roads, concrete) and more specific and potentially toxic pollutants (cooking, garbage burning)


     Dominant winds during the burning season brings air from India and Myanmar to Thailand at an average rate of 300±100 km/day at 1.5 km high. The emission sources are similar to Northern Thailand, mostly forest and agricultural fires and their impact over Thailand will depend on weather conditions in source areas and wind patterns.

     In Chiang Mai, it is estimated that 60 to 90% of background air pollution is provided by long-distance transport from Mae Hon Song and Myanmar. However, during high haze levels, transborder sources contributions are unchanged while proximal emission sources soar due to local (within the province) fires.

     The impact of transborder sources is also variable spatially and while Myanmar & Mae Hon Song are important polluters for Chiang Mai province; in Chiang Rai, a significant amount of smoke is actually produced in Northern Laos due to different wind patterns.

Fig.19 Backward_Trajectory_probability_alt_month.jpg

Figure 19: Probability map for the point of origin of pollutants in Chiang Mai after 3 days for different altitudes (10, 1000 and 1500 meters above ground) and the three months of the burning season.


The toxicity of air pollution in Northern Thailand is not fully determined

but likely less than a large Asian city

For healthy individuals, symptoms are allergy-like (blocked nose, sore throat, itchy eyes)

For individuals with pre-conditions, acute symptoms and hospitalization can occur for asthma, COPD, cerebrovascular and cardiovascular diseases

Lung cancer seems to have no relationship with the level of air pollution


A cigarette-equivalent calculation based on health effects is available here


     Air pollution and its effects on health have been studied for a long time. A large number of studies deal specifically with particulate matter which is the type of pollution seen in Chiang Mai. However, most studies are focused typically on urban pollution (traffic, industrial, urban activities, etc.) which is considerably different from biomass burning, the main pollution source in Northern Thailand. Differences exist in heavy metal contents, organic carcinogenic chemicals and the size of particulate matter itself. At this stage, it remains unclear if very high concentration of particulate matter in Chiang Mai is more toxic than an average urban pollution in large Asian cities but some data indicates that it is likely less harmful than traffic and industrial haze or smog.

     The particulate matter pollution of Northern Thailand has two modes of harmful action: infiltrating the deep respiratory and circulatory system leading to potential alveolar obstruction and the harmful effect of absorbed toxic substances. The general mechanism of harmful action occurs through stimulation of oxidative stress, inflammatory response and genotoxicity.

     Except in specific research studies, all PM fractions, from 10 microns to 0.1 and below, are considered by authorities almost equally toxic. A distinction is made between PM10 and PM2.5 but no consideration is taken for composition, surface area, shape, source, etc. It creates a situation where sea salt microdust and tobacco smoke have a similar toxicity and identical AQI while having in practice very different health impacts. In an effort to simplify data for the public and regulating bodies, air pollution is misrepresented as a simple function of concentration load in breathable air but analytical data show very significant variations too complicated to be exposed here.

Causative agents

     Most of carbonaceous particulate matter is theoretically inert but act as a mild irritant in upper and deep respiratory airways. Fine and ultrafine particulates can penetrate deeply into alveolar sacs, carrying a significant amount of toxic chemicals that can be released in blood capilaries. Adsorbed toxic chemicals on the surface of particulates are more concerning as they use particulate matter as a carrier deeply into the human body where it can be absorbed into the bloodstream.

Fig.20. Lung penetration.jpg

Figure 20: Simplified graphical representation of particulate matter penetration into the respiratory system

     Polycyclic aromatic hydrocarbons (PAHs) are among those concerning chemicals since some species are well-known carcinogenic by altering the DNA replication process. PAH sources in Northern Thailand are relatively mild and air pollution is similar to European or Japanese cities and several times lower than Bangkok. The overall risk associated with PAH in Chiang Mai haze is of 1/10000 chances of developing a cancer after 70 years of exposure.

     Dioxins (PCDD) are carcinogenic and immunotoxic compounds with the ability to bioaccumulate. The lack of large industrial complex in Chiang Mai or elsewhere in Northern Thailand makes these dangerous chemicals relatively insignificant except for smoke associated with garbage burning.

     Some heavy metals have carcinogenic properties. Biomass burning however, does not release large amounts of heavy metals compared to urban sources, so Bangkok and many large cities have systematically high levels of heavy metals than the thickest Chiang Mai haze.

General health symptoms

     For healthy individuals, heavy haze impact is often limited to allergic reactions with mild to very mild symptoms including allergic rhinitis, congestion, sore throat, itchy eyes, headaches and more rarely skin reactions. For very high levels of particulate matter, shortness of breath can also appear. These mild respiratory symptoms are quite widespread, affecting 1/3 of individuals when PM2.5 reaches 35 μg/m3.

     Regular exposure to particulate matter also indirectly leads to aeroallergen sensitization, making healthy individuals more prone to develop a histamine intolerance  from other allergens and new allergic reactions. The prolonged sunlight absorption by haze also reduce vitamin D production by 10 to 20% when PM2.5 is 35 to 100 μg/m3. On the other hand, the absorption of UV by the polluted atmosphere makes sunburns less likely during the hot season.

     Finally, the constant exposure to haze is also associated with mild psychological stress and cognitive impairment for some individuals, causing recurrent thinking about haze, irritability, insomnia and poor concentration. These psychological symptoms are worsened by constant anxiogenic media exposure and the omnipresence of AQI monitoring devices.

Short term diseases

     The toxicity of particulate matter regarding short term exposure and acute symptoms is quite established and known to increase morbidity and non-accidental mortality risk for individuals with health pre-conditions. Significant effects are observed for respiratory and cardio-vascular diseases and adult mortality in high-risk groups. Health issues also affect particularly children due to their underdeveloped respiratory system and higher breathing rate as well as the elderly.

     High level of particulate matter can increase mortality by 1 to 2% with a specific increase of 4-5 % for cardiovascular mortality and respiratory morbidity but these highly polluted day are also correlated with other weather parameters. Overall, the increase in particulate matter air pollution is strongly correlated with hospitalization rate for asthma, COPD, cerebrovascular diseases, myocardial infections and coronary and ischemic heart diseases. Other diseases such as pneumonia, influenza, pulmonary embolism also have mild indrect correlations. On the basis of increased mortality during the burning season, a cigarette-equivalent scale can be established to have a more understandable value on these health effects. On average, the air pollution in Chiang Mai follow the equivalence 1 cigarette = 66 μg/m3 (Click here for more details)

Fig.21. Aungkulanon prevalence map TH.jpeg

Figure 21:Positive (orange, red) and negative (dark blue) anomalies for several diseases based on mortality compared to national average. Significant anomalies exist in the North for COPD, asthma and lung cancer but not for pneumonia.

Long term diseases 

     The long term effects of regular exposure to biomass burning particulate matter are unclear  some studies have established that exposure to PM increases the risk of neurological and cognitive diseases (dementia, cognitive impairment, cognitive development) and metabolic (diabetes) but these research results come exclusively from urban pollution and conclusions cannot be transposed readily to biomass burning.

     Low birth weight and pre-term birth as well as higher infant mortality are correlated with high average PM2.5 but also with economic status as it affects the poorest population of continental South-East Asia (Northern Laos, Western Myanmar).

     Lung cancer is a confusing case. Initial studies show an anomaly for lung cancer mortality in Northern Thailand compared to the national average and it has been instinctively associated with air pollution. However, the lung cancer absolute risk in Northern Thailand is not different from the national average in Japan and significantly lower that the US. Further detailed analyses also show that the high lung cancer risk is particularly found in Chiang Mai and Chiang Rai and more particularly in Amphoe muang, Hang Dong, Doi Lo, San Pa Tong and Doi Saket districts of Chiang Mai province. Such spatial specificity is difficult to explain with air pollution that is broadly homogeneous over the whole northern region. While no correlation can be found between lung cancer and air pollution, the survival rate of lung cancer adult patient after remission is lower in polluted areas than elsewhere.

     Since the correlative link with the prevalence of lung cancer is weak, other causes have been investigated such as indoor radon. Northern Thailand has relatively high radon emission levels compared to neighbouring regions, but it is only mildly above the world average and not different from the European average. Similarly to air pollution, radon distribution map in Northern Thailand does not match the distribution of lung cancer particularly. So although the victims of lung cancer are overly smokers (96% male, 52% female) and have some genetic predisposition to radon sensitivity, the reasons for the national positive anomaly of lung cancer in northern Thailand are still unknown.

Fig.22. Radon_Lung prevalence map.jpg

Figure 22: Comparison between lung cancer anomalies in districts of 6 provinces of uppermost Northern Thailand (left) and radon concentration map. An air pollution map would not show significant differences between districts.


Forest fires only consume forest litter (leaves, branches) and grass

Forest fires are low intensity slow moving fires

Forest litter is rebuilt and stable after 2-3 years without fires

The essential role of fire in mushroom yield is unclear

Forest ecosystem             

     Forest fires in Northern Thailand occurs mostly in dry dipterocarp and mixed deciduous forests, two types that can be found on the Doi Suthep-Pui range for example. These forest have between 1000 and 3500 trees per hectare depending on type, slope, humidity, etc. It amount to 25 to 100 tonnes per hectare of wood and around 5 to 10% can ideally be used as fuel for wildfires. These fires are typically surface fires that never affect the canopy and only conusmes dead leaves, branches, twigs, seeds and grasses. Crown fires (spreading from tree to tree) are only known in cases of illegal logging and deforestation.

     Forest fires typically have 0.2 to 0.9 m of flame height (up to 4 m) reaching 300-500ºC and spread at an average 2 m/min (up to 3.5 m/min). The head fire moves 2 to 15x faster than back fire and 2 to 4x faster than flank fire. Only the top soil layer is affected by fires and reach 200 to 400ºC as they are quickly moving with low fuel density. A couple of centimeters in the ground, the temperature is 30ºC.

     The intensity of fires is controlled by fuel characteristics (moisture, composition, compaction, continuity, ...), atmospheric conditions (temperature, humidity, wind) and topography (slope, aspect, elevation, ...). Since fires only consumes forest litter and grasses, the biomass fuel is fully recovered after 2 to 3 years at which point it reaches stability (dead material accumulates in the dry season and an equal amount rot in the rainy season). Around 60-70% of available fuel is burned during a standard forest fire. The average forest fire is 1 to 15 ha based on satellite data and most of them are between 1 and 3 ha with thousands of burning events per season.

     Surface fires feed on twigs, dead leaves, plants, grass and undergrowth but still affect the canopy, reducing the cover from ~90% to 70% in areas affected by yearly fires. Young trees (<1.3 m high, less than 1 cm in diameter) do not survive fires and larger trees have a slower growth (a few mm less per year). In addition to killing seedlings and saplings, there is a loss of biodiversity associated with regular fires and overall lower soil nutrients levels.


     In Northern and North-East Thailand, one of the main and most discussed reason for starting fires in dry deciduous forests is to promote the production of false earthstars mushrooms (Het Thawp, Het Kra Bueang, Het Pho Fai, Het Pho Hnang). The immature fruiting bodies of these mushrooms grow in May-June and collected to be sold on local market at 300 to 500 ฿/kg with the excess preserved in saline solution for exportation.

     At this stage, the role of fire in the production of these mushrooms is still debated. It is thought that fire is not an absolute requirement and serves a role to create very dry soil conditions when the forest litter has been removed. The absence of leaves and grasses after a fire also facilitate harvesting conditions. To this date, all experiments in vitro to promote the growth of false earthstars mushrooms have failed.




Prevention is only possible through mechanical air filtering

Outdoor, the use of masks, preferably N95 rating is recommended

Indoor, the use of air purifiers or positive pressure systems

with HEPA filters are recommended

Outdoor techniques for mass filtering show no evidence to work

     Seasonal air pollution is not going to disappear anytime soon. While some years (2003, 2011, 2022) had very mild pollution, most years have several days of extreme air pollution and in the worst cases (2007, 2010, 2023), several weeks when air filtering becomes essential. Three approaches are used to reduce locally the amount of air pollution. At a personal level, the use of masks and respirators; at a room or building level, the use of large filtering devices and outdoor techniques.

     Filtering, as home devices or masks, are not just simple sieves as it was often implied during the COVID pandemic by some group of people. Several physical proceases (sieving & impaction, electrostatic trapping, Brownian motion, adsorption, ...) are at work in an air filter and makes them efficient over a large range of particulate size. For most filters, the lowest filtering efficiency is reached at ~300 nm (0.0003 mm). For larger particle size, many filters are close to 100% efficiency and the same applies to some extent for smaller particulates.

     Different scales are used to measure the efficiency of a filter, among them the Minimum Efficiency Reporting Value (MERV). It ranges from 1 to 20, with MERV 8-18 is a typical A/C filter (low EPA), the commonly used HEPA (i.e. car A/C filter) is MERV 17-18. With the exception of some specific professional filters, all these filtering membranes have their weakest point around 300 nm.

Fig.23. Filtering_size.jpg

Figure 23: Comparison of Northern Thailand haze distribution (mass: red; particulate number: dotted line) with other common pollution and micro-organisms. The range displayed by filters here is partly misleading as all filters work for the whole interval of particulate size but with different efficiency.

.Personal Protective Equipment

     When a PM-based AQI reached dangerously high levels, a mask with a high filtering efficiency is strongly recommended. Unless you have specific requirements (no negative pressure, high permeability, psychological effects, medical recommendation), a mask of N95 or equivalent rating is appropriate. Outflow valves on masks are also important as it does not get soggy due to breath humidity, reducing the risk of rash and bacterial infection. It also prevents the accumulation of CO2 in the mask and minimally improves the efficacy of the respirator by maintaining negative or neutral pressure.

     Masks are not entirely beneficial. The increase in humidity, inhaled temperature, lower oxygen intake and inhalation resistance have minor repercussions on blood glucose level, muscular abilities, cardiovascular efficiency and mental skills. It is negligible for a normal effort but during intense physical exercise, it can be significant, increasing heart rate and reaching the maximum oxygen debt in a shorter time. Some individuals are also psychologically affected when wearing masks with feelings of discomfort, anxiety, claustrophobia, etc. People wearing glasses can be affected by fogging due to increased humidity.

     Individuals with respiratory and/or cardio-vascular issues should consult a doctor before wearing a N95 or above respiratory mask. It is possible that such masks might cause more issues than breathing polluted air. Children and people with facial hair should wear a well-fitted mask.  Leaks due to a beard or improper size  will make a mask not filtering properly while making breathing more difficult.

Commercial masks (no rating)

     A lot of masks are available for sale for a variety of purposes and do not follow any filtration standard. These are made of all kinds of materials and various weave and have very different filtering abilities. While some masks will filter 70-80% of particulate matter, it can also be as low as 10%. Surveys also show that some masks are equivalent to a N99 but these are exceptions. The main issue with these masks is the lack of reliability. Since there is no standard to be enforced, filtering efficiency can only be obtained from surveys and possibly reliable brands

It is however worth noting here that some standard-rated masks from some companies fail to reach the filtering efficiency they are suppose to achieve, so a standard is not a guarantee either.

Surgical masks (EN14683, ASTMF2100, YY0469)

     These masks are traditionally used in the medical environment. They are made of a random mesh of polypropylene fibers and surprisingly, contrary to popular opinion, these materials are as effective as N95-rated fabrics in filtering particulate matter. However, since these are not negative pressure equipment, it allows airflow on the side, limiting the practical efficiency to 65-80%.

     A common medical practice in infectious environments is to tape the sides of surgical masks to seal off openings and doubling masks. In such cases, the filtering efficiency is relatively high, reaching 95 to 99.5% of filtering. Single use 'surgical-like' masks have to be considered in the above category of commercial masks.

Fig.24. SEM_Mask.jpg

Figure 24: Electron microscope imaging of various masks at different magnifications. (a) cotton flannel masks with disorganised layer (b) Woven polyester mesh with very organised layer (c) Cross-section through the filtering layer of a N95 mask made of melt-blown polypropylene fibers.


     Respirators are masks held into a semi-rigid, well-fitting structure and made of polypropylene fibers manufactured in an electrical field, producing electrostatic filtering. These masks range from the most basic air-purifying mask to air-supplied systems. This latter group is not covered here as it is used for very specific application in extremely harmful environments or life-threatening medical conditions.


N95 (also FFP2, KN95, KF94, P2, DS2, PFF2) is  a respirator that filters 95% of particulates at 300 nm and most brands are considerably above the rating they are supposed to achieve. These are the most adequate and widespread masks to filter Northern Thailand air pollution. For the purpose of air filtering in Chiang Mai during haze episodes, respirators can be reused until damaged, soiled or causing increased breathing resistance. It is only in medium biosafety settings that N95 and equivalents should be systematically discarded after use.

N99 (also FFP3) is a respirator that filters 99% of particulates at its weakest value but many brands have a practical filtering value higher than that.

N100 is a respirator that filters 99.7% of particulates at its weakest value. These masks (N99 and N100) have increased breathing resistance.

R-rating is similar to N-rated masks but oil-resistant. They have a short-life and have to be disposed after use. These are for technical work in oil-ladden atmosphere and have no use in Chiang Mai pollution.

P-rating is similar to R-rated respirators but oil-proof, giving them a long lifetime. Again, these are of no use in Chiang Mai pollution.

HE: High Efficiency respirators are similar to P100 but require a powered source. It is only for specific applications and persons medically disqualified from negative-pressure masks.

Non-mechanical respirators are masks using cartridges filled with activated carbon or resins to filter toxic chemical compounds from the air. They are not suitable to filter most particulate matter but will remove gases and nanoparticulates. They are of no particular use in Chiang Mai.

Fig.25. Mask filtration.jpg

Figure 25: Filtering efficiency curve of various masks. The dip in filtering efficiency at 0.3 microns is obvious for all masks and also show that outside that size range, (larger & smaller), many masks have ~100% efficiency. The variability (minimum and maximum measured values) for surgical masks and cloth masks is also provided.

Indoor Filters

     Particulate matter enters closed space through advection (open doors & windows) and infiltration (gaps in the structure). A variety of techniques exist to filter the air inside a closed space, including DIY and commercial air purifiers, modified A/C units and positive pressure systems.

     Air purifiers are essentially simple fans pulling ambient internal air through a filter and capturing particulate matter present. They are typically fitted with an HEPA filter but commercial devices occasionally present alternative filtering techniques exposed below. The homemade addition of an HEPA filter on an A/C unit has a similar effect but high MERV-rating filtering are not always suitable for some A/C units as it can bring considerable strain and damage to the motor.

     Positive pressure systems pull air from outside, creating a higher pressure indoor and preventing polluted air to get in from other openings.  Positive pressure systems have the advantage of replacing indoor air with clean air, maintaining environmental level of CO2, radon and other indoor pollutants but comes in direct competition with A/C system since they constantly bring hot (or cold) and humid (or dry) air inside the building. Some positive pressure system are directly hooked on the HVAC house system but this practice is relatively rare in Thailand.

EPA Filters

     It stands for Efficient Particulate Air filters and are membranes with a filtering ability from 80 to 99% (MERV11-16). These are often used in system that requires filtering but do not excessively restrict air flow. These filters are the type found in heating, ventilating and air conditions devices and pre-filters for other systems.

HEPA Filters

     It stands for High Efficiency Particulate Air filters and made of a random mesh of polypropylene or borosilicate glass fibers. Various ratings (MERV17-18) are available and will stop up to 99.97% of particulate matter. Their relatively high airflow capacity with very good filtering standard make these filters the most suitable to filter Chiang Mai air pollution during the burning season.

ULPA Filters

     It stands for Ultra-Low Particulate Air filters and is the level above HEPA. Their rating is MERV19-20, filtering 99.9999% of particles at 120 nanometers and is therefore particularly effective in that low efficiency gap that other filters have. The high filtering capability considerably reduce airflow and limit the volume of air to be treated. Since it requires several pre-filters, it has a higher cost than a standard HEPA, with more frequent changes for a minimal to zero gain in filtering natural background pollution. The typical use of such filtering devices are clean rooms used for medical, chemical and microelectronic applications where air locks, closed-up dedicated suits and strict entering procedures have to be followed for these filters to actually be effective.

Adsorbing filters

     An entirely different type of filter made of activated carbon, silica gel or zeolites. The very high surface of these porous form of carbon, silica and silicate captures nanoparticulates and some gaseous contaminants through the physico-chemical process of adsorption. These filters are not suitable without an HEPA filter as they saturate quickly but have their use to trap some volatile components and smells from biomass burning.

Silver Ion Filters

     These are filters that inactivate bacteria, fungi and viruses carried by air pollution. Operational conditions (flow rate, exposure) makes these devices largely insufficient to efficently disinfect purified air. Even if these devices were working, their purpose in Northern Thailand pollution is limited since biological organisms are almost absent from haze.

Air Ionization

     A device creating a high voltage to ionize molecules in the air. Through electrostatic attraction, ionised molecules can be removed. Such filters are used in professional environments but the lack of standard in low-cost air purifiers is problematic. The filtering efficiency only applies to the finest particulates with no guarantee on how efficient they are but more importantly, the lack of construction standard can lead to the production of ozone, nitrogen oxides, formaldehydes, etc. from an initially clean air and resulting in being more harmful than beneficial for health.

Ultraviolet purifiers

     A device emitting strong UV-C radiating pumped air to damage the DNA of bacteria and viruses passing through. The process is used in some setting such as ultraviolet germicidal irradiation but in practice, like silver ion filters, the exposure to UV of the air passing through such devices is 10 to 100x too short to significantly kill microbes and of little use in Chiang Mai haze.

As it is the case with air ionization, UV-C can produce large amount of ozone and the lack of construction standard is a source of concern.

Outdoor Filtering

     This section covers techniques sought to reduce air pollution in an outdoor situation without direct action on emission sources. It includes various use of wet deposition, scaled up mechanical filtering devices and meteorological actions.

Water spraying

     Every year, the local government promotes the use of large water canons spraying droplets in the air to catch particulate matter around Thapae Gate and various suburbs. This technique is commonly used and effective at mine sites and dust-producing factories but spraying of water in open spaces with diluted air pollution has been shown to have no effect in an urban environment.

Atomizers & fogging

     A similar techniques that relies on wet deposition and normally used for its cooling effect as countless tiny water droplets evaporates. Scientific studies of such technique for catching particulate matter show that it can make things worse on hot and dry days with mild pollution by dissolving and then reprecipitating soluble particulate matter, effectively increasing the PM2.5 by several times. The effect is unclear on a very polluted day.

Giant Air Filters

     Another regularly advertised technique by the local government in their fight against air pollution, the concept is to push air through an HEPA filter using a very large fan. A simple calculation in a theoretical static situation (no new polluted air coming from outside) shows that thousands of Airbus-380 engines working at full power for weeks would be required to clean the air over Chiang Mai metropolitan area.

Cloud Seeding

     A yearly practice in Thailand using various salts, dry ice, urea and silver iodide released in clouds to initiate precipitation, leading to rainfall. In North-East Thailand, where this technique is routinely applied, scientific scrutiny show mitigated results with either weak statistical evidence or inconclusive effects despite the Royal Rainmaking Project claiming an 89-100% success rate. Cloud seeding to fight against air pollution is attempted in other countries but the specific meteorological conditions required for a successful seeding (high humidity, mildly turbulent conditions, cloud cover, etc.) are rarely met during the burning season making this option unlikely to work.

Fig.26. Outdoor example.jpg

Figure 26: Various outdoor techniques to supposedly reduce the amount of particulate matter air pollution



     Some general forecasts on the severity of a haze season can often be made months ahead  depending on the status of the ENSO. La Niña typically brings relatively wet conditions and wind patterns that limit the size and frequency of fires and accumulaton of pollution. El Niño and neutral ENSO tend to have clear patterns that accentuate the production of smoke and its accumulation.

      Many predictive models for daily forecasts are available, using meteorological characteristics, atmospheric properties such as transparency, land use, topography, hotspots analysis, etc. and are relatively robust in their predictions.

     Long term analysis of variability of air pollution between years does not show any strong trend towards a worsening or improving situation. Howevere, urbanisation, increase of private vehicles, agriculture production growth and changes in agricultural practices could lead to variation in air pollution over time.

Fig.27. Punsompong Chantara 2018_ENSO Hysplit.png

Figure 27: Comparison of air flow in the lower atmosphere between La Niña (top) and a neutral ENSO (bottom) year. In addition to more rainfall, wind patterns are less problematic during La Niña



Haze Episode Severity

     The burning season has been described more than a century ago and probably existed for centuries before that. Until recently, the population has been quite indifferent to the issue but fear and resentment have developed in the urban population since 2007. This fear has been mostly entertained by the media, usually using misleading or false head lines such as 'continuous haze since 2007', 'worst in the world', 'extremely toxic smog', 'constantly getting worse', etc. but also government and activists who, for various purposes, target specific groups and often paint a situation worse than it is in such cases.

     The public perception of long term trends show that the urban population is convinced of a worsening situation while the rural population has a wider range of opinions (30% decrease; 20% unchanged; 30% increase and 20% fluctuating) and the only honest scientific assessment of the long term trends is that it indeed fluctuates with no clear trend.

     The urban population is also prone to racial and social stereotyping of air pollution emission sources by targetting non-Thai (hilltribes) as guilty of deforestation, open burning, etc. or rural Thai burning agricultural fields and forests. Such framing of the situation is often made to absolve a specific group or an explanation to the lack of progress in solving the haze situation. It is most obvious with urban, rural and official positions easily blaming pollution produced by neighbouring countries to explain the failure of some regional actions.

Fig.28. Long term trend.jpg
Public perception

Figure 28: Weekly variations of air pollution as PM10 concentration between 2002 and 2016 with published long term trends set at 100 μg/m3.  The strongest trend has barely a scientific statistical significance after 2 decades let alone a public significance.

Health Impact

     The health impact of haze is acknowledged by all population groups. The urban population is overall more familiar and experiences more effects from serious haze than rural population, partly due to topographic and weather features but also the level of education and have developed protection habits accordingly. However, a significant proportion of the educated urban population has developed a disproportionate and sometimes unfounded fear of haze when compared to similar populations in other polluted (ex: large cities) environments.

     Most of the rural population is aware of the haze effects on vulnerable people but rarely take action to limit emission and only apply some basic preventive habits to avoid health impacts of haze. It is partly due to financial constraints; while a basic face mask is affordable for them, air purifiers are not an option they would consider. In highlands areas (>1000 m), populations are less exposed to high air pollution and less knowledgable regarding health effects.

Measures against haze

     The first indirect public involvement with air pollution was ecological activists in the 80s and early 90s trying to protect forested areas from extensive logging & the construction of damaging public projects. In the 90s, a schism appeared between strictly conservationists views (urban) and 'forest guardians' that sees human influence has inevitable and seek community participation in sustainable forest management. For the last decade since the 2014 coup, official actions are on the side of urban conservationists, neglecting the needs of the rural population.

     In all populations (urban, low & high lands rural), the objective of preventive actions to reduce crop residue and forest burning is understood and accepted. Rural populations sees state management policies as acceptable to good while only a minority of rural and most urban populations sees these actions as insufficient with no long term management to deal with seasonal haze. Strong restrictive regulations are supported by the urban population while farmers are not supportive, particularly when no time is given for adaptation and no alternative provided. 




Traditional Methods

      Burning to clear and manage land is a regional traditional and cultural habit. In the past (pre-90s), the village council would manage fire prevention with annual clearing of fire breaks, rotation of cattle to reduce forest fuel and prescribed burning made to avoid spreading and damage to communal and private property.

     Agricultural burning required pruning and physical cleaning of fields with a decision to burn depending on meteorological factors and a semi-strict procedure for burning. In highlands, Swidden agriculture practiced by hilltribes was done with proper management and used partly as a preventive measure against larger wildfires.

     Political changes associated with the end of the cold war has considerably modified land rights with extensive expropriation of people living on forest margins. In the 90s, the state claimed authority on all natural public resources such as forests and promoted the installation of a weak community leadership. These measures put an end to the traditional management of biomass burning and a complete loss of responsibility of local communities towards wildfire management. 


Prescribed burning

     It was done traditionally and still is today with or sometimes without approved administrative oversight. Its application at a larger scale is still limited as it can only be executed by the Department of Forestry, mostly during the cooler months of the year. While prescribed burning is often practiced in some countries to avoid the accumulation of litter over several years or decades and create unmanageable wildfires, it is not the case in Thailand since fuel accumulation reaches balance after 2-3 years.

Burning bans

     Burning bans exist since 2013 but are only seriously enforced since the end of 2016. Although some data show a dramatic impact of the burning ban, it is biased and mostly due to meteorological factors. For years when the ban was ineffective (ex: 2019, 2023), the blame has been set on transborder pollution by some officials and researchers. Detailed analyses of emission makes it however obviously clear that when conditions are optimal for wildfires, the zero-burning policy has little impact on the abundance of local fires. However, its application to agricultural open-burning is more respected with some positive results.

Legal Enforcement

     With the change in land rights applied in the 90s, the state owns and manages forests but these areas remain openly accessible to anyone. With a manpower of ~100 km2 per official, it is largely insufficient for the state to monitor and controll all areas subject to fire, creating numerous, mostly uncontrolled wildfires. The yearly governmental attempt to impose to local people a duty of care towards forests to try to infuse guilt into rural folks is technically illegal since local people have legally no right to manage forested areas.

     The burning ban applied to forested areas has limited sucess, and while there is some fear of legal consequences, there is little understanding of health & environmental impact of forest fires and haze by rural people. The increased accessiblity to forested areas makes it now easy to start a fire that is quickly unnattended and while in the past, rural communities had some knowledge of the actions of outsiders and villagers within their local environment, they have now little control in identifying culprits. The zero-burning ban is also seen by farmers as a way to control their behaviour by the government rather than solving haze problems.

     In remote areas, there is also an issue of compliance of local officials who fear to lose their popularity in the local community, display some empathy towards offenders or have close relationships with them. The amount of extra work required to enforce the burning ban or fire management activities during the burning season come with no additional pay, which create a low motivation in the implementation of regulations.

     Alternative to open-burning of crops and rice paddies is applied locally or in other areas of Thailand. The valorization of biomass through biorefinery into energy products has been suggested to be economically profitable but for many remote areas, it would require considerable government support to actually create an incentive. Other agricultural practices such as triple cropping, cattle farming for straw management and leaving straws to improve soil characteristics are occasionally applied but varies between regions.


Systemic issues

     Thailand has a very centralized approach to governance. The National Environmental Quality Act of 1992 was the first legislative instrument to deal with air pollution and fires and allowed some devolved management functions to provincial governments. In 1997, following the catastrophic Indonesian fires, a new national action plan was designed but aimed at the Southern Thailand haze issue and was broadly inapplicable in Northern Thailand.

     In the 2000 decade, the central government started to impose pre-haze burning controls, patrolling and extinguishing fires but budget allocation, monitoring, post-disaster management and pre-haze management were never serious political actions. Following the 2007 burning season, more centralized policies appeared with some devolved governance to the provincial level but restricted to data collection capacity rather than an active role in tackling the haze problem. Open-burning control, forest fires and public information was all decided in Bangkok with little relevance for local communities in Northern Thailand.

     In 2010, another particularly bad burning season, the parliament passed an approval for urgent solving of haze episodes in Northern Thailand, producing strict fire prevention measures. But the highly regulatory policy regime failed to take into account local conditions, sources of haze and causes of burning, traditional practices and motivation of local people to burn fields and forest. In the government minds, their Bangkok-based policies should cascade down unaltered to provincial, district, sub-district and village authorities and somehow be applicable to local conditions  but such vertical transfer of authority never occured.

     Since 2007, regular meetings are held at provincial levels to tackle the haze problem but in practice, these meetings are systematically attended by minor representatives of various ministries and government institutions with no effective power to commit to anything. These meetings are essentially sessions of information sharing between different ministries in a pre-defined plan and no coherent management of the haze problem. Even the provincial governor authority is under the direction of the Ministry of the Interior and provincial decisions are limited within that constitutional framework.

     In 2016, the first technically effective burning ban was put in place with significant penalties for illegal burning but the top-down approach failed to initiate any true grass-root participatory action. To this date, all decisions regarding land management that include air pollution and forests rarely take local opinion into account and villagers are more coerced into agreeing with authorities decisions than discussing and solving problems.

     At the present time (2023), all top-down government regulations on burning and haze management haven't accomplished most of their objectives. The absence of pre-critical management, an ineffective early warning system, completly inappropriate budget allocation, a one-way communication between locals, academics, provate and civil stakeholders and government contribute to the lack of significant progress in reducing the severity of the haze season.

International Actions

     ASEAN members have produced some initiatives since the early 80s when transboundary pollution was acknowledged as a problematic issue. The fundamental flaw with ASEAN agreements is the lack of liability or compensation scheme to hold responsibility between member states. It is a direct result of the non-intervention aspect of the ASEAN treaty and all agreements are diplomatic, based on consensus building and each member state has the opportunity to frame the haze problem depending on national issues rather than international collaboration.

     Therefore, in practice, ASEAN policies are hardly workable in some regions and tend to try to solve the consequences of haze pollution but never address the underlying causes as it could be seen as interference in domestic policies. The only improvement that ASEAN has brought so far is the regional monitoring of air pollution in the whole continental South-East Asia.

Public Actions
bottom of page