Climate Change in Chiang Mai and Northern Thailand

This webpage is a short summary of the impact of climate change in Chiang Mai and Northern Thailand regarding weather, flooding & droughts and air pollution. Further information and references can be found in 'Floods and Water Management in Chiang Mai and the Upper Ping Catchment (Pirard, 2025)' and 'Comprehensive Review of the Annual Haze Episode in Northern Thailand (Pirard & Charoenpanwutikul, 2023)'.

Go directly to:

Other information about air pollution

Other information about the Ping river
Other information about meteorology

Climate change is often invoked in the media, but also in official and research papers for all kinds of recent extreme events, including droughts, floods, cyclones and air pollution. It is often associated with a narrative rhetoric that things are worse now than they were in the past, often done without reference or support for such ideas other than broad global statements. Even in academic research, modeling is sometimes used to support a claim, but the model itself is poorly designed or inappropriate as they have little relationship with reality or with the effect that is supposedly demonstrate. What is more telling is that all credible advocates of climate change (i.e. climatologists) are very careful not to draw simplistic and direct links between extreme events and long term trends associated with climate change.

General trends for Northern Thailand

The Global Climate Model (GCM) is, as its name shows, a global model to project the impact of climate (precipitations, frequency & intensity; temperature trends, probabilities of drought, etc.) on a global scale with very poor geographical resolution. The model is only applicable for continental or large regional scales where it can provide an average of the changes that could occur in the area. In the media and popular science, the model is often used for projection at a small scale, citing the Intergovernmental Panel for Climate Change (IPCC) to give credit to these projected trends while the IPCC itself clearly stated in 2007 that the local trends, and more particularly extreme events such as flooding or wildfires, cannot be predicted by such model. The GCM is just not designed to do that.

Based on the GCM for South-East Asia, ran for a grid size of 300*300 km2 (basically, the whole Northern Thailand), future climate is expected to increase from +1.0 to +4.5ºC and -20 to +20% of precipitation. Since the GCM is inappropriate to get a meaningful answer for anything in Northern Thailand and Chiang Mai in particular, a bias correction (dynamic downscaling) has to be applied which reduce significantly the probability and intensity of extreme events and extreme trends that could happen, taking the GCM as boundary conditions to regional extremes . The other important element to be added is called statistical downscaling. The process is to use global climate variables and local variations and assume that the statistical relationship between the two will not change in the future (e.g. If a specific year is particularly hot in South East Asia, like +2ºC above normal but the average annual temperature in Chiang Mai is only +0.5ºC hotter, then this relationship will continue to exist in the future) . Such assumption is certainly incorrect, but it’s infinitely better than assuming the whole world will follow an homogenous climate trend.

Figure 1. Future temperatures and precipitation in Chiang Mai based on all published local models for Chiang Mai and its direct surrounding.

Based on such approach, less dramatic and variable results than the GCM are obtained and provide more meaningful answers on how the local climate shift (temperatures, annual rain, length of monsoon or how dry would be the dry season, etc.) and better probabilities can be reached for extreme events with associated uncertainties. All scenarios from 2000 to 2100 give an increase in mean annual temperature of around 3ºC (between 1 and 4ºC depending on the location). The average increase is 0.04ºC/year in the next few decades, calming down after 2060-2070.

Precipitation have a larger variety of results depending on the models. Some show a decrease in rainfall from 6.3 (wet season) to 12% (dry season) (optimistic RCP+2.6 model) to 11.6 (wet season) to 27.2% (dry season) (pessimistic business-as-usual RCP+8.5 model). Other models show an increase of up to 20% in rainfall or sometimes higher. Finally, some small scale models show a slight increase in the northern provinces of Thailand while the Central Plains would see a decrease in rainfall to no significant trends. This very succinct presentation of the data available show how complicated and variable future climate prediction can be.

Climate change applied to the Ping river basin

Changes to the behaviour of the Ping river are linked to rainfall but the relationship is not linear due to various parameters. The broad GCM projection that estimate a change in precipitation from -20 to +20%, translate in runoff variation from -10 to +30% since the change in average rainfall is mostly carried by rain storms. More rainstorms outside the rainy season will produce less runoff than additional rain in the wet monsoon.

Relatively minor changes in rainfall between the rainy, cold and dry seasons, especially in a scenario where it would reduce the contrast of precipitation between seasons, would bring significant changes in water use and management for agriculture. An example is given in a model that shows that for the period 2015-2074, the drought risk appears to decrease. It is not directly linked with higher rainfall or less common drought, but to an increase in the occurrence of wet events throughout the year.

Based on some models, the annual stream flow is expected to decrease by 13 to 19% with a shift in seasonal flow later in the season from the current Aug-Sep to Oct-Nov and possibly more storms in April. Other models show an increase of around 13.7% in annual flow, still lower than the 17.3% increase predicted by the GCM.

Based on some models for extreme events, an expected increase of 89.5% for 10-year floods is expected, 91.2% for 25-year return rates, 20.8 to 30.4% for 50-year events and 10.2 to 22.1% for 100 years. While it marginally increase for extreme floods, decadal floods (i.e. 1973, 1975, 1977, 2011, 2024) could be twice more common and large floods (i.e. 1952, 2005) with a 30-year return rate could occur every 20 years.

Climate change applied to air pollution

There is a handful of papers from a single author on the effect of climate change on air pollution but these are flawed for various reasons. In summary, results show an increase of 1 to 10 mcg/m3 per year in the dry season and a decrease of 10 to 20 mcg/m3 per year in the wet season with an overall decrease per year and discuss at length the health effects of such increase.

Ultimately these results are just a projection based on the distribution of precipitation throughout the year in a future warmer climate. Although the model used is not fundamentally wrong, it is strictly meteorological with no anthropogenic input. There is little doubt that in the scenario stated above, with significantly modified rainfall distribution, rural populations and farmers will adopt a new behaviour to adapt to the new climate conditions and as a consequence, so will be the biomass sources of air pollution.

These papers reach conclusions without ever even mentioning this important confounding factor. Also essential is that the validation protocol for this model is clearly insufficient and flawed, and is designed to only provide a very good match with the very short amount of historical data. Finally, these papers suffers from a “drama bias” on historical data which is known to be factually wrong; I personally would not reach many conclusions out of these few papers as it feels closer to a garbage-in – garbage-out model than anything else.