Using AI to Map Urban Heat in Thailand

Thailand’s rapid urban growth is changing how heat is stored and released across cities. Dense buildings, dark roofs, paved roads and limited vegetation can create urban heat islands, where built-up districts remain warmer than surrounding rural areas, especially after sunset. The effect increases discomfort, electricity demand and health risks during prolonged hot weather.

Satellite imagery offers a practical way to see these patterns across large areas. Thermal sensors can estimate land surface temperature, while optical imagery identifies trees, water, bare soil, rooftops and sealed surfaces. When artificial intelligence combines these layers with weather, population and infrastructure data, city authorities can target heat reduction measures more precisely.

For an Australian audience, the issue is familiar. Residents in western Sydney, Melbourne’s growth corridors and parts of Brisbane already experience hotter neighbourhoods linked to low canopy cover and extensive hard surfaces. Air conditioning is a normal response during heatwaves, yet higher household energy use can add pressure to the electricity system and increase costs.

Thailand presents a valuable regional setting for shared digital solutions. Through an Asia-Pacific development platform such as ICTD-ASP, public agencies, universities, technology firms and development partners could use geospatial AI to improve public services, strengthen climate resilience and direct investment towards communities most exposed to extreme heat.

Why Thailand’s Cities Need Finer Heat Intelligence

Bangkok’s dense central districts, industrial areas and expanding suburbs do not experience heat in the same way. High-rise development, traffic, air-conditioning exhaust and limited open space can produce localised hotspots. In Chiang Mai, seasonal smoke, dry conditions and surrounding topography add different pressures, while cities such as Khon Kaen face rapid expansion with fewer established cooling networks.

A citywide average can conceal these differences. A neighbourhood with mature trees may be several degrees cooler at street level than a nearby zone dominated by warehouses and concrete. Mapping heat at a fine spatial scale helps planners prioritise schools, hospitals, public housing, markets and transport corridors where exposure is greatest.

What Satellite Imagery Can Reveal

Thermal satellite missions such as Landsat can estimate land surface temperature at a useful urban scale. Sentinel imagery contributes information about vegetation, impervious surfaces and water, although its standard optical sensors do not directly measure thermal conditions. Repeated observations can show how heat changes across seasons, development stages and land-use types.

The strongest analysis combines satellite observations with local information. Weather-station readings, building density, tree-canopy surveys, road networks, electricity demand and census indicators can distinguish a hot industrial estate from a hot residential area. Mobile or fixed sensors can then validate satellite estimates, especially where shade and building height make ground-level conditions different from surface temperature.

Where Artificial Intelligence Adds Value

Machine-learning models can classify land cover, identify roofs and roads, fill gaps caused by cloud cover and estimate heat patterns between satellite passes. Deep-learning systems can also detect changes in tree cover or construction, allowing authorities to monitor whether new development is increasing thermal exposure.

AI should support expert judgement rather than replace it. A model trained on one Thai city may perform poorly in another because of different roof materials, building forms, rainfall patterns or vegetation. Transparent methods, local calibration and uncertainty scores are essential when maps influence public spending or planning approvals.

The data pipeline also needs safeguards. Satellite imagery at neighbourhood scale is generally environmental information, but combining it with mobile-phone movement data or household records can create privacy risks. Australian partners would recognise the importance of privacy-by-design under the Privacy Act 1988, while Thai projects should apply clear rules for data access, retention, security and public disclosure.

Turning Heat Maps into Practical Measures

Heat-risk maps are most useful when they lead to specific interventions. Authorities could prioritise shade trees along walking routes, reflective or insulated roofs, shaded bus stops, cool pavements, pocket parks, restored canals and redesigned public squares. Green roofs and water-sensitive urban design may offer additional benefits where maintenance and water availability are manageable.

Everyday habits shape the outcome. In Thailand, outdoor markets, motorcycle travel and long waits for public transport can expose people to heat at street level. A cooling programme should therefore consider shade at market entrances, drinking-water points, rest areas and safer routes to schools and transit stations, rather than focusing only on parks.

Australian cities offer relevant examples and cautions. Shade structures at playgrounds, tree planting around train stations and cool-roof trials in western Sydney can inform Thai projects, but imported designs must suit monsoon rain, tropical vegetation, local construction practices and maintenance budgets. A material that performs well in a dry Australian suburb may weather differently in Bangkok.

Building Partnerships Across the Region

A regional programme could connect Thailand’s urban authorities with GISTDA, universities, meteorological agencies, local communities and private geospatial providers. Development partners could help fund open data infrastructure, training and pilot projects, while telecommunications companies could support sensor connectivity and public alerts.

ICTD-ASP is well placed to convene this mix of expertise. Its focus on ICT project development, investment partnerships, knowledge sharing and capacity building fits a programme that needs more than a single software purchase. A practical pilot might begin with Bangkok and one fast-growing regional city, then publish methods and reusable tools for other municipalities.

Australian organisations can contribute through climate analytics, urban forestry, engineering, cloud services and public-sector innovation. The local market already includes firms working in remote sensing, digital twins, asset management and climate-risk modelling. Collaboration should include Thai researchers and city staff from the beginning so that skills, models and maintenance capacity remain in-country.

Measuring Results and Financing Scale

A credible project needs a baseline before construction begins. Indicators might include land surface temperature, shaded walking area, canopy cover, cooling-centre access, heat-related ambulance calls, school attendance and electricity demand during hot periods. Results should be compared with similar untreated areas where possible, while recognising that weather varies from year to year.

Financing can combine municipal budgets, development finance, climate funds, research grants and private investment. Building owners may support cool roofs when energy savings are clear, while transport agencies can invest in shade where passenger comfort and network use improve. Australia’s National Construction Code and state planning systems show how building performance requirements can complement voluntary upgrades, although Thai rules would need to reflect local conditions and enforcement capacity.

Approach	Useful evidence	Likely urban action	Main limitation
Thermal satellite imagery	Surface temperature and seasonal hotspots	Prioritise cooling investments	Cloud cover and limited street-level detail
Optical imagery	Trees, roofs, roads and water	Track canopy and land-cover change	Does not directly measure experienced air temperature
AI prediction models	Heat patterns between observations	Forecast risk and test scenarios	Requires local training data and validation
Ground sensors	Air temperature, humidity and shade conditions	Verify priority locations	Small coverage area and ongoing maintenance
Community reporting	Perceived heat, access barriers and local priorities	Improve design and equity	Responses may be uneven across neighbourhoods

The most effective projects will link satellite analysis with decisions that residents can feel: a cooler walk to a bus stop, a shaded market, a safer school route or lower indoor temperatures. For Australia, Thailand’s experience can inform heat adaptation across rapidly growing Asia-Pacific cities, while Australian expertise can strengthen data governance and project evaluation.

A practical first step is to select two contrasting Thai urban districts, assemble satellite and local datasets, validate the model with ground sensors and co-design a short list of affordable cooling actions. Measure the results through the next hot season, publish the method openly and use the evidence to direct the next investment.