More focus in recent years has been given to the urban climate in fast-growing mid-sized cities in South Asia, rather than focusing on mega cities as has been done in the past. Proven approaches include the integrated Land use-Mobility-Energy-Environmental approach in urban development (e.g., Sikder et al., 2018; Alipour and Dia, 2023). Poor urban land use planning leads to ambient air pollution (e.g., ozone and nitrogen oxide), which further affects urban heat islands (UHI) (e.g., Sarrat et al., 2006; Li et al., 2018; Ulpiani, 2021). In fact, urban pollution islands (UPI) and UHI are two major concerns for the urban environment that exist simultaneously (e.g., Agarwal and Tandon, 2010; Ulpiani, 2021).

The World Health Organization (WHO) has stated that nearly the entire global population (99%) is exposed to air that exceeds the WHO’s permissible air pollution standards (WHO, 2021), and the highest suffering is experienced in low and middle-income countries (WHO, 2024a). Since 1987, WHO has been providing health-based air quality (AQ) guidelines (AQG) for governments and society to reduce human exposure to air pollution (WHO, 2021), which are related to Sustainable Development Goals (SDGs) targets 3.9.1, 7.1.2, and 11.6.2 (WHO, 2024b). In 2019, air pollution was the fourth leading risk factor for premature death (4.2 million) globally, with 89% occurring in low-and middle-income countries, and the greatest risk experienced in Southeast Asia and the Western Pacific Regions (HEI, 2020; WHO, 2022). These deaths were associated with air pollution-related ischemic heart disease and stroke (37%), chronic obstructive pulmonary disease (18%), acute lower respiratory infections (23%), and cancer within the respiratory tract (11%) (WHO, 2022). In addition, it has long-term effects, for example lower labor productivity and slowing plant development and agricultural productivity (Kapoor et al., 2024a).

Particulate Matter (PM) is the major component of air pollutants and is comprised of sulphate, nitrates, ammonia, sodium chloride, black carbon, mineral dust, and water particles (WHO, 2024c). The health risks associated with PM_2.5 (diameter ≤ 2.5 μm) and PM₁₀ (diameter ≤ 10 μm) are well documented due to their ability to penetrate deep into the lungs and the bloodstream (WHO, 2024c). A 1% increase in exposure to PM_2.5 over WHO’s AQG (5 μg/m³ annual average and 15 μg/m³ 24-h average) can cause an approximately 12.8% increase in breathing difficulties, 12.5% increase in wet coughs, and 8.1% higher risk of lower respiratory tract infection (World Bank, 2022). The cleanest city in Bangladesh, Sylhet (Figure 1), has a PM_2.5 level ⁓10 times higher than the WHO’s AQG (Islam et al., 2020).

In Bangladesh, 78,145–88,229 deaths in 2019 were caused by air pollution, costing the country 3.9–4.4% of GDP (World Bank, 2022). Poor ambient air pollution puts people at risk of breathing difficulties, cough, lower respiratory tract infections, depression, and other health conditions (World Bank, 2022). During 2018–2021, Bangladesh and its capital city, Dhaka (Figure 1), were ranked the most polluted country and second most polluted city in the world, respectively. Reasons for air pollution, e.g., for Dhaka city, are listed as smoke from brickfield kilns (58%), exhaust from transport vehicles (⁓10%), road dust (⁓8%), soil dust (⁓8%), biomass burning (⁓7%), sea salt (⁓1%), and dust from construction sites due to rapid urbanization (Begum et al., 2013). This city’s air pollution rate is increasing 10% per year.

A significant number of studies have examined ambient AQ and public health in Bangladesh’s major cities, including Dhaka and Chittagong (see Table 1). These cities have predominantly developed organically, though certain areas have undergone targeted, pocket-based development due to planning principles. This study aims to examine a mid-sized fast-growing city, where findings may offer significant insights for urban planners and policymakers. The major objectives of this study are (1) to estimate the population exposure under different levels of PM_2.5 and PM₁₀ set by local regulatory authorities using high spatial resolution in-situ observation and (2) to identify possible sources of air pollution (excluding quantification). This study also explores the relationship between PM_2.5 and PM₁₀ with other environmental and demographic factors and seasonal variation. An empirical case study has been conducted in a mid-sized city in Bangladesh named Mymensingh (Figure 1).

A semi-systematic (e.g., Snyder, 2019) literature review approach was adopted and widely used to identify the research gaps and scope (Snyder, 2019; Zunder, 2021). Google Scholar, ScienceDirect, and Web of Science were used to search the documents. The keywords searched for were “particulate matter pollution,” “air pollution trends,” “indoor and outdoor air pollution,” “impact of particulate matter on human health,” “particulate matter concentration in urban areas,” “sources of particulate matter pollutants,” and “seasonal variation of particulate matter.” Bangladesh-related peer-reviewed articles were considered for in-depth research analysis, and concepts and theories from international literature were synthesized. Backward and forward citation searching was also applied (e.g., Haddaway et al., 2022). A total of 30 documents related to ambient air pollution, PMs, potential sources, and impact on public health in Bangladesh were reviewed qualitatively, and the most recent literatures were documented in a tabular format (Table 1).

Ambient air pollution in Bangladesh is a serious issue, and thousands of people die every year from causes related to it (World Bank, 2022). However, detailed research is minimal compared to other global southern countries. Most of the research was dependent on the Department of Environment’s (DoE) AQ Index (AQI) (mainly PM_2.5), although they have very few permanent monitoring stations and remotely sensed processed data with very coarse spatial resolution (DoE, 2024). Very few studies have been conducted on in-situ observation of air pollutants with very coarse spatial and temporal resolution.

The impact of air pollution on public health was linked to stakeholders’ perceptions by using questionnaires, surveys, or interviews. One study conducted a pathological investigation using blood samples, but it has not been completed yet (Haque et al., 2024). However, the impacts of air pollutants, e.g., PM and nitrogen oxides [NOx = nitric oxide (NO) + nitrogen dioxide (NO₂)] among others (e.g., total suspended particles (TSP), potentially toxic elements (PTE), sulfur dioxide (SO₂), carbon monoxide (CO), carbon dioxide (CO₂), and ozone (O₃)), on public health are well established. Therefore, many researchers estimated the number of exposed people to the level of polluted air.

Recent studies analyzed in-situ observations with high spatial resolution but poor temporal resolution (Hossain et al., 2023). There is scope for detailed analysis to take advantage of high spatial resolution in-situ observation. Analysis of the additional data with high temporal resolution from DoE (2024) might offer insights into the ongoing air pollution conditions and exposed populations. This study used those opportunities. A summary of the most recent studies is presented in Table 1.

The level of PM_2.5 in Sylhet, the cleanest city in the country, is 9.7 times higher than the World Health Organization (WHO)-accepted maximum levels (5 μg/m³ annual mean and 15 μg/m³ 24-h mean) (WHO, 2021). Similarly, in the study area, no location met the AQ standards defined by the Department of Environment (DoE, 2024). In fact, the PM levels in the study area are more than 10 times higher than the WHO-recommended levels (WHO, 2021) (Figure 6; Supplementary Table B3). Approximately 98% of the population in the study area experienced unhealthy to extremely unhealthy air conditions for some hours each day (Supplementary Table B3). This reflects the general state of AQ in other cities in Bangladesh (Haque et al., 2024; Hassan M. S. et al., 2022; Jawaa et al., 2024; Khan M. W. et al., 2023; Saju et al., 2023).

In general, PM has a robust relationship with environmental factors (e.g., NDVI, DEM, slope, land surface temperature), meteorological factors (e.g., precipitation, wind speed, and direction), and economic factors (e.g., GDP, income level, poverty level). These findings also correspond to the work of other researchers (Hassan M. S. et al., 2022; Kulsum and Moniruzzaman, 2021). However, due to limited resources and available data sets, only NDVI, DEM, slope, density, and built-up % were considered to explore the relationship with PM. Additionally, the study area was small and relatively flat, limiting variations in land surface temperature, wind speed, and direction. In this study, PM correlated negatively with NDVI (Supplementary Tables B3, B4), similar to other findings (Hassan M. S. et al., 2022; Kulsum and Moniruzzaman, 2021). NDVI indicates green space and the health of green infrastructure (Islam, 2014). High NDVI may suggest more generation of biogenic volatile organic compounds, contributing to O₃ generation and PM_2.5 levels depending on tree species (Cai et al., 2024). Similarly, PM showed a negative relationship with DEM and slope (Supplementary Tables B3, B4), consistent with other studies (Hassan M. S. et al., 2022), despite limited variation in DEM and slope in the study area. PM is positively related to population density, which is also expected (Budde et al., 2024). Surprisingly, PM had a negative correlation with built-up % (Supplementary Tables B3, B4), contradicting other findings (Lin et al., 2013; Yuan et al., 2019; Halim et al., 2020). Typically, a higher built-up percentage indicates more urban activities and less green-blue space and green infrastructure, therefore generating more PM.

MOLSR considers a multivariable linear regression relationship, and PPC considers a one-to-one relationship. However, the relationship between a dependent and a set of independent variables may not be linear. For example, a nonlinear function may better explain PM’s relationship with NDVI (Figure 3) (Kulsum and Moniruzzaman, 2021). However, GWR is an extended version of MOLSR and considers stationary and local nonstationary (geographic) values of variables (Meik and Lawing, 2017). Therefore, GWR might be the superior method to chart a relationship between a dependent and a set of independent variables since, in geography, everything is related to everything else, but near things are more connected than distant things (Tobler, 1970) and geographical variables exhibit uncontrolled variance (Goodchild, 2004). The GWR model in this study probably gave better fits between dependent (PM) and independent variables (e.g., NDVI, DEM, population density). Coefficients obtained in GWR models tended to be lower and tended more towards zero than the MOLSR model, even though it generated a slightly higher standard error (Supplementary Tables B3, B4). However, use of other statistical and geostatistical models, e.g., structural (e.g., Bose et al., 2023; Roy et al., 2023), linear mixed-effect (e.g., Sajjad Abdollahpour et al., 2024; Venter et al., 2024), or linear mixed-effect with spatial correlation models (e.g., Halla-aho and Lähdesmäki, 2020), may change the amplitude of coefficients and the direction of relationships.

In terms of seasonal variation, PM_2.5 concentration showed a negative relationship with ambient temperature (Figure 8b), consistent with other urban environments in Bangladesh (Khan R. H. et al., 2023) but contrary to findings in the USA (Jhun et al., 2015). During the summer, PM_2.5 concentration and temperature exhibited a positive relationship, while in the winter, their relationship was negative (Figure 8a). Precipitation is the strongest factor in reducing ambient PM_2.5 concentration in Bangladesh (Hassan M. S. et al., 2022) and other parts of the world (Zalakeviciute et al., 2018; Liu et al., 2020). Except for in very hot summers, the climate in Bangladesh is influenced by the monsoon, resulting in heavy rainfall and relatively high temperatures (Figure 4; Supplementary Table B5). Therefore, precipitation may have a more significant impact than temperature on controlling ambient air pollution in the study area and country. In winter, high PM_2.5 concentrations are observed at low temperatures (Figure 8b). Relative humidity is positively related to precipitation, equally affecting the reduction of ambient PM_2.5 in the study area (Figure 8d). However, relative humidity might have an inverse impact on ambient PM_2.5 in traffic environments, as found in Ecuador (Zalakeviciute et al., 2018).

Regarding PM sources, no differences were identified when comparing other urban environments in Bangladesh. Brick production, vehicular emission, biomass burning, improper waste management, and construction activities were found in the study area that have been shown to substantially increase environmental stress (Begum et al., 2013; Begum and Hopke, 2019; Kumar et al., 2024). Brick kilns are widespread in the landscape and are a significant contributor to PM generation in the study area (Figure 1). PM emissions occur due to the combustion of solid fuels during brick-making processes (Ahmad et al., 2022). Vehicular emissions are now a major source of PMs due to the increasing combustion of fossil fuels caused by high traffic volumes (Kumar et al., 2021). The burning of biomass materials (e.g., wood burning for domestic cooking and, in commercial areas, domestic and communal waste burning) releases considerable amounts of PMs into the atmosphere (Figure 8) (Johnston et al., 2019). Ongoing construction and demolition operations substantially contribute to PM concentration, particularly in areas experiencing rapid urbanization and infrastructure development (Muleski et al., 2005). Dust particles released during excavation, material handling, and other construction processes aggravate the problem of declining AQ and lead to localized pollution hotspots (Cheriyan and Choi, 2020). Moreover, Central Business District (CBD) areas, characterized by a large population density and a wide range of human activities, are hotspots for elevated PM concentrations (Menon and Nagendra, 2018). Dust accumulation from vehicular traffic, coupled with industrial activity, leads to elevated PM concentrations in CBD areas. Modelling might help in quantifying ambient PM_2.5 concentration with high temporal and spatial resolution (e.g., Suri et al., 2023; Kapoor et al., 2024b). However, not all PMs are locally generated. PMs can stay in the atmosphere for a long period (e.g., hours for PM₁₀, weeks for PM_2.5, and even longer for ultra-fine particles) and can travel thousands of km in dry conditions (Ali et al., 2019; Kumar et al., 2024; Pima County, 2024). The lifespan and traveling distance of PMs are highly dependent on particle size, aerodynamics (e.g., wind pressure, speed, and direction), and meteorological conditions (e.g., temperature, precipitation, and humidity). Further, climate change impacts trigger higher PM concentrations, affecting public health (Jacob and Winner, 2009; Doherty et al., 2017; Pienkosz et al., 2019).

Ambient air pollution in urban environments largely varies in relation to city size in countries around the world. Generally, urban air pollution increases in relation to city size; therefore, AQ in mid-sized cities is much better than in large cities and/or megacities in China and South Asia (e.g., in India, Pakistan) (Liu et al., 2018; Tabinda et al., 2020). On the other hand, the AQ in large cities is better in Europe, North America, and Latin America (Han et al., 2016). Therefore, the relationship between urban AQ is determined by urban function and land use rather than city size alone. Even though our case is a mid-sized city, it is one of the fastest-growing cities in South Asia. Therefore, ambient AQ should be identical to the large cities and/or megacities in the country (e.g., Jawaa et al., 2024; Kulsum and Moniruzzaman, 2021; Islam et al., 2020).

The urban areas in Bangladesh have turned into contaminated gas chambers due to severe air pollution which has increased the UHI effect, posing a significant public health risk that we are currently grappling with. However, this is common for mid-sized and fast-growing cities in South Asia, e.g., in India (Roy and Singha, 2020, 2021), Pakistan (Anwar et al., 2021), and Sri Lanka (Ileperuma, 2020). Legislatively, AQ in Bangladesh is protected and promoted by several acts, rules, and regulations by the DoE (DoE, 2024). However, the reality is that air pollution is occurring in the study area and, in general, in urban areas in Bangladesh, similar to particularly fast-growing cities in South Asia. There are big gaps in policy execution in this regard, e.g., managing traffic effectively, controlling pollution-generating backdated unfit vehicles, and instating environmentally friendly brick production (Figure 8). Poor practice in plan execution is another example, e.g., Dhaka city’s green space reduced from 56% in 1989 to ~2% in 2020 (Khan M. W. et al., 2023). Many authors have suggested nature-based solutions, e.g., ensuring enough green infrastructure by promoting proper land use planning, which would have mitigate UPI (e.g., Ren et al., 2023; Wu and Chen, 2023) and UHI (e.g., Peng and Jim, 2015; Zardo et al., 2017), improve urban flood management (Jarden et al., 2016) and biodiversity and ecological services (e.g., Capotorti et al., 2019; Nakamura et al., 2020), and positively affect physical and mental health (e.g., Felappi et al., 2020; Moreira et al., 2022). However, studying such issues is out of the scope of this research. Therefore, immediate actions have to be taken to secure the interest of public health. Decisive, transparent, and accountable leadership, with strong collaboration and cooperation amongst public-private organizations and agencies, is a pre-requisite to governing the actions for controlling air pollution.

Data-driven urban public policy can help to promote a clean urban environment. This study adopted a geospatial approach to studying ambient AQ in a mid-sized Bangladeshi city. This empirical case study estimates the population’s exposure to air pollution using high spatial resolution empirical in-situ observations to identify possible sources of air pollution and explore a relationship with other environmental and demographic factors, including seasonal variation. Findings suggested that, even though the AQ in Bangladesh cities should be protected to ensure public health, it is being polluted due to increased anthropogenic factors that in turn are facilitated by poor land use planning. Snapshot measurements in the field campaign between 27 March and 02 April 2023 confirm that

The observations and data analysis on air pollution between 13 February 2023 to 10 April 2024 suggest that

Some avoidable limitations were not possible to overcome. The PM measurements were taken over seven consecutive days with three time slots each day, making it a snapshot measurement. Continuous measurements over a 2-h slot would likely provide a more representative dataset. Additionally, in-situ observation only reflects conditions during the dry winter season. For seasonal analysis, the study relied on observed data from a fixed station operated by the DoE.

Prevention of and protection from pollution sources is necessary. As mentioned earlier, the environment and, therefore, AQ in Bangladesh is protected by legislative acts and regulations, however, they need to be properly enforced. Many people do not know that little actions could protect from the worsening of AQ, e.g., burning residential and communal waste. Private, non-governmental, and community-based organizations could increase awareness in this regard. To control air pollution, the researchers suggested a package of activities for short-, mid-, and long-term measures (Hossain et al., 2023; Khan R. H. et al., 2023). However, many of these suggestions do not align with local experiences and knowledge, logistics capabilities, and cultural orientation. Intensive research is required for practical policy suggestions, their implications, and the realities posed by their implementation, which is out of the scope of this study.