Low-cost NO2 monitoring and predictions of urban exposure using universal kriging and land-use regression modelling in Mysore, India
Introduction
Air pollution exposure studies and assessments of the corresponding health outcomes are needed in Low- and Middle-Income Countries (LMICs) worldwide, but especially in heavily populated nations that are undergoing unprecedented urbanization (Allender et al., 2010). According to the World Health Organization (WHO), 98% of cities in low- and middle-income countries with more than 100,000 inhabitants do not meet WHO health-based air quality guidelines, as compared with 56% of similar cities in high-income countries (World Health Organization, 2019a). Rapid growth of population in cities, development of heavy industries, and intensification of road traffic present significant challenges to ambient air quality in urban areas of LMICs. In addition, many basic needs such as cooking, household heating, and transit take place in these urban areas without availability of clean fuels or alternative technologies, resulting in poor air quality (Krzyzanowsk et al., 1999).
India represents a LMIC setting that experiences several of these challenges to urban air quality. Air pollution likely differs in Indian cities compared with Western cities due to variations in traffic pollution, which can be caused by fuel differences, technology differences, the types of vehicles on the roads, and the presence of heavily trafficked roadways (O'Neill et al., 2003). The results of several studies further agree that in future, the demand for transportation will exceed improvements to emissions reduction technologies (Vasconcellos and Urban Transport, 2013; Delucchi, 2000). Despite regulatory interventions, higher exposure to traffic pollution with distinct within-city gradients around major roads and highways has been shown in studies of US and European urban areas, suggesting higher exposure differentials within cities than between cities (Zhu et al., 2002; Gilbert et al., 2005). Further, the presence of heavy industry within urban areas of India is likely to impact air quality and the heterogeneity of urban air pollution (Khillare et al., 2004; Reddy et al., 1993). Additionally, the spatial gradient of air pollution may vary due to multiple industrial point sources of pollution. Compared to Western metropolitan areas, cities in India likely have diverse built structures, topography, weather and land use patterns that may influence the spatial distribution of air pollution. Actual exposure may also vary through distinctions in lifestyle and culture that affect indoor/outdoor activity patterns, as well as housing characteristics.
Understanding the variation of pollution that occurs within urban environments is important to improve the estimation of air pollution exposure, and thereby improve our knowledge of the health effects of air pollution. Many methods of estimating air pollution concentrations (including atmospheric dispersion models and satellite data-based models) are very data intensive, and the input data that are required to successfully build and implement those models may not be readily available in low-resource settings (Jerrett et al., 2005). Additionally, current satellite-based data on air pollution at the surface level for India are not highly spatially resolved, and may lead to exposure misclassification by missing the fine-scale changes or seasonal patterns in the distribution of air pollution exposure in Indian cities (Somvanshi et al., 2019). Studies using simpler methods of exposure, such as distance to the nearest major roadway or the data collected by a local regulatory monitor as a proxy of exposure at the individual's location, have potential for exposure misclassification and do not incorporate other variables that may influence exposure, such as local point sources, small industry, and highly trafficked markets (Jerrett et al., 2005).
The development of models to assess air pollution exposures within cities for health studies has been identified as a priority area for future research (Brunekreef and Holgate, 2002; Brauer et al., 2003). Locally tailored approaches for estimating heterogeneous air pollution exposures within LMICs are needed, as methods used for studies in Western countries with denser monitoring networks may not be appropriate or feasible. Multiple methods exist to address intraurban variation of air pollutants. Traditionally, urban studies estimate air pollution exposure by assigning values from the nearest available monitor or averaging values from multiple nearby monitors, an approach that uses readily available data from regulatory monitors and provides large spatial and temporal coverage (Ritz et al., 2000; Marshall et al., 2008; Bell, 2006). This approach either requires a large network of monitors, and/or relies on the assumption that the pollutant of interest is homogeneous within urban areas. However, such extensive monitoring networks may not exist in all areas of interest in India, and previous studies (Gurung et al., 2017; Weissert et al., 2018) suggest that there may be spatial heterogeneity within small areas in urban cities of developed countries as well as LMICs such as India, leading to high potential for exposure misclassification using traditional methods.
Alternative approaches have made use of low-cost, spatially dense monitoring networks for collection of field samples of health relevant gaseous pollutants (such as nitrogen dioxide, NO2), as well as spatial analysis tools such as Geographical Information Systems (GIS), to estimate exposures to pollutants at high spatial resolution using multiple regression models (Wheeler et al., 2008), interpolation (e.g. kriging (Mulholland et al., 1998; Jerrett et al., 2001; Pikhart et al., 2001; Finkelstein et al., 2003; Mercer et al., 2011; Young et al., 2014; Adam-Poupart et al., 2014), inverse distance weighting (Marshall et al., 2008; Abbey et al., 1999)), dispersion modeling (Panasevich et al., 2009), and LUR (Gurung et al., 2017; Jerrett et al., 2005; Finkelstein et al., 2003; Mercer et al., 2011; Young et al., 2014) models.
The objectives of this study were to characterize seasonal intra-urban variation of NO2 in Mysore, Karnataka State, India, in order to facilitate our study of the long-term respiratory health effects of NO2 exposure in the city. We collected seasonal field samples of NO2 utilizing multiple sampling technologies. We applied LUR modeling and universal kriging to the collected field data to build seasonal and annual average models of NO2 air pollution levels throughout the city, with a focus on explaining the influence of important covariates on the spatial relationships in the levels of NO2 in the city. The methods were designed to address issues central to India and similar LMICs, including limited data availability and high intra-urban spatial variability of air pollution. NO2 was selected as the pollutant of interest given the potential for adverse health outcomes associated with short-term exposure to NO2, including airway inflammation in healthy people (Blomberg et al., 1997) and increased respiratory symptoms in people with underlying health conditions (Weinmayr et al., 2010). Exposure to NO2 at high concentrations is of particular concern among sensitive populations, including children, the elderly, and people with chronic respiratory ailments (Weinmayr et al., 2010; Annesi-Maesano et al., 2003). Additionally, NO2 has high feasibility and reliability of field measurements, as well as relatively low cost of equipment and sampling.
Section snippets
Study area
The study area covers the urban center of Mysore (excluding Nanjanguda, a nearby city that is considered part of the Mysore Conurbation), Karnataka State, India. Fig. 1 illustrates the location and major features of the city of Mysore. Mysore is located in the foothills of the Chamundi Hills, 145 km (90 mi) to the southwest of Bangalore. The extent of the urban area, which is divided into 63 wards for administrative purposes, is characterized by the Ring Road, which circumnavigates the city,
Results
During sampling, loss was due primarily to curiosity of passersby, who frequently removed monitors from sampling location to check for useful parts and then tossed them on the ground nearby. Additionally, unusual weather patterns with strong rain lead to the losses of some monitors; infestation by spiders or insects was a third cause of sampler loss. Table 2 shows the timing of the seasonal air pollution sampling campaigns, as well as the monitor coverage (percentage of sites with completed
Discussion & conclusions
We developed and implemented a low-resource air quality monitoring campaign and used the collected data to develop LUR and kriging models of NO2 in Mysore. In general, our study demonstrates the practical application of spatial interpolation approaches, developed for cities in Europe and North America, to urban areas in LMICs. We further illustrate the differences and challenges inherent in applying these methods to Indian cities. Studies that aim to characterize the variability in air
CRediT authorship contribution statement
Amruta Nori-Sarma: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing - original draft, Writing - review & editing, Visualization, Project administration, Funding acquisition. Rajesh K. Thimmulappa: Methodology, Investigation, Resources, Writing - review & editing. G.V. Venkataramana: Methodology, Investigation, Resources, Writing - review & editing. Azis K. Fauzie: Investigation, Data curation. Sumit K. Dey: Investigation,
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors would like to acknowledge Dr. P. Barry Ryan (Emory University Rollins School of Public Health), Dr. Ana Rule (Johns Hopkins Bloomberg School of Public Health), and Timothy Green (Johns Hopkins Bloomberg School of Public Health) for their assistance with laboratory analysis.
References (60)
- et al.
Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe – the ESCAPE Project
Atmos. Environ.
(2013) The use of ambient air quality modeling to estimate individual and population exposure for human health research: a case study of ozone in the northern Georgia region of the United States
Environ. Int.
(2006)- et al.
Air pollution and health
Lancet
(2002) - et al.
A land use regression for predicting NO2 and PM10 concentrations in different seasons in Tianjin region, China
J. Environ. Sci.
(2010) - et al.
Modeling the intraurban variation in nitrogen dioxide in urban areas in Kathmandu Valley
Environ. Res.
(2017) - et al.
Spatial variation of PM 2.5 chemical species and source-apportioned mass concentrations in New York city
Atmos. Environ.
(2004) - et al.
Evaluation of land-use regression models used to predict air quality concentrations in an urban area
Atmos. Environ.
(2010) - et al.
Establishing an air pollution monitoring network for intra-urban population exposure assessment: a location-allocation approach
Atmos. Environ.
(2005) - et al.
Within-urban variability in ambient air pollution: comparison of estimation methods
Atmos. Environ.
(2008) - et al.
Comparing universal kriging and land-use regression for predicting concentrations of gaseous oxides of nitrogen (NOx) for the multi-ethnic study of atherosclerosis and air pollution (MESA air)
Atmos. Environ.
(2011)
A distance-decay variable selection strategy for land use regression modeling of ambient air pollution exposures
Sci. Total Environ.
Development of a microscale land use regression model for predicting NO2 concentrations at a heavy trafficked suburban area in auckland
NZ. Sci. Total Environ.
Intra-urban variability of air pollution in windsor, ontario-measurement and modeling for human exposure assessment
Environ. Res.
Development of land-use regression models for metals associated with airborne particulate matter in a North American city
Atmos. Environ.
Study of ultrafine particles near a major highway with heavy-duty diesel traffic
Atmos. Environ.
Long-term inhalable particles and other air pollutants related to mortality in nonsmokers
Am. J. Respir. Crit. Care Med.
Spatiotemporal modeling of ozone levels in quebec (Canada): a comparison of kriging, land-use regression (LUR), and combined bayesian maximum entropy–LUR approaches
Environ. Health Perspect.
Estimation of outdoor NOx, NO2, and BTEX exposure in a cohort of pregnant women using land use regression modeling
Environ. Sci. Technol.
An assessment of air pollution and its attributable mortality in ulaanbaatar, Mongolia
Air Qual. Atmos. Heal.
Level of urbanization and noncommunicable disease risk factors in Tamil nadu, India
Bull. World Health Organ.
Subpopulations at increased risk of adverse health outcomes from air pollution
Eur. Respir. J.
Natural cause mortality and long-term exposure to particle components: an analysis of 19 European cohorts within the multi-center ESCAPE Project
Environ. Health Perspect.
The inflammatory effects of 2 ppm NO 2 on the airways of healthy subjects
Am. J. Respir. Crit. Care Med.
Estimating long-term average particulate air pollution concentrations: application of traffic indicators and geographic information Systems
Epidemiology
Environmental measurement and modelling
Population enumeration data (final population)
A study on modeling nitrogen dioxide concentrations using land-use regression and conventionally used exposure assessment methods
Environ. Res. Lett.
Intra-urban spatial variability in wintertime street-level concentrations of multiple combustion-related air pollutants: the New York city community air survey (NYCCAS)
J. Expo. Sci. Environ. Epidemiol.
Environmental externalities of motor vehicle use in the US
J. Transp. Policy
Development of land use regression models for nitrogen dioxide, ultrafine particles, lung deposited surface area, and four other markers of particulate matter pollution in the Swiss SAPALDIA regions
Environ. Heal.
Cited by (13)
Application of land use regression model to assess outdoor air pollution exposure: A review
2023, Environmental AdvancesNeighborhood-scale ambient NO<inf>2</inf> concentrations using TROPOMI NO<inf>2</inf> data: Applications for spatially comprehensive exposure assessment
2023, Science of the Total EnvironmentCitation Excerpt :These emerging sources may require comprehensive regulatory surveillance due to the health risks of NO2 and further O3 (NO2 as a precursor gas of O3). Previous studies have reported an association between ambient NO2 concentrations and population density (Nori-Sarma et al., 2020; Hoek et al., 2015; Hoek et al., 2008). This spatial association has not been recognized to indicate emissions from specific source types.
Integration of GIS, improved entropy and improved catastrophe methods for evaluating suitable locations for well drilling in arid and semi-arid plains
2021, Ecological IndicatorsCitation Excerpt :When the variable has spatial autocorrelation, kriging interpolation can be performed (Piccini et al., 2014). The spatial autocorrelation of the data can be expressed using the Moran's I Index (Nori-Sarma et al., 2020), which can be calculated using ArcGIS (v10.2) software (package, Spatial Statistics Tools; toolset, Global Moran's I). The tool computes the Moran's I Index value along with the z-score and p-value to further validate the Index statistically.
Deep learning-based downscaling of tropospheric nitrogen dioxide using ground-level and satellite observations
2021, Science of the Total EnvironmentCitation Excerpt :Ahmed et al. (2018) analyzed the spatiotemporal variability of air pollutants, SO2, NO2, and PM10, in Egypt from the air quality monitoring network, and employed the spatiotemporal kriging to interpolate the monthly averages. Nori-Sarma et al. (2020) monitored, modeled and interpolated nitrogen dioxide (NO2) in Mysore, a rapidly urbanizing city in India using two distinct models: land use regression (LUR) approach and universal kriging methods. They concluded that the influence of pollution factors (e.g., point sources) and highly localized characteristics of the urban environment (e.g., proximity to religious points of interest) are the major contributors to ambient air pollution levels.