as major roads with heavy traffic and industrial and residential sites. The recorded
measurements at these stations or locations can only quantify the air pollution level
in that particular region or nearby small areas. In order to estimate the concentrations
of the air pollutants at unmonitored areas using the available measurements, a variety
of methods, including spatial averaging, nearest neighbour, inverse distance
weighting (IDW), kriging, land-use regression (LUR) modelling, dispersion modelling, neural networks, etc., have been proposed by researchers (Xingzhe et al. 2017).
Several study designs have opted for the use of a single methodology to estimate
the concentrations of air pollutants (Ritz et al. 2000; Ritz et al. 2007; Miller et al.
2007; Brauer et al. 2008; Henderson et al. 2007; Beckerman et al. 2012), while some
have attempted a combination of several methods for pollution estimation at different geographical scales (Brunekreef et al. 2009; Beelen et al. 2007).
With the rapid development of pollution-detecting sensors and global positioning
system (GPS) technology over the recent decades, mobile sensing systems with
devices integrated with these systems have been extensively applied in urban areas.
Hence this leads to an increase in the intensity of pollutant measurements, but does
not discount the sparse measurements since the coverage of the mobile sensing
systems mainly explore specific areas or provide roads wise information (Xingzhe
et al. 2017).
The regular measurement of precise ground level PM 2.5 concentrations is of great
significance to epidemiological studies whereby it provides valuable information for
an effective management and forecasting of air quality (Al-Saadi et al. 2005). Air
quality monitoring networks have been established in many industrial countries to
measure pollutant concentration at different locations on a daily or hourly basis. The
temporal resolution and accuracy of air quality data vary depending on the methods
of measurement (Cyrys et al. 2001). Point measurements collected at monitoring
units are not necessary representative of regional concentrations. The regional
variability of air pollutants is difficult to assess from point measurements alone
due to various other impending factors (van Donkelaar et al. 2010).
Ground-based measurements are generally considered to be accurate and they
provide measurements for relatively small areas around the point stations. Often, the
limited spatial coverage and irregular distribution of ground-based monitoring
stations largely restrict the study on space–time dynamics of air pollution and its
impact on human health and the environment (Tian and Chen 2010). Application of
satellite data to observe the surface air quality has advanced considerably (Martin
2008; Hoff and Christopher 2009). In fact, global aerosol observations from satellite
remote sensing has proven to substantially improve estimates of population that are
exposed to PM 2.5 .
Regression models were developed by researchers to find whether any correlation
exists between MODIS-derived AOD products with ground-measured PM 2.5
(Diao et al. 2019). North American Land Data Assimilation System (NLDAS)
data was also integrated to account for the meteorological variables while estimating
the ground-level PM 2.5 from AOD. It was observed that the relationships varied for
different AOD products and also for different regions in the USA.
18 Geospatial Modelling of Air Pollution and its Impact on Health of Urban. . .
379
measurements at these stations or locations can only quantify the air pollution level
in that particular region or nearby small areas. In order to estimate the concentrations
of the air pollutants at unmonitored areas using the available measurements, a variety
of methods, including spatial averaging, nearest neighbour, inverse distance
weighting (IDW), kriging, land-use regression (LUR) modelling, dispersion modelling, neural networks, etc., have been proposed by researchers (Xingzhe et al. 2017).
Several study designs have opted for the use of a single methodology to estimate
the concentrations of air pollutants (Ritz et al. 2000; Ritz et al. 2007; Miller et al.
2007; Brauer et al. 2008; Henderson et al. 2007; Beckerman et al. 2012), while some
have attempted a combination of several methods for pollution estimation at different geographical scales (Brunekreef et al. 2009; Beelen et al. 2007).
With the rapid development of pollution-detecting sensors and global positioning
system (GPS) technology over the recent decades, mobile sensing systems with
devices integrated with these systems have been extensively applied in urban areas.
Hence this leads to an increase in the intensity of pollutant measurements, but does
not discount the sparse measurements since the coverage of the mobile sensing
systems mainly explore specific areas or provide roads wise information (Xingzhe
et al. 2017).
The regular measurement of precise ground level PM 2.5 concentrations is of great
significance to epidemiological studies whereby it provides valuable information for
an effective management and forecasting of air quality (Al-Saadi et al. 2005). Air
quality monitoring networks have been established in many industrial countries to
measure pollutant concentration at different locations on a daily or hourly basis. The
temporal resolution and accuracy of air quality data vary depending on the methods
of measurement (Cyrys et al. 2001). Point measurements collected at monitoring
units are not necessary representative of regional concentrations. The regional
variability of air pollutants is difficult to assess from point measurements alone
due to various other impending factors (van Donkelaar et al. 2010).
Ground-based measurements are generally considered to be accurate and they
provide measurements for relatively small areas around the point stations. Often, the
limited spatial coverage and irregular distribution of ground-based monitoring
stations largely restrict the study on space–time dynamics of air pollution and its
impact on human health and the environment (Tian and Chen 2010). Application of
satellite data to observe the surface air quality has advanced considerably (Martin
2008; Hoff and Christopher 2009). In fact, global aerosol observations from satellite
remote sensing has proven to substantially improve estimates of population that are
exposed to PM 2.5 .
Regression models were developed by researchers to find whether any correlation
exists between MODIS-derived AOD products with ground-measured PM 2.5
(Diao et al. 2019). North American Land Data Assimilation System (NLDAS)
data was also integrated to account for the meteorological variables while estimating
the ground-level PM 2.5 from AOD. It was observed that the relationships varied for
different AOD products and also for different regions in the USA.
18 Geospatial Modelling of Air Pollution and its Impact on Health of Urban. . .
379