are exposure hot spots due to the concentrations of
both pollution sources and people, and 96% of
people in large cities are exposed to pollution
levels above recommended limits. There are
over three million deaths each year associated
with poor outdoor air quality [93]. Urban air quality is highly dependent on emissions of, for example, fine and ultrafine PM and NO 2 from road
vehicles, which also contribute to global air
pollution.
A study by Carslaw et al. (2013) [94] reports
measurements of NO, NO 2 , and NH 3 from almost
70,000 vehicles (taxies, buses, and passenger
cars) from 1985 to 2012 in the UK made using a
remote sensing technique. In this study, emissions
are recorded depending on specific vehicle technologies that impact emissions performance. It is
possible then to compare different vehicles with
respect to emissions, for example, the results for
TfL buses and diesel cars show important variations in NO 2 emissions depending on the vehicle
manufacturer. By analyzing trends in NO x and
NO 2 at 23 roadside sites in London, the study
found that total NO x and NO 2 emissions have
only decreased by about 1 and 0.5% annually,
respectively, over the past 10 years. The trends
are in agreement with the previously determined
trends between 2004 and 2009 measured by
Carslaw et al. (2011) [95], which suggest that
NO x and NO 2 concentrations are decreasing
more slowly than indicated by emission inventory
estimates.
In addition, new vehicle after-treatment technologies have increased the ratio of NO 2 to NO
compared with older vehicles, in particular for
increased engine loads, indicating a challenge in
emissions source management [94].
Low-cost portable sensors are promising for
personal exposure monitoring, although such sensors not yet have achieved their full market potential. Many researchers are working on improved
wearable sensors. Piedrahita et al. [17] developed
low-cost, wearable, air quality monitors (M-Pods)
using the Arduino platform. The M-Pods have two
types of sensors – commercially available metal
oxide semiconductor (MOx) sensors for NO 2 , CO,
O 3 , and VOCs, and NDIR sensors for CO 2 . The
MOx sensors show high sensitivity and are low in
cost; however, they are cross-sensitive and display
nonlinear output signals [17]. The model performance has been evaluated, and the relationship
between reference data and collocation-calibrated
sensor readings displays a minor negative bias at
the higher end of observed concentration levels,
possibly driven by the small number of data points.
The approach gives access to air quality monitoring to a wider group of scientists and citizens;
however, there are some uncertainties regarding
the actual personal exposure levels due to the
level of user engagement [30].
Low-cost pollution monitors have been placed
at many stationary locations and are also
employed with mobile platforms. Installations in
the field are necessary to monitor concentrations
of the key pollutants. One example of a field
installation is the Airlabs AirNode-T (#3 and
#4), developed by the company Airlabs in Copenhagen. The AirNode is a low-cost and low-power
self-contained air quality monitoring platform
measuring PM 2.5 , PM 10 , NO 2 , O 3 , and CO. The
two AirNodes are mounted at DCMR’s Schiedam
air quality monitoring station in Rotterdam, NL,
and they have been running onsite since
December 19, 2018. A field validation of the
AirNodes shows the uncorrected readings from
AirNodes #3 and #4 compared to reference measurements at the air quality station over a 40-day
period. The reference data was collected by the
station operated by the DCMR Environmental Protection Agency. The station monitors an array of air
pollutants including PM 2.5 , PM 10 , NO 2 , and O 3 .
The uncorrected readings from AirNodes #3
and #4 have been plotted against the references
and R
2 for the different plots are seen on the
graph. The R
2 values show that AirNodes #3 and
#4 are most accurate toward O 3 and least with
PM 2.5 (Fig. 12).
Calibration
The sensors were calibrated by three different
methods whose performance was then compared.
The three methods were linear regression, multivariate linear regression, and an artificial neural
network.
Low-Cost Sensors for Indoor and Outdoor Pollution
445
both pollution sources and people, and 96% of
people in large cities are exposed to pollution
levels above recommended limits. There are
over three million deaths each year associated
with poor outdoor air quality [93]. Urban air quality is highly dependent on emissions of, for example, fine and ultrafine PM and NO 2 from road
vehicles, which also contribute to global air
pollution.
A study by Carslaw et al. (2013) [94] reports
measurements of NO, NO 2 , and NH 3 from almost
70,000 vehicles (taxies, buses, and passenger
cars) from 1985 to 2012 in the UK made using a
remote sensing technique. In this study, emissions
are recorded depending on specific vehicle technologies that impact emissions performance. It is
possible then to compare different vehicles with
respect to emissions, for example, the results for
TfL buses and diesel cars show important variations in NO 2 emissions depending on the vehicle
manufacturer. By analyzing trends in NO x and
NO 2 at 23 roadside sites in London, the study
found that total NO x and NO 2 emissions have
only decreased by about 1 and 0.5% annually,
respectively, over the past 10 years. The trends
are in agreement with the previously determined
trends between 2004 and 2009 measured by
Carslaw et al. (2011) [95], which suggest that
NO x and NO 2 concentrations are decreasing
more slowly than indicated by emission inventory
estimates.
In addition, new vehicle after-treatment technologies have increased the ratio of NO 2 to NO
compared with older vehicles, in particular for
increased engine loads, indicating a challenge in
emissions source management [94].
Low-cost portable sensors are promising for
personal exposure monitoring, although such sensors not yet have achieved their full market potential. Many researchers are working on improved
wearable sensors. Piedrahita et al. [17] developed
low-cost, wearable, air quality monitors (M-Pods)
using the Arduino platform. The M-Pods have two
types of sensors – commercially available metal
oxide semiconductor (MOx) sensors for NO 2 , CO,
O 3 , and VOCs, and NDIR sensors for CO 2 . The
MOx sensors show high sensitivity and are low in
cost; however, they are cross-sensitive and display
nonlinear output signals [17]. The model performance has been evaluated, and the relationship
between reference data and collocation-calibrated
sensor readings displays a minor negative bias at
the higher end of observed concentration levels,
possibly driven by the small number of data points.
The approach gives access to air quality monitoring to a wider group of scientists and citizens;
however, there are some uncertainties regarding
the actual personal exposure levels due to the
level of user engagement [30].
Low-cost pollution monitors have been placed
at many stationary locations and are also
employed with mobile platforms. Installations in
the field are necessary to monitor concentrations
of the key pollutants. One example of a field
installation is the Airlabs AirNode-T (#3 and
#4), developed by the company Airlabs in Copenhagen. The AirNode is a low-cost and low-power
self-contained air quality monitoring platform
measuring PM 2.5 , PM 10 , NO 2 , O 3 , and CO. The
two AirNodes are mounted at DCMR’s Schiedam
air quality monitoring station in Rotterdam, NL,
and they have been running onsite since
December 19, 2018. A field validation of the
AirNodes shows the uncorrected readings from
AirNodes #3 and #4 compared to reference measurements at the air quality station over a 40-day
period. The reference data was collected by the
station operated by the DCMR Environmental Protection Agency. The station monitors an array of air
pollutants including PM 2.5 , PM 10 , NO 2 , and O 3 .
The uncorrected readings from AirNodes #3
and #4 have been plotted against the references
and R
2 for the different plots are seen on the
graph. The R
2 values show that AirNodes #3 and
#4 are most accurate toward O 3 and least with
PM 2.5 (Fig. 12).
Calibration
The sensors were calibrated by three different
methods whose performance was then compared.
The three methods were linear regression, multivariate linear regression, and an artificial neural
network.
Low-Cost Sensors for Indoor and Outdoor Pollution
445