=Paper=
{{Paper
|id=Vol-3006/59_short_paper
|storemode=property
|title=Methods for calculating the average particulate matter concentrations in the Krasnoyarsk city ground layer
|pdfUrl=https://ceur-ws.org/Vol-3006/59_short_paper.pdf
|volume=Vol-3006
|authors=Maxim I. Malimonov,Alexander A. Pushkarev,Oxana V. Sokolova
}}
==Methods for calculating the average particulate matter concentrations in the Krasnoyarsk city ground layer==
https://ceur-ws.org/Vol-3006/59_short_paper.pdf
Methods for calculating the average particulate matter
concentrations in the Krasnoyarsk city ground layer
Maxim I. Malimonov1 , Alexander A. Pushkarev2 and Oxana V. Sokolova1
1
Federal Research Center Krasnoyarsk Science Center SB RAS, Krasnoyarsk, Russia
2
Institute of Computational Modelling SB RAS, Krasnoyarsk, Russia
Abstract
The paper considers a method based on the Voronoi diagram for calculating the average particulate
matter concentration in the surface layer of the Krasnoyarsk city air environment. We have used two
methods to achieve this goal. The first method relied on buffer zones built in the immediate vicinity
around monitoring posts. The second one uses the boundaries of the city of Krasnoyarsk.
Keywords
JavaScript, JSON, air pollution, PM2.5.
1. Introduction
Analysis of the results of monitoring the state of atmospheric air, determining the level of air
pollution in the city as a whole is the subject of numerous discussions. At present, the level
of atmospheric pollution at time intervals from 20 minutes to twenty-four hours researchers
estimate by the value of the maximum value of the parameter recorded for the selected interval.
This approach does not evaluate the average concentration of the measured impurity for the city
as a whole. The paper considers a method for calculating the average concentration of suspended
particles in the surface layer of the Krasnoyarsk air environment based on the Voronoi diagram.
The average value is calculated as follows. According to the original point data, a Voronoi
diagram is constructed. After that, boundary conditions are imposed on the diagram, which
is used as the contour of the city territory. Next, the weighted average is calculated, where
the weight is the area of the Voronoi polygons. The application of such a calculation for the
analysis of data from the air monitoring system of the Krasnoyarsk Scientific Center of the
SB RAS is demonstrated.
1.1. Data processing
This paper discusses an approach for calculating the average concentration of suspended
particles using the Voronoi diagram. The data for the work were taken from the environmental
monitoring network of the Federal Research Center of the KSC SB RAS. There is software and
hardware for collecting and storing data to store all the information [1].
SDM-2021: All-Russian conference, August 24โ27, 2021, Novosibirsk, Russia
" malimonov.mi@ksc.krasn.ru (M. I. Malimonov); pushkarev@icm.krasn.ru (A. A. Pushkarev)
ยฉ 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073
CEUR Workshop Proceedings (CEUR-WS.org)
499
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
The average daily data for the whole month is taken for the analysis. Visualization subsystem
environmental data based on Javascript libraries is used to display the data [2]. The dline.js
library is responsible for displaying polygons and contours [3].
2. Results and Discussion
The first step for analyzing the data is to select the period with the most noticeable contamination.
These periods are February 2019, November 2020 and from the end of December 2020 to the
beginning of January 2021. The unfavorable meteorological conditions (UMC) regime was
announced from 19:00 February 8, 2019, to 19:00 February 13, 2019 [4]. Next time the UMC
was announced from November 27 to November 30. In the third selected period, the UMC was
announced from December 26, 2020, to January 2, 2021.
The average value is calculated as follows. Interpolation is built using Voronoi diagrams
based on the initial point data, and this makes it easy to divide the city into some areas of
influence of each monitoring station. Voronoi diagrams are built using the turf.js library, which
provides many different functions for processing spatial data in JavaScript
After building Voronoi diagrams, the layer is cropped along the required contour. In this
work, two variants of the competition were used: the first one is along the buffer zones around
each monitoring station with a radius of 3 km; the second one is along the administrative
boundaries of the city of Krasnoyarsk. The buffer zone is the area of influence of one monitoring
post. Competing is carried out using the npm package polygon-clipping, which allows one
to perform operations to trim polygons, merge polygons or subtract the boundaries of one
polygon from another.
The next step is to calculate the area of each resulting polygon, which is also carried out
using the truf.js library. The result is a GeoJSON file containing a set of polygons, the area
polygon, and the value of the point around which each Voronoi polygon was built.
Next, the average pollution value for the entire city is calculated. The calculation is made
according to the arithmetic weighted average formula (1), where the weight is the area of each
Voronoi polygon.
๐
โ๏ธ
๐ค๐ ยท ๐ฅ ๐
๐=1
๐ฅ
ยฏ= ๐ , (1)
โ๏ธ
๐ค๐
๐=1
where ๐ค๐ is the area of each polygon obtained as a result of constructing Voronoi diagrams, ๐ฅ๐
is the value of a point that hits this polygon.
The second approach uses not buffer zones but the boundaries of the city of Krasnoyarsk.
Figure 1 shows the result of building buffer zones around monitoring posts. This figure shows
the peak of pollution on February 12, 2019.
Figure 2 shows the construction of a Voronoi diagram within the boundaries of the city of
Krasnoyarsk for the same period. Monitoring posts outside the city boundaries are taken into
account when constructing a pollution map.
Figure 3 displays the comparison of mean concentrations of suspended particles Card PM2.5
in February 2019, obtained by using different approaches.
500
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
Figure 1: Buffer zones separated by a Voronoi diagram.
Figure 2: The city of Krasnoyarsk, divided by the Voronoi diagram.
Figure 3: Comparison of average concentrations in February 2019.
501
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
The most significant differences begin at high concentrations. If we consider the average
concentrations together with the maximum and minimum concentrations, then some differences
say that the cityโs pollution is highly uneven. One can also replace that averages and highs
show the same trend, as opposed to lows.
Fifteen posts in the city of Krasnoyarsk were used for the previous calculations, located in
different districts of Krasnoyarsk, and their number is sufficient to cover most of the city. In
some areas, there are several monitoring posts at once. For the following calculations, some
posts were removed from the districts of the city. Figure 4 shows the result of this construction.
In this case, other results were obtained, which show that this number of posts is enough
to calculate average concentrations and display a trend, but in this case, some monitoring
posts pull over a large share of pollution. In Figures 1 and 2, there were yellow zones in the
Akademgorodok microdistrict, located in the Oktyabrsky district. We removed posts from this
microdistrict and set up one monitoring post located in the Nikolaevka microdistrict in the
Figure 4: Buffer zones built with the exclusion of some posts and separated by a Voronoi diagram.
Figure 5: Krasnoyarsk city borders, separated by Voronoi diagram, except for some posts.
502
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
Oktyabrsky district. The minimum values shifted quite strongly towards the average values,
and the differences between the average values of the different approaches became smaller but
increased in value. Figure 6 shows the calculation results without taking into account several
monitoring posts.
These results show that the placement of one post per district is not enough. The city has
a rather tricky terrain, and monitoring posts located in neighboring micro-districts can show
completely different values, which differ tenfold.
Figure 7 shows the results of November 2020. The data below is calculated using all monitoring
posts. Tables 1 and 2 show the results of comparing the average values obtained by different
methods.
Based on the results shown in the tables, we can conclude that the difference between the
โArithmetic meanโ and โBuffer zonesโ is not significant. The difference between Borders and
Arithmetic mean is more tangible.
Figure 6: Comparison of average concentrations in February 2019 (๐g/m3 ) excluding some monitoring
stations.
Figure 7: Comparison of average concentrations in November 2020 taking into account all monitoring
stations.
503
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
Table 1
Two-sample ๐ก-test for November 2020. Comparison of โArithmetic meanโ and โBuffer zonesโ values.
Arithmetic mean Buffer zones
Average 28.33333333 27.16666667
Variance 934.0229885 834.7643678
Observations 30 30
Combined variance 884.3936782
Hypothetical difference of averages 0
df 58
๐ก-statistics 0.151939119
๐ (๐ โค ๐ก) one-sided 0.439881121
๐ก critical one-sided 1.671552762
๐ (๐ โค ๐ก) two-sided 0.879762243
๐ก critical two-sided 2.001717484
Table 2
Two-sample ๐ก-test for November 2020. Comparison of โArithmetic meanโ and โBoundariesโ values.
Arithmetic mean Buffer zones
Average 28.33333333 23.4
Variance 934.0229885 699.0068966
Observations 30 30
Combined variance 816.5149425
Hypothetical difference of averages 0
df 58
๐ก-statistics 0.668657964
๐ (๐ โค ๐ก) one-sided 0.25318317
๐ก critical one-sided 1.671552762
๐ (๐ โค ๐ก) two-sided 0.50636634
๐ก critical two-sided 2.001717484
Figure 8 shows the results of the calculation, in which several monitoring posts were also
removed. It can be seen that the minimum values have not changed much compared to Figure 7.
That is because the post in Akademgorodok showed high values, and even with its exclusion,
there were no significant changes. The values at this post are dependent on the wind. If the
wind is blowing to the east, this post shows shallow values compared to the rest. During the
UMC in November, the wind blew mainly to the north and at a low speed.
In the examples above, it is noticeable that the maximum values are not changed as when
considering all the posts as without taking into account several monitoring posts. This is due to
the fact that the maximum values are usually shown by the posts located in Solontsy or the city
center. The posts in the city center show different results, but not significantly different from
each other. Even if we take one monitoring post from the center, it will not change the overall
picture. There is only one monitoring post in Solontsy, and it was not ruled out.
Figure 9 shows the results for the third selected period. In this case, the UMC affects two
504
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
Figure 8: Comparison of average concentrations for November 2020 excluding several monitoring
stations.
Figure 9: Comparison of average concentrations for November 2020 excluding several monitoring
stations.
months at once, and we decided to take a thorough month, starting from December 15 to
January 15.
Figure 9 shows that the minimum values correlate with the average values only on the days
of the UMC regime. On other days, the wind was more than 3 m/s, and Akademgorodok shows
the minimum values that are cut off from the cityโs values.
3. Conclusions
These results lead to several conclusions. 15โ16 monitoring posts are enough to calculate average
concentrations in the city, but one can lose the detail of the spread of pollution if reducing their
number. The cases considered show that the wind has a relatively strong effect on the spread of
pollution. The rugged terrain of the city requires the installation of many monitoring posts to
display a high-quality picture. Even the exclusion of neighboring monitoring posts significantly
changes the minimum values. Different methods for calculating average concentrations do not
differ dramatically in values.
505
Maxim I. Malimonov et al. CEUR Workshop Proceedings 499โ506
The subsequent work will explore how to place monitoring posts in the city to reflect a
quality picture and how many monitoring posts are needed to achieve this goal.
References
[1] A. Tokarev, Software and technology for collecting and storage serial observa-
tions of the environment, InterCarto. InterGIS 20 (2014) 241โ249. doi:10.24057/
2414-9179-2014-1-20-241-249.
[2] M. Malimonov, Software tools for interactive visualization of environmental data, in:
Open Conference of Young Scientists of the ICM SB RAS on Mathematical Modeling and
Information Technologies, ICM SB RAS, Krasnoyarsk, 2019, pp. 30โ32.
[3] A. Pushkarev, Formation of isolines on a regular data in the web environment, in: Open Con-
ference of Young Scientists of the ICM SB RAS on Mathematical Modeling and Information
Technologies, ICM SB RAS, Krasnoyarsk, 2019, pp. 44โ46.
[4] O. Yakubailik, V. Zavoruev, M. Malimonov, A. Pushkarev, Spatial analysis of air pollution in
Krasnoyarsk, in: 2019 All-Russian Conference "Spatial Data Processing for Monitoring of
Natural and Anthropogenic Processes", SDM 2019, Berdsk, CEUR Workshop Proceedings,
volume 2534, 2020, pp. 491โ496.
506