Performance assessment of air quality monitoring networks using principal component analysis and cluster analysis

doi:10.1016/j.buildenv.2010.09.004

Building and Environment

Volume 46, Issue 3, March 2011, Pages 577-583

https://doi.org/10.1016/j.buildenv.2010.09.004 Get rights and content

Abstract

This study aims to evaluate the performance of two statistical methods, principal component analysis and cluster analysis, for the management of air quality monitoring network of Hong Kong and the reduction of associated expenses. The specific objectives include: (i) to identify city areas with similar air pollution behavior; and (ii) to locate emission sources. The statistical methods were applied to the mass concentrations of sulphur dioxide (SO₂), respirable suspended particulates (RSP) and nitrogen dioxide (NO₂), collected in monitoring network of Hong Kong from January 2001 to December 2007.

The results demonstrate that, for each pollutant, the monitoring stations are grouped into different classes based on their air pollution behaviors. The monitoring stations located in nearby area are characterized by the same specific air pollution characteristics and suggested with an effective management of air quality monitoring system. The redundant equipments should be transferred to other monitoring stations for allowing further enlargement of the monitored area. Additionally, the existence of different air pollution behaviors in the monitoring network is explained by the variability of wind directions across the region. The results imply that the air quality problem in Hong Kong is not only a local problem mainly from street-level pollutions, but also a region problem from the Pearl River Delta region.

Introduction

Hong Kong is one of the most developed metropolitans and characterized by the highest population and traffic densities in the world. With the continuous economic development and the population expansion, several environmental problems especially air quality problem, have become severe and attracted much attention in recent years [2], [3], [5], [6], [7], [10], [20]. To control and improve air quality, Hong Kong Environmental Protection Department (HKEPD) has established fourteen monitoring stations to monitor and manage. These stations are distributed in different area of Hong Kong and equipped with various instruments for detecting corresponding pollutants. Some of them are located closely in geography or characterized by similar monitoring condition, which generally present similar behaviors. It probably results in an inefficient usage of the resource and extra expense cost. Hence, it is necessary and significant to optimize the air quality monitoring network using the practical alternative methods such as principal components analysis (PCA) and cluster analysis (CA).

PCA is a statistical technique that transforms the original set of inter-correlated variables into a new set of an equal number of independent uncorrelated variables or principal components that have linear combinations to the original variables. The multicollinearity which probably implied between original variables can be removed through application of it [16], [17]. On the other hand, CA is a classification method used to split a data set into a number of groups of observations which are distinct in terms of typical group values of the variables [11], [12], [13]. The aim of it is to maximize between-group variance and to minimize within-group variance.

In previous studies, these two approaches are combined to explore significant information from the origin data. Gramsch et al. [4] used the PCA and CA to determine the seasonal trends and spatial distribution of PM₁₀ and O₃ in Santiago de Chile, concluding that the city had four large sectors with dissimilar air pollution behaviors. Shah and Shaheen [14] employed them to identify the major source of airborne trace metals in area of Islamabad. Pires et al. [12], [13] applied them in the mass concentration of SO₂, PM₁₀, CO, NO₂ and O₃ for the efficient management of air quality in Oporto Metropolitan Area. They all verified that these two methods could complement each other and the combination of them could provide a practical alternative approach for analyzing and solving environmental problem.

In this study, PCA approach was employed as well as CA for optimizing and managing the air quality monitoring stations in Hong Kong. The aim of this paper was to assess the performance of PCA and CA for the management of SO₂, RSP and NO₂ mass concentrations monitoring with the following specific objectives: (i) to identify city areas with similar air pollution behavior; and (ii) to locate emission sources.

Section snippets

Sample location and available data

Air quality in Hong Kong is greatly affected by the vehicle density on roads, coupled with the hilly geography and cavernous streets. For a better management of air quality in Hong Kong, fourteen monitoring stations have been established by HKEPD since 1999. They are generally classified into two categories in term of the experimental location, roadside stations, i.e., Causeway Bay (CB), Central (CT), Mong Kok (MK), and general stations, i.e., Central/Western (CW), Sham Shui Po (SS), Eastern

Results and discussion

Table 2 shows the main results of the PCA application at all stations for pollutants of SO₂, RSP and NO₂. From the table, it can be seen that the first three PCs explain 69.1% of the original data variance for analyzing the SO₂ concentrations, followed by PC4, PC5 and PC6 components with cumulative variance of 75.4%, 80.0% and 83.7% respectively. In the first PC (PC1), the stations of KC, TW, and ST seem to present similar characteristics and they are combined to take up much loading on

Comments and suggestions

Aiming at the identification of city areas with similar pollution behaviors in Hong Kong, two statistical methods, PCA and CA, were applied on three pollutants at fourteen stations. The grouped clusters for SO₂ and RSP pollutants were evidently dependent on the geography while that for the NO₂ pollutant was dependent on the monitoring condition. The identification of city areas with the same specific pollutant behaviors at many stations suggests an ineffective management of the AQMN. The

Acknowledgement

The work presented in this paper was partially supported by research grants from City University of Hong Kong, HKSAR, China (Projects No. CityU-SRG 7002370) and the National Natural Science Foundation of China (Grant Nos 10662002, 10672098 and 10532020).

References (21)

S.A. Abdul-Wahab et al.
Principal component and multiple regression analysis in modelling of ground-level ozone and factors affecting its concentrations
Environ Modell Softw
(2005)
P. Berghmans et al.
Exposure assessment of a cyclist to PM₁₀ and ultrafine particles
Sci Total Environ
(2009)
E. Gramsch et al.
Examination of pollution trends in Santiago de Chile with cluster analysis of PM₁₀ and ozone data
Atmos Environ
(2006)
W.Z. Lu et al.
Investigation of respirable suspended particulate trend and relevant environmental factors in Hong Kong downtown areas
Chemosphere
(2008)
L. Paoletti et al.
Physico-chemical characterisation of the inhalable particulate matter (PM₁₀) in an urban area: an analysis of the seasonal trend
Sci Total Environ
(2002)
J.C.M. Pires et al.
Management of air quality monitoring using principal component and cluster analysis—Part I: SO₂ and PM₁₀
Atmos Environ
(2008)
J.C.M. Pires et al.
Management of air quality monitoring using principal component and cluster analysis—Part II: CO, NO₂ and O₃
Atmos Environ
(2008)
B. Škrbić et al.
Principal component analysis for soil contamination with organochlorine compounds
Chemosphere
(2007)
Y. Song et al.
Source apportionment of PM2.5 in Beijing using principal component analysis/absolute principal component scores and UNMIX
Sci Total Environ
(2006)
S.I.V. Sousa et al.
Multiple linear regression and artificial neural networks based on principal components to predict ozone concentrations
Environ Modell Softw
(2007)

There are more references available in the full text version of this article.

Cited by (97)

Interval-valued functional clustering based on the improved Euclidean distance with application to air quality index
2023, Applied Mathematical Modelling
Interval function clustering is a statistical method used to classify functional data based on interval number similarity measurements. However, existing similarity measurements focus on measuring the similarity of the curves in terms of numerical distance. This ignores the changing characteristics of the curve shape, which may lead to unreasonable clustering results when clustering interval-valued functional data. To address this issue, an improved Euclidean-distance-based interval-valued functional clustering method is proposed in this study. By deducing a specific calculation formula for the Euclidean distance of the interval function under the basis function and derivative information, the absolute difference in the numerical value was reflected based on the distance of the basis function, and the curve shape difference was reflected based on the distance of the derivative function information. Furthermore, a similarity measurement method combining numerical distance and curve shape was constructed to reflect the change characteristics of the functional data more completely and improve the K-means clustering process. Finally, the air quality index of different cities is taken as an example to cluster, and the differences and variation characteristics of different types of air quality are analyzed, which verifies the effectiveness of the proposed method.
Multi-site, multi-pollutant atmospheric data analysis using Riemannian geometry
2023, Science of the Total Environment
We demonstrate the benefits of using Riemannian geometry in the analysis of multi-site, multi-pollutant atmospheric monitoring data. Our approach uses covariance matrices to encode spatio-temporal variability and correlations of multiple pollutants at different sites and times. A key property of covariance matrices is that they lie on a Riemannian manifold and one can exploit this property to facilitate dimensionality reduction, outlier detection, and spatial interpolation. Specifically, the transformation of data using Reimannian geometry provides a better data surface for interpolation and assessment of outliers compared to traditional data analysis tools that assume Euclidean geometry. We demonstrate the utility of using Riemannian geometry by analyzing a full year of atmospheric monitoring data collected from 34 monitoring stations in Beijing, China.
Comprehensive satisfaction assessment of thermal and light environment in a primary school classroom in Qingdao: A field study
2023, Journal of Building Engineering
Various factors, such as operative temperature and illuminance, can affect primary school students' satisfaction with their classroom environment. This study investigated these environmental parameters in a primary school classroom in Qingdao and assessed primary school students' subjective satisfaction. Additionally, a comprehensive satisfaction evaluation method was developed that considers the characteristics of the winter climate. The study first obtained a dimensionless function of each environmental factor through field test. The weights of heat and light in the model were then investigated and counted, and a comprehensive satisfaction assessment model for primary classrooms was proposed and verified. The results indicated that primary school students were more satisfied with the thermal environment of the classroom than with the light environment. The comprehensive satisfaction was higher when the operative temperature was between 14–23 °C, and the illuminance was between 300–550 lux. However, the comprehensive satisfaction was slightly lower when the operative temperature was between 10–14 °C or 23–27 °C, and the illuminance was between 550–700 lux. This model provides a framework for studying the comprehensive effects of multiple factors on primary school students' satisfaction with their classroom environment. Additionally, it offers a theoretical basis and design reference for creating a more comfortable classroom environment.
Using the big data analysis and basic information from lecture Halls to predict air change rate
2023, Journal of Building Engineering
School lecture halls are often designed as confined spaces. During the period of COVID-19, indoor ventilation has played an even more important role. Considering the economic reasons and the immediacy of the effect, the natural ventilation mechanism becomes the primary issue to be evaluated. However, the commonly used CO₂ tracer gas concentration decay method consumes a lot of time and cost. To evaluate the ventilation rate fast and effectively, we use the common methods of big data analysis - Principal Component Analysis (PCA), K-means and linear regression to analyze the basic information of the lecture hall to explore the relation between variables and air change rate. The analysis results show that the target 37 lecture halls are divided into two clusters, and the measured 11 lecture halls contributed 64.65%. When analyzing the two clusters separately, there is a linear relation between the opening area and air change rate (ACH), and the model error is between 6% and 12%, which proves the feasibility of the basic information of the lecture hall by calculating the air change rate.
Comparison of different multicriteria decision-making methodologies for sustainability decision making
2023, Hydrogen Economy: Processes, Supply Chain, Life Cycle Analysis and Energy Transition for Sustainability
This chapter aims to compare different weighting methodologies and multicriteria decision-making methodologies for sustainability decision making. It mainly introduces and compares four weighting methodologies and four multicriteria decision-making methodologies. The weighting methodologies comprise two objective methodologies (i.e., entropy method and ideal point method) and two subjective methodologies (i.e., analytic hierarchy process and Delphi method). The multicriteria decision-making methodologies consist of four options, namely, technique for order of preference by similarity to an ideal solution, data envelopment analysis, preference ranking organization method for enrichment evaluation, and principal component analysis. The proposed weighting methodologies and multicriteria decision-making methodologies have been compared by applying them to sustainability decision making among five energy scenarios.
Hybridization of hierarchical clustering with persistent homology in assessing haze episodes between air quality monitoring stations
2022, Journal of Environmental Management
Haze has been a major issue afflicting Southeast Asian countries, including Malaysia, for the past few decades. Hierarchical agglomerative cluster analysis (HACA) is commonly used to evaluate the spatial behavior between areas in which pollutants interact. Typically, using HACA, the Euclidean distance acts as the dissimilarity measure and air quality monitoring stations are grouped according to this measure, thus revealing the most polluted areas. In this study, a framework for the hybridization of the HACA technique is proposed by considering the topological similarity (Wasserstein distance) between stations to evaluate the spatial patterns of the affected areas by haze episodes. For this, a tool in the topological data analysis (TDA), namely, persistent homology, is used to extract essential topological features hidden in the dataset. The performance of the proposed method is compared with that of traditional HACA and evaluated based on its ability to categorize areas according to the exceedance level of the particulate matter ( ${PM}_{10}$ ). Results show that additional topological features have yielded better accuracy compared to without the case that does not consider topological features. The cluster validity indices are computed to verify the results, and the proposed method outperforms the traditional method, suggesting a practical alternative approach for assessing the similarity in air pollution behaviors based on topological characterizations.

View all citing articles on Scopus

View full text

Performance assessment of air quality monitoring networks using principal component analysis and cluster analysis

Abstract

Introduction

Section snippets

Sample location and available data

Results and discussion

Comments and suggestions

Acknowledgement

Environ Modell Softw

Sci Total Environ

Atmos Environ

Chemosphere

Sci Total Environ

Atmos Environ

Atmos Environ

Chemosphere

Sci Total Environ

Environ Modell Softw