Spatial Assessment of Air Pollution Index Using Environ Metric Modeling Techniques
The quest for industrial and urban development over the years has changed the pattern and source of atmospheric air pollution in Malaysia. This study aims to investigate the spatial variation in the source of air pollution; identify the most significant parameterscontributing to the air pollution...
Saved in:
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
2014
|
Subjects: | |
Online Access: | http://eprints.unisza.edu.my/4948/1/FH02-ESERI-16-04958.pdf http://eprints.unisza.edu.my/4948/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | The quest for industrial and urban development over the years has changed the pattern
and source of atmospheric air pollution in Malaysia. This study aims to investigate the
spatial variation in the source of air pollution; identify the most significant
parameterscontributing to the air pollution and to develop the best receptor model for
predicting air pollution index (API). Data from five monitoring stations base on five
year's observation (2007-2011) were used. Multivariate techniques such as cluster
analysis (HACA), discriminate analysis (DA), principal component analysis (PCA),
factor analysis (FA) and modeling techniques comprising of artificial neural network
(ANN) and multiple linear regression (MLR) were used in this study. HACA was able
to group the five monitoring stations into three clusters, indicating that one station in
each cluster can provide a reasonable accurate spatial assessment of air quality within
the study area. The result for standard mode, forward stepwise and backward stepwise
DA gave a correct assignation of 82.37% (p˂ 0.05) which indicate that all the
parameters significantly discriminate spatially. PCA and FA for the three clusters
account for more than 62%, 56% and 58% of the total variance respectively indicating
that the source of air pollution are from anthropogenic induced point source and nonpoint
source. ANN gave a better prediction at R2 = 0.93 and RMSE = 4.87 compared
with MLR AR2 = 0.70 and RMSE = 9.57. This indicates that ANN can be applied with
more precision in modeling and understanding the dynamic nature of API compared
with MLR. However, it can be concluded that the atmosphere exhibit a complex
reaction which is made more complicated by man through his daily activities and
multivariate analysis, modeling techniques will help to understand the dynamic
structure of the source of pollution, reduce the cost of monitoring unnecessary sampling
site, save time and accurately predict air pollution index. These findings can provide
stakeholders a basic for understanding and developing air pollution control strategies in
order to reduce their possible health and environmental effect. |
---|