Air Quality Monitoring and Advanced Bayesian Modeling introduces recent developments in urban air quality monitoring and forecasting. The book presents concepts, theories, and case studies related to monitoring methods of criteria air pollutants, advanced methods for real-time characterization of chemical composition of PM and VOCs, and emerging strategies for air quality monitoring. The book illustrates concepts and theories through case studies about the development of common statistical air quality forecasting models. Readers will also learn advanced topics such as the Bayesian model class selection, adaptive forecasting model development with Kalman filter, and the Bayesian model averaging of multiple adaptive forecasting models.
Author(s): Yongjie Li, Ka In Hoi, Kai Meng Mok, Ka Veng Yuen
Publisher: Elsevier
Year: 2023
Language: English
Pages: 316
City: Amsterdam
Air Quality Monitoring and Advanced Bayesian Modeling
Copyright
Introduction
Clean versus polluted air
Sources and impacts of air pollutants
Air quality monitoring strategies
Modeling and forecasting of air pollution
About this book
References
Current air quality monitoring methods
Methods for criteria air pollutants
Carbon monoxide (CO)
Sulfur dioxide (SO2)
Nitrogen oxides (NO and NO2)
Ozone (O3)
Particulate matters (PM10 and PM2.5)
Real-time chemical composition monitoring
Particulate matters
Mass spectrometry for real-time PM measurement
Mass spectrometry based on electron impact (EI)
Mass spectrometry based on laser ionization desorption (LDI)
Ion chromatography for real-time PM measurement
Ion chromatographic systems for particles only
Ion chromatographic systems for gases and particles
Real-time measurement of trace elements in PM
Volatile organic compounds
Gas chromatography for real-time VOC measurement
Mass spectrometry for real-time VOC measurement
Other real-time techniques
Optical techniques for real-time measurements of gases
Thermal and optical techniques for real-time measurements of PM
Conclusions
References
Emerging air quality monitoring methods
Low-cost sensors
Electrochemical sensors
Metal oxide sensors
Optical sensors for PM
Sensors for VOCs
New considerations for low-cost sensors
Analytical merits
Potential interferences
Lab calibrations and field comparisons
Data correction
Data transmission and sensor networks
Mobile measurement platforms
On-road air quality monitoring
Powered and nonfixed-route vehicles
Powered and fixed-route vehicles
Nonpowered and nonfixed-route platforms
Requirements on monitoring method and data analysis
Air-borne air quality monitoring
Balloon-borne measurements
Manned-aircraft measurements
Unmanned-aircraft measurements
Other mobile measurement platforms
Conclusions
References
Traditional statistical air quality forecasting methods
Multiple linear regression (MLR)
Overview
Basics of multiple linear regression
Ridge regression and LASSO
Example: Estimation of AR(2) parameters with the multiple linear regression, the ridge regression, and the LASSO r ...
Classification and regression tree (CART)
Overview
Regression tree
Classification tree
Bagging and random forests
Example: Estimation of CO2 emissions from vehicle features with random forest
Multilayer perceptron
Overview
Basics of multilayer perceptron
Training algorithm of MLP
Example: Imputation of missing air quality data based on multilayer perceptron
Support vector regression (SVR)
Overview
Formulation of support vector regression
Case study
Overview
Prediction of PM2.5 and ground-level O3 concentrations of Macau
References
Advanced Bayesian air quality forecasting methods
Overview of technique limitations and advanced topics for improvement
Choice of model complexity
Necessity of model adaptiveness
Bayesian model class selection of linear regression model
Overview
Basics of Bayesian model class selection in linear regression model
Modeling of Keeling curve
Kalman filter-based adaptive air quality model
Overview
Basics of Kalman filter-based adaptive air quality model
Selection of perturbation matrix and measurement noise variance
Revisiting example 5.2.3 (modeling of Keeling curve) with the adaptive linear model
Time-varying multilayer perceptron
Overview
Basics of time-varying multilayer perceptron
Example: Prediction of Mackey-Glass time series by using the TVMLP model
Adaptive Bayesian model averaging of multiple time-varying regression models
Overview
Basics of dynamic Bayesian model averaging
Modeling of measured PM2.5 concentration of the low-cost sensor
Case study
Overview
Air quality forecasting in Macau with the adaptive linear models
References
Index
