Title Page

Abstract

Contents

Chapter 1. Introduction and outlines 25

1.1. Background and motivations 25

1.1.1. Backgrounds 25

1.1.2. Motivations 28

1.2. Outline 29

Part Ⅰ. An integrated estimation of polycyclic aromatic hydrocarbons (PAHs) environmental fate and the potential cancer risk 31

Chapter 2. Literature review of multimedia fugacity model 32

2.1. Introduction 32

2.2. Multimedia fugacity model 34

2.3. Fugacity approach-based chemical fate modelling software 36

2.4. Websites of fate model Tools 42

Chapter 3. Quantitative assessment of human health risks induced by vehicle exhaust PAHs (VEPAHs) 44

3.1. Introduction 44

3.2. Materials and methods 48

3.3. Results and discussion 64

3.4. Summary 76

Chapter 4. Multiple land-use fugacity model for the description of PAHs fate at a regional scale 77

4.1. Introduction 77

4.2. Materials and methods 81

4.3. Results and discussion 99

4.4. Conclusion 124

Part Ⅱ. Public perception of climate change and the awareness of climate change induced nature events 125

Chapter 5. Background and literature review of public perception on climate change 126

5.1. Researches on public perceptions regards climate change 126

5.2. Twitter 134

Chapter 6. Assessment of public perceptions of climate change in social media using big data-based machine learning techniques 144

6.1. Introduction 144

6.2. Materials and methods 147

6.3. Results and discussion 159

6.4. Summary 178

Chapter 7. Finding the relationships between public awareness of climate change and the consequential natural events 179

7.1. Introduction 179

7.2. Materials and methods 185

7.3. Results and discussion 199

7.4. Summary 225

Chapter 8. Environmental management of mitigating environmental pollution and climate change using fuzzy-SWOT analysis 227

8.1. Introduction 227

8.2. Linkage between climate change and environmental pollution 228

8.3. Methodology of fuzzy-SWOT analysis 233

8.4. SWOT views of environmental managements in environmental pollution and climate change strategies 238

8.5. Summary 245

Conclusion 247

References 252

Curriculum Vitae 282

Table 2.1. Four types of fugacity models. 36

Table 2.2. Summary of the notable application of the reviewed models. 43

Table 3.1. The motor vehicle populations from 1999 to 2015 of Zhengzhou city and the annual mileage of different types of vehicles (Zhengzhou Statistic Yearbooks). 50

Table 3.2. On-road emission factors of particle-phase PAHs with different types of vehicles. 51

Table 3.3. Physicochemical properties of 16 PAHs. 55

Table 3.4. The distribution information of each parameter in the BAP Level III model. 56

Table 3.5. The updated value of half-life degradation time and compartment depth of each VEPAH in air. 58

Table 3.6. The standard (according to the observed PAHs concentration from literature) and modeled VEPAHs concentration in air phase. 59

Table 3.7. The parameters of the ILCR model. 61

Table 3.8. Environmental compartment parameters of Zhengzhou downtown area. 63

Table 3.8. Polynomial regression parameters of sixteen VEPAHs. 69

Table 3.9. The mortality rate of respiratory diseases and lung cancer. 73

Table 4.1. The mass balance equations for each compartment in multiple land-use fugacity model. 83

Table 4.2. Equations of fugacity capacities 84

Table 4.3. Equations of transfer processes 84

Table 4.4. Model parameter definition of multiple land use fugacity model. 89

Table 4.5. Environmental parameter list used in multiple land use fugacity model. 89

Table 4.6. Emission factors of PAHs by different emission sources. 94

Table 4.7. Emission activity of PAHs by different emission sources from 2000 to 2017 (×10⁴ t) (Statistics Bureau Staff). 95

Table 4.8. Environmental compartment parameters of the Beijing downtown area and Tongzhou district. 97

Table 4.9. Maximum permissible concentration of each PAH in water (μg/L), sediment (mg/kg dw, sediment), and soil (mg/kg dw, soil).[이미지참조] 98

Table 4.10. Estimated total PAH concentrations in the study area. 108

Table 4.11. Comparison of the concentration and mass of PAHs accumulated in various media in 2010, 2015, and 2020. 114

Table 4.12. Overall concentration reduction of ∑₁₆PAHs during the COVID-19. 123

Table 5.1. Literature review of climate change perception. 132

Table 5.2. The description of tweets data extract by Tweepy. 142

Table 6.1. The data information of climate change related tweets. 148

Table 6.2. A example of Twitter text pre-processes. 150

Table 6.3. Example of positive, neutral, and negative sentiment of tweets. 152

Table 6.4. Summary of the text feature extraction methods. 155

Table 6.5. Summary of the classifiers used in the study. 157

Table 6.6. The class information of the labelled climate change tweets. 164

Table 6.7. The class information of the labelled sentiment tweets. 168

Table 6.8. The Pro opinion percentage data information in each month. 174

Table 6.9. The one-way ANOVA table of the Pro opinion percentage of climate change tweets. 175

Table 6.10. Multiple comparison tests using Tukey HSD post-hoc test with 95% confidence intervals of the Pro opinion percentage data. 175

Table 6.11. The N/P ratio data information in each month. 176

Table 6.12. The one-way ANOVA table of the N/P ratio of climate change tweets. 177

Table 6.13. Multiple comparison tests using Tukey HSD post-hoc test with 95% confidence intervals of N/P ratio data. 177

Table 7.1. Summary of the lexicon used in sentiment analysis. 184

Table 7.2. The Twitter datasets information. 188

Table 7.3. Influential factors to the consequence of climate change related natural events. 190

Table 7.4. The list of heat index equation parameters in Rothfusz regression equation. 192

Table 7.5. Summary of the lexical-based algorithms on text similarity analysis. 197

Table 7.6. The happened major natural events of each topic in the study period. 202

Table 7.7. The standardized regressing coefficients of natural events factor on the daily number of tweets in each state. Intercept is not shown. 215

Table 7.8. The multiple linear regression results of the daily number of climate change tweets in California. 217

Table 7.9. The standardized regressing coefficients of natural events factor on the daily net sentiment rate of climate change tweets in each state. Intercept is not shown. 220

Table 7.10. The multiple linear regression results of the daily net sentiment rate of climate change tweets in California. 222

Table 8.1. Transformation rules for linguistic variables. 237

Table 8.2. SWOT mix analysis strategies for environmental management of both climate change and environmental pollution. 243

Fig. 1.1. The dissertation outline. 30

Fig. 2.1. Evaluative environments (Mackay 2001). (a) Four compartments system. (b) Eight compartments system. 35

Fig. 2.2. The input data panel of the EQC model. (a) Chemical properties, (b) emission rates, and (c) environmental properties. 38

Fig. 2.3. Transport and transformation processes in the QWASI model. 40

Fig. 2.4. An illustration of mass-exchange processes modeled in the CalTOX. 42

Fig. 3.1. (a) Satellite-derived annual average surface-level PM₂.₅ concentration (μg/m³) of China in 2008 (NASA); (b) vehicle population per unit area (/km²) of china in 2008 (Chinese Statistic... 46

Fig. 3.2. Framework for predicting the environmental behavior and health risk of VEPAHs. 49

Fig. 3.3. The reported data and fitted curve of time-trend BAP emission rate based on fourth-order polynomial function. 62

Fig. 3.4. Location of the study area - Zhengzhou urban area. 63

Fig. 3.5. BaP concentration attributions in each compartment of air, water, soil, and sediment (bar chart) with the VEPs concentration proportion (pie graphs). 64

Fig. 3.6. A schematic representation of multimedia fugacity Level III model results. Steady-state mass balance diagram of BAP in air, water, soil, and sediment. 66

Fig. 3.7. Total and individual lifetime cancer risks to humans caused by sixteen VEPAHs through various exposure routes of dermal contact, inhalation, and ingestion. 67

Fig. 3.8. Sensitivity results of BaP posed health risk against the input parameters through different exposure routes, including (a) total risk, (b) dermal contact, (c) inhalation, and (d) ingestion. 68

Fig. 3.9. (a) Total BAP equivalent concentration and (b) the estimated ILCR of the VEPAHs in Zhengzhou city for 17 years. 71

Fig. 3.10. Total lifetime cancer caused ILCR results of local residences in Zhengzhou during last 17 years from 1999 to 2015 and the prediction of the future 10 years based on the history ILCR. 72

Fig. 3.11. The respiratory diseases and lung cancer mortality rate and the estimated air phase VEPAHs concentration from 1999 to 2015. 74

Fig. 3.12. The relation between the respiratory diseases mortality rate (a), lung cancer mortality rate (b) with the air phase VEPAHs concentration. 76

Fig. 4.1. Research diagram of the multiple land-use fugacity model and the multivariate interpretation and system analyses. 81

Fig. 4.2. Diagram of the mass transfer in multiple land-use fugacity model (upper plot) and typical multimedia fugacity model (lower plot). 83

Fig. 4.3. The proposed systematic model calibration method for the multiple land-use model. 87

Fig. 4.4. Location of the study area - downtown and the Tong Zhou district area of Beijing. 96

Fig. 4.5. (a) Annual emissions of 16 PAHs in Beijing from 2000 to 2017. (b) The contribution of PAHs by ring number to the total PAH emissions in 2010. 100

Fig. 4.6. The relationship between the BaP concentration in each compartment and the model parameters of (a) chemical properties, (b) environmental properties, (c) compartment size, and... 102

Fig. 4.7. The selected model parameter in multiple land-use fugacity model via V-plot for BaP. 103

Fig. 4.8. The selected model parameter in multiple land-use fugacity model via VIP scores for each PAH. 105

Fig. 4.9. (a) Total PAH concentrations and (b) the PAH equivalent concentration estimated by the multiple land-use fugacity model. Simulated value (bar plot) and observed values... 107

Fig. 4.10. Comparison of the (a) total PAH concentrations and (b) PAH equivalent concentrations between the values estimated by the Level III fugacity model (bar plot) and... 108

Fig. 4.11. Percentage accumulation of the (a) ∑₁₆PAH concentration and (b) PAHseq concentration in multiple compartments in 2010.[이미지참조] 109

Fig. 4.12. Estimated PAH transformation and fate inside the compartments, as determined by the multiple land-use fugacity model. 111

Fig. 4.13. Land use of Beijing city area and Tong Zhou district from 2012 to 2017. 113

Fig. 4.14. Risk assessment of (a) human and (b) ecological system exposure to PAHs in the study area. 116

Fig. 4.15. Spatial distributions of ∑₁₆PAHs caused ILCR of the study area based on multiple land-use fugacity model. 117

Fig. 4.16. Timeline of COVID-19 epidemic in Beijing and global. 121

Fig. 4.17. Trend of traffic intensity in Beijing. 121

Fig. 4.18. The estimate traffic factor according to the travel intensity index in Beijing. 121

Fig. 4.19. The dynamic variation of ∑₁₆PAHs concentration during the COVID-19 pandemic. 123

Fig. 5.1. Comparing the global temperature variation and solar activity within the last 140 years [NASA]. 127

Fig. 5.2. The number of tweets posted per day from Twitter established year until now. 135

Fig. 5.3. The number of Twitter users in the top 20 countries with the largest Twitter users. 136

Fig. 5.4. The total percentage of male and female Twitter users. 138

Fig. 5.5. The profiles regarding the number of tweets and user age and gender. 138

Fig. 5.6. Creating a new App for Twitter APIs in developer. 140

Fig. 5.7. The generated API keys after registration. 140

Fig. 5.8. The generated access token and access token secret of Twitter APIs. 141

Fig. 5.9. The example of extracting tweets information from Twitter by "tweepy". 142

Fig. 6.1. People as a sensor on climate change mitigation. 146

Fig. 6.2. Research flowchart of climate change Twitter sentiment analysis. 147

Fig. 6.3. The daily number of climate change tweets of the study period. 149

Fig. 6.4. The diagram of CBOW and skip-gram architectures in Word2Vec. 155

Fig. 6.5. The data of daily tweeting number. (a) The number of totals, retweets, and original tweets. Normalized tweeting number of (b) total tweets, (c) retweets, and (d) original tweets. 161

Fig. 6.6. The time trend of the number of climate change tweets. The weekly trend for (a) original tweets and (b) retweets. The daily trend for (c) original tweets and (d) retweets. 162

Fig. 6.7. Word cloud of tweets related to climate change. (a) Raw data, and (b) processed data. 163

Fig. 6.8. The distribution of four classes labelled data. 164

Fig. 6.9. Comparing F1 scores of various classification model with three types of feature extraction methods: BOW, TF-IDF, and Word2Vec. 165

Fig. 6.10. The F1 score, training time, and test time of each classifier with the BOW feature extraction method. 166

Fig. 6.11. The pie chart of the three emotion classes of the climate change related tweets. 167

Fig. 6.12. The distribution of three classes labelled data. 168

Fig. 6.13. Comparing F1 scores of various classification model with three types of feature extraction methods: BOW, TF-IDF, and Word2Vec. 169

Fig. 6.14. The F1 score, training time, and test time of each classifier with the BOW feature extraction method 170

Fig. 6.15. The pie chart of the three emotion classes of the climate change related tweets. 170

Fig. 6.16. The Pro tweets proportion in each country. 172

Fig. 6.17. The number of news tweets from the obtained data in each country. 172

Fig. 6.18. The ratio of negative/positive tweets. 173

Fig. 6.19. The proportion of emotion classes of the most tweeting countries. 173

Fig. 7.1. The Global reported natural disasters from 1970 to 2019. 180

Fig. 7.2. The characteristics of big data. 183

Fig. 7.3. The research flowchart. 186

Fig. 7.4. (a) The proportion of tweeting amount of each investigating corpus in all corpora. (b) the proportion of tweeting number in each month from April to October. 189

Fig. 7.5. The number of tweets in each state of the USA. 190

Fig. 7.6. The average daily AQI of US in 2019. 191

Fig. 7.7. The temperature, relative humidity (RH), and heat index (HI) of California in 2019. 193

Fig. 7.8. The natural disaster factors of floods and hurricanes for indicating the possible close relationships. 195

Fig. 7.9. The daily number of tweets in each corpus along with the time changes. (a) climate change, (b) air pollution, (c) heatwave, (d) floods, and (e) hurricane. 202

Fig. 7.10. The word clouds of each corpus. (a) climate change, (b) air pollution, (c) heatwave, (d) floods, and (e) hurricane. 204

Fig. 7.11. The 10 most frequent hashtags in each corpus. (a) climate change, (b) air pollution, (c) heatwave, (d) floods, and (e) hurricane. 206

Fig. 7.12. The description of the correlations of hashtags with the corpora. 207

Fig. 7.13. The cosine similarity of tweets text in each corpus. 208

Fig. 7.14. The similarity between climate change corpus and consequential events corpora in different months. 209

Fig. 7.15. The daily net sentiment rate of each corpus. 210

Fig. 7.16. The correlation coefficient of the net sentiment rate of each corpus. 211

Fig. 7.17. The climate change consequence events evidence in the US, including air pollution, heatwave, flood, and hurricane. (a) Air pollution concentration (Korosec 2020). (b) NOAA/ESRL... 214

Fig. 7.18. The largest positive effect factors of each state in the US contribute to the number of tweeting on climate change. 218

Fig. 7.19. The largest negative effect factors of each state in the US contribute to the number of tweeting on climate change. 219

Fig. 7.20. The largest positive effect factors of each state in the US contribute to the sentiment of tweets on climate change. 223

Fig. 7.21. The largest negative effect factors of each state in the US contribute to the sentiment of tweets on climate change. 223

Fig. 7.22. The coefficient value of the AQI factor regress to the daily number of tweets on climate change in each state of the US. 225

Fig. 8.1. The interaction between climate change and environmental pollution. 229

Fig. 8.2. The measures of climate change mitigation and environmental pollution control. 231

Fig. 8.3. The hierarchical structure schematic representation of SWOT model. 235

Fig. 8.4. SWOT analysis formation of environmental management of both climate change and environmental pollution. 240

Fig. 8.5. The weights and ranking of SWOT factors and sub-factors (a) SWOT factors, (b) Strength sub-factors, (c) Weakness sub-factors, (d) Opportunity sub-factors, and (e) Threat sub-factors. 242

Fig. 8.6. Fuzzy ideal distances and closeness coefficients for the SWOT strategies. 245

 Environmental pollution and climate change are the serious environmental issues facing by mankind civilization and have received important concerns in recent years. With the development of human society, the adverse impacts of environmental pollution and climate change are increasingly vital. The intrinsic linkages exist between environmental pollution and climate changes which can be used to manage these two tough issues efficiently. The interconnections between typical environmental pollution (air pollution) and climate change have been investigated by many researchers, which can be summarized in four aspects.

• Causes

• Interactions

• Mitigation measurements

• The synergies of climate change and environmental pollution policies

Until now, with the efforts scientists made on understanding and mitigating the problems of climate change and environmental pollution, many technology and policy are introduced to control and reduced the adverse effects of these two issues. Currently, lots of issued policies that are dealing with serious environmental concerns are working based on the designation from different science communities and they are separately executed in different policy frameworks. Besides, the execution of policies among the public is highly related to their perception and awareness of climate change and environmental pollution, which may affect by the unclear or conflict environmental regulation, policy, or suggestions. Therefore, the objective of this dissertation is to reveal the adverse effect of environmental pollution on human health and assess public perception between climate change and environmental pollution.

In Part I, the assessment of pollutants' environmental fate and the potential risk are carried out to help the public figure out the adverse effect of pollutants from surrounding environment. A group of hazardous environmental pollutants, polycyclic aromatic hydrocarbons (PAHs), are selected as the target analytes in this part. Not only because of their strong carcinogenic, mutagenic, and teratogenic effects on humans, also the inconvenient and inefficient monitoring measurements. Chapter 2 reviewed the approaches of estimating organic pollutants environmental fate.

In Chapter 3, to find the relationship between the vehicle population and human cancer risk of the study area, multimedia fugacity models and incremental lifetime cancer risk (ILCR) model is implemented. The heavy traffic burden is always an intractable problem in most of the megacity, not only in making difficulties in commute but also cause pollution problem. The vehicle exhaust PAHs (VEPAHs) is estimated based on the vehicle population and the type of fuel. Furthermore, the multimedia fugacity model is applied to calculate VEPAHs concentrations which shows a high value in the soil compartment than other environmental phases. The multiple pathways of exposure (dermal contact, inhalation, and ingestion) and multiple sources (air, water, soil, and food) of the ILCR model indicate that the cancer risks of VEPAHs follow the order of inhalation ＞ dermal contact ＞ ingestion. The Monte Carlo technique is used to analyze the sensitivity of the model parameters on the final outputs. And the results confirmed that the emission rate of benzo[a]pyrene (BAP) was the most influential factor on the health risks from these VEPAHs, and the half-life in air, partition coefficient, and cancer slope factor are the main factors that influence the ILCR of BAP. The dynamic ILCR model showed that the increased vehicle population between 1999 and 2015 exacerbated the cancer risks caused by VEPAHs in Zhengzhou.

In Chapter 4, a multiple land-use fugacity (MLUF) model was proposed to investigate pollutants' distribution fates comprehensively. In the MLUF model, the soil phase variances are considered in several detailed compartments. Subsequently, this study successfully estimated the fate and transfer of PAHs in Beijing, China through the designed MLUF implementation. Therefore, the proposed MLUF modelling process is interpreted in detail and a multivariate interpretation and systematic model framework are demonstrated for calibration. And several significant model parameters selected in MLUF through PLS-derived VIP scores versus the correlation coefficient values approach. In further, the spatial distributions of Σ16PAHs concentration in the soil phases of the study area acquires from the MLUF model are presented in the map. The PAHs movement and transformation behavior inside the compartments are determined and represented with the estimated mass transport rates. After the pollutants risk assessment, this study obtains the ILCR values for different land covers that follow the order of urban green space ＞ agricultural area ＞ forest and semi-nature area (FSNA). In terms of ecological risk, the risk quotient (RQ) values for different land covers follow the order of urban green space ＞ agricultural area ＞ FSNA, which indicates the relatively higher e hazardous risk in urban green space soil than agricultural and FSNA soil. At last, the proposed dynamic MLUF model implemented into the evaluation of the PAHs concentrations quantitively changes in the study area during the COVID-19 pandemic. The significant influence of the COVID-19 outbreak period on PAHs concentration reveals in the air, water, vegetation, and organic film compartments.

The part II of this dissertation conducted the analyses of the public perception about climate change with Twitter data and analyzed whether the climate change consequence events (such as air pollution, heatwave, floods, and hurricane) will affect people's tweeting behavior about climate change message. In Chapter 5, a review of climate change and public perception is summarized. Besides, an introduction of Twitter and the value of Twitter on information extraction is clarified.

In Chapter 6, the sentiment analysis of climate change related tweets is conducted by implementing machine learning classification algorithms. Totally 3,100,543 climate change related tweets are used to analyze the people's opinion and emotion when mentioning climate change. In this study, the Linear SVM classifier with the BOW feature extraction method shows the best performance on opinion detection and emotion discrimination. The classified climate change tweets show that most tweets believe climate change is caused by human activity. When people mention climate change on Twitter, most are showing negative emotion. The spatial and temptation variance of climate change tweets sentiments from different countries and different month are clearly shown in this research. Detecting public's sentiment on climate change is necessary for both policymaking and guide on the policy implementation.

In Chapter 7, the relationship between people's awareness of climate change and climate change caused natural events (including air pollution, heatwave, floods, and hurricane) were analyzed. The relation analysis is conducted from two aspects, Twitter content analysis and filed data impact analysis. The big data analysis and text mining algorithm are implemented to find the similarity between the climate change corpus and the consequence corpora. This study shows that when serious events occurred, people are more likely to link the consequence events with climate change when they are tweeting. The regression model indicated that climate change consequence events influence the people's tweeting activity of climate change related messages on Twitter. The dominant influence factors on the number of people's climate change related tweets are varied from state to state. From this study, it can be found that people's experience on climate change consequences events can lead them to link the climate change with the events easily than they recognize the relation with other events.

In Chapter 8, based on the relationship between the typical environmental pollution (air pollution) and climate change, the environmental management strategies are introduced and evaluated to provide the information to the public, policymaker, manufacturing industries, and so on. The SWOT (strengths, weaknesses, opportunities, and threats) view analysis on the environmental pollution and climate change combination managements is used to provide feasible strategies with considering the strengths, weaknesses, opportunities, and threats that human-being faced recently. An integrated environmental management framework for mitigation climate change and environmental pollution using a SWOT analysis augmented with analytical network process (ANP) to obtain criteria weights was implemented and a technique for ordering performance by similarity to ideal solution (fuzzy-TOPSIS) methodology was used to prioritize alternative environmental management strategies. Combining environmental pollution (EP) and climate change (CC) has several typical strengths in building a better sustainable society. More efficient mitigation performance can be derived by targeting in both EP and CC rather than separately deal with them. The policies regarding the EP and CC are an essential part of environmental management. By tracking the origins of EP and CC, some common aspects indicated influences in both EP and CC, especially the energy modes in the combustion of fossil fuels. Therefore, it is a strong strength that the sociality has started to replace the traditional energy modes by policy supporting, technology exploring, and building renewable energy harvesting infrastructures.