Estimating short- and long-term impact of COVID-19 on air quality, human health, and economic losses in China through machine learning counterfactual simulations Público
Cheng, Haoran (Spring 2022)
Abstract
The short-term reduction in air pollutant concentrations due to acute lock-down during COVID- 19 has been widely studied but few have quantified the amount of reduction in pollutant concentrations due to COVID. Fewer studies have analyzed the long-term impact of COVID-19 on local air quality or studied their social impacts. Our study uses machine learning counterfactual simulations to analyze both the acute (~weeks) and chronic (~year) impacts of COVID-19 on air quality, human health, and the economy in China. We analyzed concentrations of six air pollutants (PM2.5, PM10, SO2, NO2, CO, O3) in 39 major cities in China during and after COVID-19 lockdown (January 28th, 2020 to April 30th, 2021) and predicted the pollutant concentrations for each pollutant in a counterfactual scenario – with no “COVID-19” lockdown, using local meteorological data with XGBoost machine learning model. We calculated the associated health and economic impacts in the counterfactual world and compared them with the real-world impact. Among the cities surveyed, 64%, 93%, 82%, 95%, 81% of the cities showed a statistically significant reduction between the observed PM2.5, PM10, SO2, NO2, CO during the lockdown, compared to the model prediction. We observed an increase in O3 concentrations in all cities during the lockdown but this increase was only significant for 34% of the cities analyzed. The total amount of observed gaseous oxidants (Ox= NO2+O3) remained mostly the same, compared to the counterfactual scenario. In both the observed and predicted simulations, NO2 and O3 were the leading cause of excess health and economic burdens. For the post-lockdown period, the observed concentrations of each pollutant were still lower than the counterfactual scenario for all regions combined but the differences were much smaller (more than 90% reduction in differences) compared to those during the lockdown period. The health and economic burdens due to air pollution continued to be lower with COVID than without in the post-lockdown period, except for O3 in the Northeast (NE), Southeast (SE) and Southwest (SW) regions. The implementation of COVID lockdown in China resulted in a significant reduction of various air pollutant concentrations. Long-term impacts varied among the cities and pollutants studied. The chronic, post-lockdown impact of COVID-19 on China’s air quality is yet to be determined and further research is needed.
Table of Contents
1 Introduction 1
2 Data and Method 3
2.1. Data 3
2.1.1. Air Quality Data 3
2.1.2 Meteorological Data 3
2.1.3 Demographical Data 3
2.2 Method 4
2.2.1 Sample Inclusion and Grouping 4
2.2.2 Air Quality Change 7
2.2.3 Health Impact Analysis 9
2.2.4 Economic Impact Analysis 11
3 Results 11
3.1 Model Validation 11
3.2 During Lockdown 13
3.2.1Air Quality Change 13
3.2.2 Premature Mortality Health 16
3.2.3 Economic Losses 18
3.3 Post Lockdown 20
3.3.1 Excess Mortality Health 20
3.3.2 Economic Losses 23
3.4 Limitations and Strengths 24
3.5 Future Steps 26
4 Reference 26
5 Appendix 32
About this Honors Thesis
| School | |
|---|---|
| Department | |
| Subfield / Discipline | |
| Degree | |
| Submission | |
| Language |
|
| Research Field | |
| Palabra Clave | |
| Committee Chair / Thesis Advisor | |
| Committee Members |
Primary PDF
| Thumbnail | Title | Date Uploaded | Actions |
|---|---|---|---|
|
|
Estimating short- and long-term impact of COVID-19 on air quality, human health, and economic losses in China through machine learning counterfactual simulations () | 2022-04-12 23:46:57 -0400 |
|
Supplemental Files
| Thumbnail | Title | Date Uploaded | Actions |
|---|