Combining deep learning and machine learning techniques to track air pollution in relation to vegetation cover...

Area of Science:

• Environmental Science
• Urban Planning
• Data Science

Background:

• Rapid urbanization in Dhaka has led to severe air pollution and a decline in green spaces.
• Air pollution, particularly particulate matter (PM2.5 and PM10), poses significant environmental and health risks.
• Urban green spaces are crucial for environmental quality and mitigating pollution impacts.

Purpose of the Study:

• To investigate the relationship between air pollution and urban green space changes in Dhaka from 1990 to 2022.
• To apply machine learning (ML) and deep learning (DL) techniques for environmental assessment and prediction.
• To establish the causal links between declining vegetation and rising air pollution levels.

Main Methods:

• Utilized ML algorithms (XGB, SVM, RF) to predict high air pollution zones.
• Employed DL models (Unet, Unet++, MAnet, Linknet) to forecast vegetation cover trends.
• Analyzed data spanning from 1990 to 2022 to track environmental changes.

Main Results:

• ML models accurately identified areas with high air pollution concentrations.
• DL models effectively forecasted trends in vegetation cover decline.
• A significant negative correlation was confirmed between increased air pollution and reduced vegetation cover.

Conclusions:

• The study confirms a strong inverse relationship between air pollution and green spaces in Dhaka.
• Findings underscore the urgent need for integrated pollution management and green infrastructure development.
• ML and DL are valuable, cost-effective tools for monitoring environmental degradation and informing urban planning.