PM_2.5 mapping using integrated geographically temporally weighted regression (GTWR) and random sample consensus...

Area of Science:

• Environmental Science
• Atmospheric Science
• Remote Sensing

Background:

• The relationship between aerosol optical depth (AOD) and fine particulate matter (PM2.5) is complex, affected by model uncertainties, resolution mismatches, and data integration challenges.
• Accurate mapping of PM2.5 is crucial for environmental monitoring and public health assessments.

Purpose of the Study:

• To develop and apply an integrated model for high-resolution PM2.5 mapping over Taiwan.
• To address uncertainties and outliers in AOD data for improved PM2.5 estimation.

Main Methods:

• Utilized dark target (DT) and merged DT/Deep Blue (DB) AOD observations at 3-km resolution.
• Integrated geographically temporally weighted regression (GTWR) with RANdom SAmple Consensus (RANSAC) for robust data fitting.
• Applied spatial variability and hotspot analysis to the resulting PM2.5 maps.

Main Results:

• The integrated GTWR-RANSAC model demonstrated a fine goodness-of-fit between observed PM2.5 and AOD data.
• The model effectively handled spatiotemporal heterogeneity and data uncertainty, overcoming estimation challenges.
• PM2.5 mapping revealed significant spatial variability and identified hotspots across Taiwan.

Conclusions:

• The integrated GTWR-RANSAC approach is a powerful tool for inferring PM2.5 patterns from AOD data, even with observational uncertainties.
• The developed mapping technique provides valuable insights into the spatiotemporal distribution of PM2.5 in Taiwan.
• Hotspot analysis of PM2.5 maps aids in understanding pollution variations and informing environmental management strategies.