IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (Jan 2024)
Estimating Ground-Level Hourly PM<sub>2.5</sub> Concentrations in Thailand Using Satellite Data: A Log-Linear Model With Sum Contrast Analysis
Abstract
This study introduces a novel method for estimating hourly concentrations of particulate matter 2.5 μm (PM2.5) using satellite data. The pollution control department of Thailand collected hourly PM2.5 data nationwide in 2020. NASA's Earth Observing System Data and Information System encompasses all moderate resolution imaging spectroradiometer satellite data. We employed aerosol optical depth (AOD), land surface temperature (LST), normalized difference vegetation index (NDVI), and elevation (EV) in our analysis. The approach incorporates a weighted sum contrast log-linear regression model that integrates satellite data, allowing for the examination of small-scale hourly variations in PM2.5 concentrations. The results reveal a high correlation between hourly PM2.5 levels and AOD, LST, NDVI, EV, time, and week of the year in terms of spatial distribution, with an R2 value of 53.8%. The mean hourly PM2.5 concentration was 23.1 μg/m3, displaying elevated concentrations during the dry season (November to March) and peak hours (8 to 11 a.m. and 8 to 12 p.m.). Positive correlations between AOD and PM2.5, especially when AOD exceeded 0.52, and between LST and PM2.5, particularly when LST exceeded 33.9 °C, along with NDVI ranging from −0.08 to 0.18 and EV above 67.9 m, resulted in higher PM2.5 levels than the overall mean. The proposed model proved valuable for interpretation and practical application, offering comparable estimated hourly PM2.5 concentrations at a 1-km resolution with monitoring stations. This suggests that researchers or policymakers may use the model to understand hourly PM2.5 fluctuations and their impact on human health and the environment.
Keywords
