Suitable model for estimation of PM2.5 concentration using aerosol optical thickness (AOT) and ground based station: Under the dome in upper Northern, Thailand

• Main Authors: A. Charoenpanyanet, P. Hemwan
• Format: Journal
• Published: 2020
• Subjects: Earth and Planetary Sciences; Physics and Astronomy; Social Sciences
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85073335857&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/67784
• Institution: Chiang Mai University

© Geoinformatics International. The upper northern areas of Thailand have faced the problem of smog for more than a decade due to open-air burning, geographical features, and climate which all allow the problem to occur continually. One of the air pollutants causing smog is fine particulate matter, less than PM2.5 in diameter, which affects people’s health. This research aims at finding a suitable model to be used in estimating PM2.5 concentration from the data collected by the Moderate Resolution Imaging Spectroradiometer, MODIS. The areas of study covering 8 provinces in the upper north of Thailand were used in this research. The data used in this study were Aerosol Optical Thickness (AOT) values, MOD04 data of the MODIS system, which all of these were qualified to be used in detecting this particulate matter in the atmosphere. According to the research methodology, PM2.5 concentrations from 3 ground stations, on the same positions of AOT values from MODO4 produced in a total period of 10 days from February to May 2016, were calculated to examine the suitable model out of 7 models by linear regression and non-linear regression which included Cubic model, S-curve model, Quadratic model, Logarithmic model, Linear model, Power model, and Exponential model. After that, the suitable model was analyzed for its validation and applied in estimating PM2.5 concentration in the areas of upper northern region and Bangkok. The results showed that Cubic model had the highest coefficient of determination (R2) at 0.770 and its model validation was 83.33%.