Prediction of Hourly Particulate Matter Concentrations in Chiangmai, Thailand Using MODIS Aerosol Optical Depth and Ground-Based Meteorological Data

Various extreme events recorded over the world have been recognized as scientific-based evidence from possible climate change and variability. The incidence of increasing forest fires and intensive agricultural field burning in Chiangmai and Northern Thailand due to favor conditions may also due to a likely increase of droughts caused by the changing climate. Smog from biomass burning, particularly particulate matter (PM) seriously affects health and the environment. Lack and sparse of ground monitors may cause unreliability for warning information. Satellite remote sensing is now a promising technology for air quality prediction at ground level. This study was to investigate the statistical model for predicting PM concentration using satellite data. Aerosol optical depth (AOD) data were gathered from MODIS-Terra platform while hourly PM2.5 and PM10 data were collected from the Pollution Control Department. The relationship between AOD and hourly PM over Chiangmai was addressed by Model I-Simple linear regression and Model II-Multiple linear regression with ground-based meteorological data correction. The data used for the statistical analyses were from smog period in 2012 (January-April). Results revealed that AOD and hourly PM in Model I were positively correlated with the coefficient of determination (R2) of 0.22 and 0.21, respectively for PM2.5 and PM10. The relationship between AOD and hourly PM was improved significantly when correcting with relative humidity and temperature data. The model II gave R2 of 0.77 and 0.71, respectively for PM2.5 and PM10. To investigate the validity of model, the regression equation obtained from Model II was then applied with smog data over Chiangmai in March 2007. The model performed reasonably with R2 of 0.74. The model applications would provide supplementary data to other areas with similar conditions and without air quality monitoring stations, and reduce false warning the level of air pollution associated with smog from intensive biomass burning. However, further investigation in different locations should be conducted to confirm the applicability of the model