The Prediction of Particulate Matters 2.5 Micron (PM2.5) Concentration using Geographic Weighted Regression Modeling in Nan Province

Abstract

This study has the main objective to study prediction of Particulate Matters 2.5 micron (PM2.5) with Geographic Weighted Regression Modeling: GWR in Nan Province. The study had analyzed the relationship of concentration of PM2.5 from ground-based station in March 2018 together with influencing factors that were hotspot from VIIRS of NASA Suomi NPP satellite and humidity from the same such station. The study had determined these factors using predictive analysis with mathematic equation from Geographic Weighted Regression. Result model had a coefficient of decision (R2) of 0.893832, meaning that the model has very good relationship for prediction of PM2.5 concentrations in the study area. In model validation of this model, it was the comparison of predictive PM2.5 concentration data from the model and PM2.5 concentration data from ground-based station of the Pollution Control Department at 36t Station, Muang District, Chiang Mai Province. The result of evaluation was 89.23 percent, which is considered to be a high prediction accuracy.