Estimation of nitrogen dioxide concentrations in Inner Bangkok using Land Use Regression modeling and GIS

Abstract

In Bangkok, nitrogen dioxide (NO2) concentrations have long been measured hourly by the Pollution Control Department (PCD) at 12 monitoring stations covering 430 km2 of Inner Bangkok. In the past, to estimate NO2 concentrations at any unmeasured location, the proximity model, interpolation model, or dispersion model was employed. These models used distance from a measured location as the sole determinant of any estimation. Toward the end of the 1990s, the more sophisticated land use regression (LUR) model was introduced. This model with its built-in geographic information system (GIS) and multiple regression analysis enabled the inclusion of other important determining variables such as land use types, traffic volume, and selected meteorological variables. This study aims to apply the LUR model for the estimation of NO2 concentrations over the study area covering Inner Bangkok. Monthly average NO2 concentrations, traffic volume, land use types, road areas together with humidity, temperature, wind speed, and rainfall data, measured at or within the vicinities of the 12 PCD stations, were input into the model. Only humidity, temperature, wind speed, rainfall, residential land use, and industrial land use were found to have influenced the NO2 concentrations in inner Bangkok. The resulting coefficient of determination (R squared) of 0.759 implies that 76 % of the variations in NO2 concentrations in inner Bangkok can be explained by the model. The study will, however, continue to obtain more precise traffic volume data in terms of time scale to improve the model.