Dispersion mapping of Carbon Monoxide (CO) derived Sentinel 5P and evaluation with LUR model during winter in Ulaanbaatar, Mongolia

In recent years, air pollution caused by coal combustion in Ulaanbaatar city is becoming a priority health concern for the city residents and many researchers are investigating its ingredient and impact on people's health. This study aims to map Carbon Monoxide (CO) dispersion from remote sensing data during the winter season and evaluate it with the Land Use Regression (LUR) model result and fixed station measurements. Fixed station data from the National Agency for Meteorological and Environmental Monitoring (NAMEM) were utilized as a reference point for remote sensing data, a dependent variable for LUR model. Sentinel 5P TROPOMI CO data was correlated moderately-positive with fixed station measurements (R=0.56). And LUR model performance was relatively higher compared to Sentinel data or its determination coefficient R2=0.71, Adjusted R2=0.53, Root Mean Square Error (RMSE)=0.84 mg/m3 and Mean Absolute Error (MAE)=0.7 mg/m3. These statistical evaluation coefficients are relatively lower than other similar studies, and authors have adduced it to a lack of reference data and satellite data spatial resolution. Finally, these two result maps were compared, which have shown lower correlation-positive or R=0.44. Even though these two map results are not exactly the same, Pearson’s correlation coefficient shows that both image results move in the same direction or they are moderately showing the same results.