Statistical assessment of the influence of hydrometeorological factors on fine dusts

Fine dust, also known as particulate matter, refers to very small matters floating in the atmosphere, and has become one of major concerns in modern society. Fine dust arises not only by anthropogenic emissions, but also by natural sources, and previous studies have revealed that it intensifies, migrates and decreases by several hydrometeorological factors, such as solar radiation, wind and precipitation. The people of South Korea get information of fine dust concentration from Air Quality Monitoring Systems (AQMS). However, due to the spatial imbalance of the AQMS observatories, those who live in places without observatories are using nearest-neighbor-interpolated fine dust information. Therefore, it is crucial to monitor fine dusts across large area to assess spatial variability of fine dust concentration. In this study, we tried to analyze the influences of hydrometeorological factors on fine dust concentration and investigate the capability of the developed model in estimating spatial variability of fine dust, by using Bayesian Model Averaging (BMA). By analyzing Posterior Inclusion Probabilities (PIP) of the hydrometeorological variables chosen from Global Land Data Assimilation System (GLDAS), we could assess positive/negative influences of each variable on the variance of fine dust concentration (e.g. positive influence of solar radiation, negative influence of precipitation). The optimal regression model from weighted average of the models developed by BMA showed moderate correlation with ground-based observation of fine dust. Compared with the nearest neighbor method, the model performed better in expressing spatial variability, implying the possibility of global scale monitoring of fine dust.