Spatiotemporal characteristics of PM2.5-meteorological factor relationship in winter of Korea

Abstract

Worsening air quality, especially the episode of high concentration of Particulate Matter (PM) during winter and spring seasons has been a societal issue in Asia including India, China, and Korea in recent decades due to its potential for adverse effects on human health. PM concentration of Korea is known to be affected by both internal and external sources under certain weather conditions. With the increasing demand on the long-term PM forecast, covariability between meteorological conditions and PM concentration is suggested as a potential predictor for such forecast. In this study, the relationship between PM2.5 and covarying meteorological factors were characterized by their spatial and temporal characteristics. To achieve our goal, the PM2.5 concentration observation in the period of 2000 to 2016 in the Seoul metropolitan area is used during winter season (December-January-February) and spring season (March – April – May). Covarying meteorological pattern is constructed with correlation map for both seasons. Dominant meteorological patterns of winter that cause high PM2.5 concentration of Korea were characterized with (1) the mid-level high-pressure center at the East Sea indicated by positive 500hPa geopotential height anomalies over (2) positive temperature anomalies near-surface (1000hPa) with (3) the corresponding wind anomalies pattens (500hPa u-wind and 850hPa v-wind). These patterns were consistent with the favorable conditions for the high PM events (Cai et al. 2017; Lee et al. 2018). Compare to winter, spring patterns showed overall displacement of the center from the East Sea to the Korean Peninsula. Based on the dominant patterns of winter, year-to-year variabilities of correlation were analyzed and 5-day and 7-day running means were utilized as the time filters for the synoptic patterns. Although they showed distinct annual variability, time filters showed no distinct tendency or improvement for the correlation. Power spectrum and coherence of PM2.5 and factors demonstrated that they are largely composed of cycles of 5 to 10days or more with minor short cycles. Prediction on the frequency of the high-PM events may be possible by considering these results. However, building an index represents the meteorological conditions with the factors used in this study may have limitation since they are dependent to each other.