Abstract CSEOF analysis is applied for the springtime (March, April, May) daily PM10 concentrations measured at 23 Ministry of Environment stations in Seoul, Korea for the period of 2003–2012. Six meteorological variables at 12 pressure levels are also acquired from the ERA Interim reanalysis datasets. CSEOF analysis is conducted for each meteorological variable over East Asia. Regression analysis is conducted in CSEOF space between the PM10 concentrations and individual meteorological variables to identify associated atmospheric conditions for each CSEOF mode. By adding the regressed loading vectors with the mean meteorological fields, the daily atmospheric conditions are obtained for the first five CSEOF modes. Then, HYSPLIT model is run with the atmospheric conditions for each CSEOF mode in order to back trace the air parcels and dust reaching Seoul. The K-means clustering algorithm is applied to identify major source regions for each CSEOF mode of the PM10 concentrations in Seoul. Three main source regions identified based on the mean fields are: (1) northern Taklamakan Desert (NTD), (2) Gobi Desert and (GD), and (3) East China industrial area (ECI). The main source regions for the mean meteorological fields are consistent with those of previous study; 41% of the source locations are located in GD followed by ECI (37%) and NTD (21%). Back trajectory calculations based on CSEOF analysis of meteorological variables identify distinct source characteristics associated with each CSEOF mode and greatly facilitate the interpretation of the PM10 variability in Seoul in terms of transportation route and meteorological conditions including the source area.

中文翻译：

---

韩国首尔春季PM 10浓度变化的源区及气象因素分析

摘要 CSEOF 分析适用于 2003 年至 2012 年期间在韩国首尔 23 个环境部站测量的春季（3 月、4 月、5 月）每日 PM10 浓度。还从 ERA Interim 再分析数据集获取了 12 个压力级别的六个气象变量。对东亚的每个气象变量进行 CSEOF 分析。回归分析在 PM10 浓度和各个气象变量之间的 CSEOF 空间中进行，以确定每种 CSEOF 模式的相关大气条件。通过将回归载荷向量与平均气象场相加，获得前五种 CSEOF 模式的每日大气条件。然后，HYSPLIT 模型在每种 CSEOF 模式的大气条件下运行，以便回溯到达首尔的气团和灰尘。应用 K-means 聚类算法识别首尔 PM10 浓度各 CSEOF 模式的主要源区。根据平均场确定的三个主要源区是：（1）塔克拉玛干沙漠北部（NTD），（2）戈​​壁沙漠和（GD），以及（3）华东工业区（ECI）。平均气象场的主要源区与前人研究一致；41% 的源位置位于 GD，其次是 ECI (37%) 和 NTD (21%)。