集合最优插值同化方法在珠三角空气质量模拟中的初步应用

Preliminary application of ensemble optimal interpolation data assimilation method on air quality numerical modeling in the Pearl River Delta

本文旨在探讨集合最优插值(EnOI)同化方法对MM5-STEM空气质量模式污染物浓度预报场的修正能力,先从局地化尺度(L)及经验系数(α)的敏感性试验中获得NO2、SO2和PM10各自的"最优L和α组合",然后对此参数设置下的同化结果进行分析.研究结果显示,EnOI在NO2、SO2及PM10的同化试验中均取得较好的效果,检验站点均方根误差(RMSE)的平均下降比例分别可达33%、32%和42%,RMSE值下降的站点占检验站点总数的比例分别为86%、84%和91%.表明该方法能够有效地应用在珠三角空气质量模拟中,产生与实际更为接近的污染物浓度预报场.

This paper mainly evaluated the ability of Ensemble Optimal Interpolation Data Assimilation Method （EnOI） in modifying the pollutants concentration forecast field in MM5-STEM air quality numerical model. Optimal settings of localization scale （L） and empirical coefficient （α） of NO2, SO2 and PM10 were acquired by sensitivity tests respectively. Assimilation results under these coefficient settings were analyzed, the results showed that EnOI had a good performance in the data assimilation experiments of NO2, SO2 and PM10, with RMSE decreasing percentage of 33%, 32% and 42%, respectively. The proportion of verification stations with decreased RMSE were 86%, 84% and 91%, respectively, which proved that EnOI produced a pollutant concentration forecast field closer to the true situation, therefore can be effectively applied in air quality modeling in the PRD.