欢迎访问《环境科学学报》编辑部网站！

快速检索			高级检索

研究论文

A WRF-RTIM-based model output statistics method for haze event prediction in Shanghai

摘要：应用基于系统辨识理论的实时迭代模式（real-time iterative model，RTIM）对WRF模式预报结果进行后处理，建立了上海地区霾天气的模式输出-统计（model output statistics，MOS）方法.首先，根据WRF模式的气象输出资料，结合大气污染观测数据，筛选出霾事件的预报因子；其次，运用系统辨识实时迭代模型，建立依据MOS预报方法的PM_2.5、PM₁₀和能见度预报模式；最后根据霾事件的判别标准，对上海2012年11月—2013年1月的霾日进行24 h和48 h预报.结果表明，PM_2.5模式预报成功率为75.0%~63.9%，PM₁₀模式预报成功率为87.5%~81.8%，能见度模式预报成功率为71.0%~74.2%，霾日预报成功率为73.7%~72.7%.分析表明，研究期间上海的气溶胶以细颗粒物为主，低能见度主要由导致霾现象的PM_2.5引起.该方法能较准确地预报霾日的发生，可为我国城市大气环境业务预报提供参考依据.

Abstract：Model Output Statistics (MOS), generated from post-processing of the WRF model output using the real-time iterative model(RTIM)based on system identification theory, is employed to predict haze events in Shanghai. Prediction factors were first selected from a candidate set through a stepwise regression analysis according to the WRF outputs in combination with observation data of air pollutants. RTIM was then applied to establish models for PM_2.5, PM₁₀ and visibility prediction with respect to MOS. In the final step, these models were empirically tested on the basis of haze event criteria to produce 24 h and 48 h forecasts of haze days from Nov. 2012 to Jan. 2013 in Shanghai. The results indicated that the prediction success rate reached 75.0%~63.9% for PM_2.5, 87.5%~81.8% for PM₁₀, 71.0%~74.2% for visibility, and 73.7%~72.7% for haze pollution events. During the study period, the aerosol in Shanghai was dominated by fine particulates, and the low visibility was mainly caused by the accumulation of fine particulates such as PM_2.5 in the haze formation. The model seemed to perform well for haze prediction during the study period and presented great potential to be applied to operational environmental prediction.