大气自净能力指数的气候特征与应用研究

为了定量地评估污染气象条件对空气污染的作用并实现对空气污染潜势的预报，本文在城市大气污染数值预报系统（CAPPS）预报原理的基础上，定义了大气自净能力指数，并分别给出了采用气象站观测资料和通过数值模拟计算大气自净能力指数的方法.基于气象站观测资料的全国大气自净能力指数分析计算表明，全国大气自净能力最差的地区分布在四川盆地和新疆塔里木盆地，大气自净能力最强的地区分布在青藏高原、蒙古高原、云贵高原、以及东北平原和三江平原、山东半岛和海南岛；1961~2017年，京津冀、长三角和珠三角地区的大气自净能力指数呈下降的变化趋势，全年低自净能力日数呈上升的变化趋势.采用大气自净能力指数评估2014年北京APEC会议期间大气污染防控效果，表明在11月8~10日极端不利扩散气象条件发生时，减排措施使北京市空气质量AQI平均降低77%，使京津冀平原地区11个城市的空气质量AQI平均降低37%.基于国家气候中心月动力延伸气候预测模式（DERF2.0）的预报产品和中尺度模式（WRF），建立了可以预测全国未来40d逐日大气自净能力指数的延伸期-月尺度大气污染潜势预测系统，回报实验表明，在大多数情况下可以提前15d预报出大气重污染过程；月尺度的大气重污染过程预报效果更大程度上取决于月动力延伸气候预测模式（DERF2.0）的预报准确率.

The climate characteristics of atmospheric self-cleaning ability index and its application in China

In order to assess the effect of meteorological conditions on air pollution quantitatively and predict the air pollution potential, the Atmospheric Self-cleaning ability Index (ASI) was defined based on the prediction principle of city air pollution prediction system (CAPPS). Both computational methods of ASI using observational data of meteorological stations and the meso-scale modelling result were introduced. The analysis of ASI in China showed that the lowest ASI was located in Sichuan Basin and Talimu Basin in Xinjiang, and the highest ASI was located in Plateau, island and peninsula areas. The decrease of ASI and the increase of the low ASI days of a year were found in regions of Jing-Jin-Ji, Yangtze River Delta and Pearl River Delta from 1961 to 2017. The effect evaluation of air pollution prevention and control during the Beijing APEC conference using ASI showed that the AQI in Beijing decreased by 77% and the mean AQI of 11cities in Jing-Jin-Ji Plain decreased by 37% because of the emission reduction while the worst air pollution meteorological condition happened in 8^th~10^th Nov. 2014. An air pollution potential prediction system on extended and monthly scales, which can predict nationwide daily ASI of 40 days in advance, was set up based on the combination of the production of the extended and monthly dynamical climate model (DERF2.0) and the downscaling of WRF model. The historical simulation experiment showed that the system could forecast the process of heavy air pollution events 15days in advance in most cases, and the forecast accuracy depended on the extended and monthly dynamical climate model (DERF2.0) to a considerable extent.