京津冀地区细颗粒物污染来源实时预报系统的构建和初步应用

摘要

The forecasting system of pollutant origins was established based on numerical model of regional air quality and tagging technology of emission sources. It realized the realtime forecasting of pollutant origins over Beijing-Tianjin-Hebei area. To cope with the urgent time restriction in forecasting, a series of special designs were made in emission preprocessing, source apportionment prediction and automatic routine operation. These included developing the fast emission preprocessing technology, determining the solution of source apportionment according to regions and sectors by trial calculation, compiling and running the regional air quality model via a hybrid usage of two mainstream solutions of parallel computation, and designing a distributed solution to handle multiple tasks in routine forecasting running for sufficient use of computation resources. All of the optimization measures effectively reduced the time of forecast runs. At present, the forecasting system is in routine operation stage. It predicts the air quality for the 3 days ahead and the origins of major pollutants according to regions and sectors each day before 0800LCT. The system had been applied in some big events, such as for the air quality guarantee and post-assessment for the period of victory parade of war against Japan, the pre-warning of severe air pollution from the end of November to the early of December, 2015.%基于区域空气质量数值模型和源示踪技术，构建污染来源预报系统，实现了京津冀地区污染来源的实时预报。针对预报中最迫切的时效限制，在系统的源排放预处理、污染物来源贡献计算方案、业务化运行等方面进行特别的设计，主要包括：开发出快速源排放前处理技术；通过试算确定了分区域和行业的污染来源追踪方案；采用目前2种主流的并行计算方式混合编译区域空气质量模型并运行；设计了业务化运行中多任务的分布式计算方案，以充分利用计算资源。这些优化措施有效地缩短了预报时间。目前系统已处于业务化运行阶段，每天08：00之前，预报出未来3 d的区域空气质量，同时给出所关注区域主要污染物分区域和行业的来源贡献状况；系统已应用在庆祝抗战胜利70周年阅兵期间空气质量保障后评估，以及2015年11月底—12月初京津冀重污染预警等应急工作中。