CAPPS模式中二次污染物的预报研究

【作者】 张恺；

【导师】 徐大海；

【作者基本信息】 中国气象科学研究院 ， 气象学， 2004， 硕士

【摘要】 本文在CAPPS原有模式基础上研究了城市大气中臭氧等二次污染物的预报问题。 为了使CAPPS模式具备预报二次污染物的能力，本文从考虑了化学反应的大气平流扩散方程出发，经过推导给出了体积分形式下污染物浓度随时间变化的方程组，并根据实际要求采用GRS光化学方案和解刚性微分方程的QSSA算法建立起光化学模式，为二次污染物浓度的模拟建立预报模型。同时针对VOC浓度缺测及氮氧化物源排放强度难以确定的问题采用了遗传算法反演的方法来解决。模式很好的考虑了业务预报的需要，对计算条件、污染物监测条件没有过高要求。 本文还利用973项目监测的共48天的污染物浓度资料和相应的气象资料进行单箱预报检验，其中分别用MM4(实验一)和MM5(实验二)预报的气象场作为CAPPS模式的输入来模拟日间臭氧浓度的变化。实验二中VOC的浓度采用四种方案给出，即基于逐日反演统计结果的固定值、早晨逐时反演值、以及根据工作日和周末差别分类反演的统计结果对上述取值加以限定而得到的两种新方案。通过对比不同方案对预报效果的影响，分别检验了他们的正确性。预报试验结果表明，即使VOC的浓度只采用基于逐日反演统计结果的固定值，三个监测时段模拟的日最大臭氧小时平均浓度与监测值相关系数也分别达到0.843，0.774，0.630。选用MM5预报的气象场作为CAPPS模式的输入时的预报效果比选用MM4的结果好。另外，采用了VOC浓度工作日和周末分类反演的统计结果以后，预报准确度有相当大的提高，三个监测时段模拟的日最大臭氧小时平均浓度与监测值相关系数都达到了0.7以上。相比之下，早晨逐时反演值的应用效果起伏较大。 由于监测资料不足，对二次气溶胶浓度只挑选晴天且臭氧浓度监测值较高的两个个例进行了模拟，并与监测的气溶胶数据以及原CAPPS预报结果进行了对比。结果表明，在光化学污染比较典型的两天中，考虑了化学反应的模式对气溶胶浓度的模拟效果有所改善。更多还原

【Abstract】 A study on the prediction of secondary pollutants’ concentration in cities are carried out in this paper, based on a former version of the City Air Pollution Prediction System (CAPPS).A set of volume integrated equations for pollutant concentration are derived from the advection and diffusion equations with chemical reaction terms involved. GRS scheme of the chemical reactions and QSSA solver for the stiff ordinary differential equations are selected for the photochemical module. The generic algorithm is employed to inverse the concentration of VOC, observation data of which are usually not available in most cities in China. With careful consideration of the features of operational prediction, the new model version is demanding on neither the computational ability nor the capacity of pollutant monitoring systems.Single-box simulations are carried out using observational data of the 973 project. Products from both MM4 and MM5 are passed on to CAPPS to detect the effects of meteorological fields prediction on the veracity of ozone forecast. Moreover, since VOC concentration plays an important role in ozone formation in the urban area, a series of experiments are performed under four different VOC concentration scenarios. It shows that the modified CAPPS model serves as a good choice for urban ozone prediction. According to these results, substitution of MM5 for MM4 as the meteorological field generator has a positive effect on the ozone prediction, and the distinguishing between weekdays and weekends in specification of the VOC concentration also improves the accuracy of prediction significantly.Due to the scarcity of observational data, predictions of the secondary aerosols are practised on only two cases with fine weather and high ozone concentration. The results show that model prediction is acceptable during the two typical days of photochemical air pollution.更多还原