Understanding of the Heavily Episodes Using the MM5-Model-3/CMAQ in Handan city, China Fenfen Zhang ab1, Litao Wang* ab, Zhe Wei ab, Pu Zhang ab, Jing Yang ab, Xiujuan Zhao ab a Department of Environmental Engineering, School of City Construction, Hebei University of Engineering, Handan, Hebei 056038, China b State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China 1. Introduction Among all air pollutants, attentions of particulate matters pollution (as PM 2.5, PM 10 ) have been prompted and become one of key pollution priorities facing the municipal government (Wu et al, 2010). The Hebei province is a heavily populated and industrialized area with surround by Beijing, Tianjin, Shanxi, Shandong and Henan provinces, which are located in the area of the North China Plain (NCP). According to Wang, et al, approximately 65% of PM 2.5 in Shijiazhuang and Xingtai (typical cities in Hebei province) originated from the local southern Hebei emission and Shanxi Province and the northern area of Hebei are two major regional contributors (Wang et al, 2012). It was reported that the Handan city (another cities in Hebei province) was one of the most serious pollution areas from the end of year 2012 to the beginning of the year 2013 according to the report from the China National Environmental Monitoring Centre (CNEMC). Most areas in China suffered from the particulate matters pollution and continuous haze days during that time. The monitoring site located at the campus of Hebei University of Engineering is used to monitor the value about air quality for analyzing heavy episodes, which is compared with the simulated value of some pollutants. The target month, January in 2013 with the seriously air quality, was selected to simulate the air quality using the MM5-Model-3/CMAQ. At that time, the chemical characteristics of PM 2.5 in Handan city were analyzed to help understand the CMAQ model. 1 Author address: Fenfen Zhang, Department of Environmental Engineering, School of City Construction, Hebei University of Engineering, Handan, Hebei 056038, China. Email: senger2008@163.com. Phone number: (+86)18330030758
2. Summary of the simulation results 2.1 Temporal variations Fgure1: Spatial variation of simulated monthly-average concentration for SO 2, NOx, PM 10 and PM 2.5 at 36-km in Jan. 2013. The Figure 1 showed that heavy episodes appeared over the most areas in China in Jan. 2013. Overall, the concentrations of pollutants were relatively higher than ever before and lasted for a long time, which resulted in a widely regional and seriously haze days over the most areas in China. Especially for the PM 2.5 concentrations, the most areas in China suffered from the high value of the PM exceed the GradeⅡstandard (75µg/m 3 ), which actually aroused the reduction of the air visibility. 2.2 Model evaluation The time series of observed and predicted hourly concentrations about four pollutants: PM 2.5, PM 10, SO 2 and NO 2 at our station (located in Handan city) were given in Figure 2, which highlights a consistent over- and under-prediction.
Figure 2: the four pollutants observed hourly-mean concentration compared with the predicted concentration in the monitoring site. It was worth noting that the air quality in Handan city was very serious during the January in 2013, with the some serious under-predictions for pollutants occurring. The Handan city was chosen for the representative city on account of its similar variation for pollution compared with the most areas in NCP, the central regions in China, Sichuan Basin, Yangtze River delta (YRD) and the Pearl River delta (PRD). The performance statistics of the four pollutants hourly-average concentrations, as SO 2, NO 2, PM 2.5, PM 10, at a 4-km horizontal grid resolution show in Table 1. Table 1: Performance statistics of the four pollutants hourly-mean concentrations at a 4-km horizontal grid resolution Variables SO 2 (ppb) NO 2 (ppb) PM 2.5 (μg m -3 ) PM 10 (μg m -3 ) n 744 744 727 727 Max. Obs. 335.61 82.76 824.95 1280.98 Min. Obs. 11.56 6.42 14.06 42.52
Obs. 78.19 45.48 230.65 345.19 Max. Sim. 371.77 139.69 953.05 1121.06 Min. Sim. 11.19 13.39 29.47 33.28 Sim. 79.72 57.44 322.62 360.46 Corr. Coeff. 0.198 0.246 0.225 0.116 RMSE 66.98 24.75 217.83 256.60 MFB -0.103 0.219 0.442 0.119 MFE 0.621 0.371 0.686 0.582 NMB 0.020 0.263 0.399 0.044 NME 0.633 0.415 0.758 0.555 n Number of data pairs. Max. Obs.- maximum observations. Min. Obs.- minimum observations. Max. Sim.- maximum simulations. Min. Sim.- minimum simulations. Obs. mean observation. Sim. mean simulation. Corr. Coeff. correlation coefficient. RMSE the root mean square error; MFB the mean fractional bias; MFE the mean fractional error; NMB the normalized mean bias; NME the normalized mean error. Overall, the diurnal variations and magnitudes of the NO 2 mixing rations were captured by the model. At that time, the predicted concentration of NO 2 was significantly overestimated. The statistics for NO 2 show much larger bias and error when compared to the same statistics for SO 2 due to the generally higher sensitivity of NO 2 to error in emission and meteorology, especially under stagnant conditions. As gaseous pollutants, the mean monthly NO 2 concentration for observation was 45.48 ppb, which predicted concentration was 57.44 ppb, with a NMB of 26.3% and a MFB of 21.9%. Regarding SO 2, it was underestimated with a MFB of -10.3% and a NMB of 20%, which appeared a related well performance. As mentioned PM 10 and PM 2.5, they present similar variation. For PM 10, the CMAQ accurately captured the temporal variations in the PM 10 concentrations. The rank of the observed concentration by region is between 42.52μg/m 3 and 1280.98μg/m 3, with the predicted value is from 33.28μg/m 3 to 1121.06 μ g/m 3. At that time, the observed and simulated monthly averages were 345.19 μ g/m 3 and 360.46 μ g/m 3, respectively, with a NMB of 44%, a NME of 55.5%, a MFB of 11.9% and a MFE of 58.2%. Analogously, the most of the predicted concentration of PM 2.5 was overestimated except some heavily pollution episodes. The rank of the predicted PM 2.5
concentration was from 29.47μ g/m 3 to 953.05μ g/m 3, which the observed value was between 14.06μ g/m 3 and 824.95μ g/m 3. Although the peak of the PM concentration was extremely higher than usual, the monthly mean concentrations of PM were nearly close to each other. What s more, the statistics for PM 2.5 showed that a relatively good correlation between observed and predicted value, with a NMB of 39.9%, a NME of 75.8%, a MFB of 44.2% and a MFE of 68.6%. According to the Attainment Modeling Guidance (EPA, 2007) for fine particulate matter, the MFB, MFE, NMB, NME were recommended as the performance statistical indicators. Fractional bias and mean fractional error were selected as the most appropriate metrics to summarize model performance (Boylan and Russell, 2006) and an MFB of less than or equal to ±60% and an MFE of less than or equal to 75% was recommended for PM 2.5 as the model performance criteria. It is worth mentioning that the MM5-Model-3/CMAQ modeling was feasible and acceptable except for the PM 2.5 with a NME of 75.8%, though the hourly average PM concentrations were up to an unimaginable peak. Several factors could contribute to the under-predicted of PM mass concentration, including uncertainties in emissions and the model treatment of chemistry and thermodynamics of aerosols. References Wu, D., Wu, X.J., Li, F., Temporal and spatial variation of haze during 1951-2005 in Chinese mainland, Acta Meteorologica Sinica, vol. 5, no. 68, pp. 680-688, 2010 (in Chinese). CMA, Specifications for the Surface Meteorological Observations. Meteorological Press, Beijing, China, 2003 (in Chinese). Wang, L.T., Xu, J., Yang, J., Understanding haze pollution over the southern Hebei area of China using the CMAQ model, Atmospheric Environment, vol. 56, no. 1, pp. 69-79, 2012. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality Goals for Ozone, PM 2.5, and Regional Haze; U.S. Environmental Protection Agency; Office of Air Radiation/Office of Air Quality Planning and Standards: Research Triangle Park, NC, 2007. Boylan, J. W. Russell, A.G., PM and light extinction model performance metrics, goals, and criteria for three-dimensional air quality models, Atmospheric Environment, vol. 40, no. 26, pp. 4946-4956, 2006.