Modeling of air quality with a modified two-dimensional Eulerian model: A case study in the Pearl River Delta (PRD) region of China

Transcription

1 Journal of Environmental Sciences 19(2007) Modeling of air quality with a modified two-dimensional Eulerian model: A case study in the Pearl River Delta (PRD) region of China CHENG Yan-li, BAI Yu-hua, LI Jin-long, LIU Zhao-rong College of Environmental Sciences, Peking University, Beijing , China. ylcheng@pku.edu.cn Received 29 May 2006; revised 13 July 2006; accepted 14 August 2006 Abstract A modified two-dimensional Eulerian air quality model was used to simulate both the gaseous and particulate pollutant concentrations during October 21 24, 2004 in the Pearl River Delta (PRD) region, China. The most significant improvement to the model is the added capability to predict the secondary organic aerosols (SOA) concentrations because of the inclusion of the SOA formation chemistry. The meteorological input data were prepared using the CALMET meteorological model. The concentrations of aerosol-bound species such as NO 3, NH 4 +, SO 4 2, and SOA were calculated in the fine particle size range (<2.5 µm). The results of the two-dimensional model were compared to the measurements at the ground level during the PRD Intensive Monitoring Campaign. Overall, there were good agreements between the measured and modeled concentrations of inorganic aerosol components and O 3. Both the measured and the modeled results indicated that the maximum hourly O 3 concentrations exceeded the China National Air Quality Standard. The predicted 24-h average SOA concentrations were in reasonable agreement with those predicted by the method of minimum OC/EC ratio. Key words: air quality model; numerical simulation; secondary organic aerosol; inorganic aerosol Introduction Fine particulate matter with aerodynamic diameters less than 2.5 µm (PM 2.5 ) is of research and regulatory interest because of the impacts that such particles have on human health, acid deposition, atmospheric visibility, and the earth s radiative budget. On a regional scale, the secondary components of PM 2.5, including sulfate, nitrate, ammonium, and organic carbon (OC), typically constitute the majority of the PM 2.5 mass. Among these components, organic aerosols are the least understood. Organic aerosols consist of both primary and secondary compounds. The former is emitted directly into the atmosphere from emission sources, and the latter is produced via photo-oxidation of hydrocarbon precursors. The secondary component of organic aerosols is referred to as secondary organic aerosol (SOA). SOA formation depends on the atmospheric mix of anthropogenic and biogenic precursors, the formation of condensable products by the gas-phase oxidation of these precursors, and the subsequent gas/particle partitioning of those products. Although a number of laboratory studies have been conducted to determine the aerosol-forming potential of individual hydrocarbons, it is still difficult to directly identify the two components (Odum et al., 1997; Griffin et al., 1999; Kamens and Jaoui, 2001). Project supported by the National Natural Science Foundation of China (No ) and the National Basic Research Program of China (No. 2002CB410802, 2002CB410801). * Corresponding author. yhbai@pku.edu.cn. Several modules have been developed for the prediction of SOA formation in atmospheric models (Strader et al., 1999; Pun et al., 2002; Griffin et al., 2002). In most cases, aerosol species are generally assessed as independent passive tracers, and no interaction with their gaseous environment is taken into account. This limiting constraint is overcome with the use of an aerosol module, which may simulate, in some way, gas-particle interactions and the chemical and size evolution (ageing) of the resulting heterogeneous aerosols (Cousin et al., 2005). Until now, general efforts have been made to model gaseous pollution, however, there is at present an urgent need for incorporating aerosol processes in air quality simulations. For this purpose, a two-dimensional air quality model including gas-phase chemistry and inorganic aerosol chemistry developed in the laboratory is extended. The PRD (Pearl River Delta) region is one of the fastest growing regions in China. Rapid industrialization and urbanization in the region has resulted in a drastic increase in pollutant emissions into the air. Particulate matter (PM) levels have increased in the past few years because of the increase of motor vehicles, urban construction, heating installation, and industrial combustion. In addition to these primary emissions, secondary anthropogenic and biogenic aerosols formations are also of concern. All these conditions have resulted in a unique opportunity to test the modified two-dimensional air quality model. In the present study, an organic aerosol module, which may simulate in some way gas-particle interactions, was

2 No. 5 Modeling of air quality with a modified two-dimensional Eulerian model: A case study in the Pearl River Delta 573 implemented in the 2-dimensional air quality model. The modified model has been tested and improved using the data during an episode of the PRD IMC (Intensive Monitoring Campaign). The PRD IMC research focused on the production, transport, and distribution of ozone and other photo-oxidants over the Pearl River Delta Region of China. The PRD IMC observation period was from October 1 to November 5, 2004 during which extensive measurements were taken for gaseous pollutants, including SO 2, O 3, CO, NO, NOy, VOCs, and particle-bound pollutants, including SO 4 2, NO 3 and NH 4 +, TC (total carbon), OC (organic carbon) and EC (element carbon). 1 Model descriptions Initially developed by Li et al. (1988) and later modified by Tang (2003), the 2-dimensional air quality model is a multi-scale Eulerean atmospheric chemistry model capable of predicting the ground-level concentrations of gas-phase pollutants and inorganic aerosol components. It comprises three modules: meteorological module, emission module, and chemical transport module. The meteorological module provides the meteorological fields, and the emission module produces emissions required for the chemical transport module. In this study, the 2- dimensional air quality model was enhanced by adding nine new reaction mechanisms to the CBM-IV (Gery et al., 1988) to allow the prediction of SOA formation. The key parts of the model are discussed below. 1.1 Meteorological model The meteorological data required as input for the photochemical model were generated from a CALMET meteorological model, which adjust input winds to a Lambert Conformal Projection coordinate system to account for the Earth s curvature. The diagnostic wind field module uses a two step approach to compute the wind fields (Douglas and Kessler, 1988). The basic meteorological quantities required by CALMET are time and space dependent surface measurements of pressure, temperature, relative humidity, wind speed, and direction. The hourly observations of parameters were obtained from nine surface weather stations. Details of the CALMET meteorological model were described by Scire et al. (2000). 1.2 Photochemical model The 2-dimensional air quality model is an Eulerian photochemical grid model that allows integrated assessment of gaseous and particulate air pollution in multiple scales ranging from urban to super-regional. The host model simulates gas-phase chemistry, dry deposition, and primary emissions. The CBM-IV gas-phase chemical mechanism has been extended to include the formation of organic precursors. In this study, secondary organic aerosol models based on the Odum/Griffin module (Pun et al., 2003) were incorporated into this framework. A detailed aerosol speciation has been adopted to better represent the aerosol composition. Secondary organic aerosols are formed from condensable volatile organic compounds using aerosol yields (Moucheron and Milford, 1996) and partition coefficients (Pankow, 1994). For anthropogenic VOCs, two condensable products were added to the existing reactions of each of the aromatic species: TOLPM1 and TOLPM2 (high secondary organic aerosol yield products) for toluene oxidation, and XYLPM1 and XYLPM2 (low secondary organic aerosol yield products) for xylene oxidation. The biogenic VOCs were calculated by α-pinene, β- pinene, caryophyllene, and humulene reactions (Table 1). Here, pinenes (α-pinene, β-pinene) are the most abundant monoterpenes in biogenic emissions; caryophyllene and humulene represent sesquiterpenes. Since aromatic and biogenic compounds are already represented in the original lumped structure formulation of CBM-IV to simulate O 3 formation, the reactions added for biogenic SOA formation do not alter the O 3 chemistry. Accordingly, OH, O 3, and NO 3 are artificially regenerated in the reactions forming SOA, and the addition of these reactions has no net effect on the original O 3 chemistry (Pun et al., 2003). Furthermore, equilibrium inorganic aerosol modules are included in the model modified by Tang (2003). In the equilibrium approach, a thermodynamic inorganic model, ISORROPIA by Nenes et al. (1998) determines the bulk inorganic mass to be transferred between gas and aerosol phases. The inorganic aerosol module calculates the transformations of SO 2 to sulfate via gaseous reactions, and gaseous HNO 3 to aerosol nitrate. Ammonium is considered a primary species converted into the aerosol phase by ammonia neutralization to nitric and sulfuric acids. In this study, a particle diameter range of less than 2.5 µm was Table 1 Modification to CBM-IV Anthropogenic reactions with new products TOL+OH 0.08XO2+0.36CRES+0.44HO2+0.56TO TOLPM TOLPM2 XYL+OH 0.7HO2+0.5XO2+0.2CRES+0.8MGLY+1.1PAR+0.3TO2+0.18XYLPM1+0.79XYLPM2 New biogenic reactions Rate constants a (cm 3 /(molecule s)) Rate constants a (cm 3 /(molecule s)) α-pinene APIN+OH 0.231APIPM1+1.98APIPM2+OH APIN+O APIPM3+0.62APIPM4+O β-pinene BPIN+OH 0.79BPIPM1+0.25BPIPM2+OH BPIN+O BPIPM3+2.94BPIPM4+O BPIN+NO BPIPM5+NO Caryophyllene CRP+OH 9.11CRPPM+OH Humulene HUM+OH 9.2HUMPM+OH a Reference Lamb et al., 1999.

3 574 CHENG Yan-li et al. chosen for these secondary aerosol species. 2 Simulation region descriptions The main part of the Pearl River Delta Region is in the Guangdong Province. It also includes Hong Kong and Macao. The region has a total land area of km2 and a population of over 38 million. It is situated in a transitional zone of the East Asian monsoon system, where the southwesterly summer monsoon arrives from the oceans (South China Sea and the tropical Pacific), while the northeasterly winter monsoon arrives from Mainland China. The region has humid subtropical weather with an annual average temperature of 22 C and rainfall of 1690 mm (Cao et al., 2004). The model domain contains grid cells, with a horizontal resolution of 15 km. The center of the domain is located at E, N (Fig.1). The anthropogenic emissions data were compiled from various sources including the emissions from traffic, industrial processes, and residential areas. The biogenic emissions were included in the emission inventory as well (Hu, 2000). Both boundary conditions and initial conditions of the chemical species were derived from the Vol. 19 results of the measurements and the previous model runs (Tang, 2003). 3 Results and discussion In this study, the modified 2-dimensional air quality model was used to simulate the air quality in the PRD region during the period of October 21 24, Data from the first two days (October 21 22) was used to initialize the model. Therefore, only the results of October 23 24, 2004 are discussed in this article. In general, a northeasterly or northerly wind almost covered the entire domain during the simulation period. The wind speed measured at the Xinken. Meteorological Station was considerably higher on October 23 than that on October 24 (The 24-h averages were 1.3 and 0.1 m/s, respectively). Contrarily, the wind speed at Guangzhou on October 23 was lightly lower than that on October 24 (The 24-h averages were 1.6 and 2.0 m/h, respectively). The wind fields simulated by CALMET are shown in Fig.2. The calculated wind directions are in quite good agreement with the measurements for the two days. 3.1 O3 concentrations Ozone is a secondary air pollutant produced by the chemical reactions of nitrogen oxides (NOx) and volatile organic compounds (VOCs) in the presence of sunlight. It is hazardous to human health and plant growth. Fig.3 shows the temporal profiles of ozone for two sites in the PRD region. The measured and predicted concentrations agree with each other reasonably well at both locations on October 23. However, the model overpredicted the peak concentration in Guangzhou and underpredicted it in Xinken on October 24. These results may have been caused by the overestimation/underestimation of the vertical mixing as well as of the deposition processes. The measurements showed a maximum O3 concentration of 263 µg/m3 at Xinken on October 24, The predicted maximum 1-h O3 concentration was 197 µg/m3 on October 24, 2004 (approximately 25% bias). Fig. 1 Simulation domain. 1: Guangzhou station; 2: Xinken station. Fig. 2 Afternoon (15:00) wind fields at the CALMET simulation (a) for October 23 and (b) for October 24.

4 No. 5 Modeling of air quality with a modified two-dimensional Eulerian model: A case study in the Pearl River Delta 575 Fig. 3 Temporal O 3 profiles at Guangzhou (a) and at Xinken (b), 23 to 24 October (Dots: measurement and solid lines: simulation). 3.2 Particle components simulation Inorganic aerosol Fig.4 compares the simulated and observed 6-h average aerosol-phase concentrations for sulfate, nitrate, and ammonium for two sites in the PRD region. Here, the concentration values were obtained by averaging all both day values over the same time period. (1) Sulphate concentrations: As shown in Fig.4a, the model overpredicted the sulfate levels in Guangzhou by 46% in the early part of the day but underpredicted the levels by 8% 50% in the latter part of the day. The model overestimated the sulfate levels in Xinken. Several reasons can explain these discrepancies. First, this is most likely caused by the uncertainties of the emission inventory. In the Guangzhou City, with strong industrial and urban sources and intense traffic, sulphate formation occurs rapidly near the emission sources. Another possibility can be heterogeneous sulphate formation at the particle surfaces in connection with the high-aerosol concentrations in these plumes. The model did not simulate these processes explicitly and, thus, caused underestimation. Relative humidity (RH) is another probable cause of the biases. Cousin et al. (2005) observed that the mean sulfate concentration increased with the increased RH. In this study, the meteorological analyses forced the model to provide low RH at Guangzhou stations and high RH at Xinken. Homogeneous sulphate production is considerably slower and less efficient than heterogeneous aqueous production. The high RH values for Xinken resulted in the domination of heterogeneous reactions and, therefore, overestimated the sulfate concentrations there. Fig. 4 Sulphate (a), nitrates (b) and ammonium (c) aerosol concentrations for different time periods during simulations at: (a1, b1, c1) Guangzhou; (a2, b2, c2) Xinken.

5 576 CHENG Yan-li et al. Vol. 19 Fig. 5 SOA model results (solid lines) for October at Guangzhou (a) and Xinken (b) as a function of time. (2) Nitrates concentrations: the model underestimated the nitrate aerosol concentrations in both locations. The predicted and measured particulate nitrate concentrations show similar diurnal trends: high in the morning and low in the afternoon, especially at Guangzhou. The meteorological condition during morning was propitious for the transformations of gaseous nitrate to aerosol. Some studies have supposed that terrigenous mineral aerosols can react with and neutralise gaseous HNO 3 and contribute to the formation of nitrate aerosols (Dentener et al., 1996; Tabazadeh et al., 1998). In this study, carbonates were not considered in the inorganic aerosol model, which is probably one of the major factors for the underestimation of nitrates. (3) Ammonium concentrations: simulated and observed diurnal behaviors of ammonium are similar to those of sulphate. Particulate NH 4 + was influenced by the NH 3 concentration. The bias in the simulated ammonium concentrations is mainly because of the uncertainty of the NH 3 emission data Secondary organic aerosol One of the major goals of this study is to compare the SOA levels predicted by the 2-dimensional model with those measured in the PRD region. Only the simulated SOA results are shown in Fig.5, as the hourly concentrations of OC were not measured during the period of simulation. A significant peak of SOA concentration is predicted in the afternoon of this day. The predicted 24- h average SOA concentrations for this period are 7.7, 7.8 µg/m 3 at Guangzhou and 4.4, 10.5 µg/m 3 at Xinken, respectively. The 24-h average SOA concentrations estimated by the method of minimum OC/EC ratio (Turpin and Huntzicker, 1995; Castro et al., 1999) for the same period and sites are 9.4 µg/m 3, 10.5 µg/m 3 and 3.3 µg/m 3, 9.1 µg/m 3, respectively. The difference may be because of the uncertainties of the method of minimum OC/EC ratio. The predicted contributions of the main precursor groups to the predicted 24-h average total SOA mass show that biogenic SOA are the dominant component of the total SOA in the modeled region, accounting for about 57% of the SOA. SOA concentrations in PM 2.5 predicted by this model are comparable with the nitrate and ammonia concentrations. 4 Conclusions A photochemical air pollution episode in the PRD region during October 21 24, 2004 was simulated by a modified two-dimensional air quality model, which includes both gas-phase and aerosol chemistry. The meteorological input data were generated by the CALMET meteorological model. The simulation results were compared to the field measurement data obtained from the PRD region Intensive Monitoring Campaign. The results showed overall good agreement between the measured and modeled time-varying concentrations of inorganic aerosol components and ozone. Owing to the lack of hourly OC measurements, only simulated SOA concentrations are presented. The predicted 24-h average SOA concentrations were in reasonable agreement with those predicted by the method of minimum OC/EC ratio. Furthermore, both measured and modeled maximum hourly O 3 concentrations exceeded the China National Air Quality Standard. Even though the modified model performed reasonably well, the meteorological model and emission inventory still need to be further improved. The aerosol chemistry module, which is still under development seems promising. More measurements on aerosol species are required for model evaluation. Acknowledgements The authors thank all the participants in the Pearl River Delta Region Intensive Monitoring Campaign in October 2004 and also thank the Panyu Meteorological Office for providing the meteorological data for the CALMET calculations and the Guangdong Provincial Environmental Monitoring Station for providing the emission sources data. References Cao J J, Lee S C, Ho K F et al., Spatial and seasonal variations of atmospheric organic carbon and elemental carbon in Pearl River Delta Region, China[J]. Atmos Environ, 38: Castro L M, Pio C A, Harrison R M et al., Carbonaceous aerosol in urban and rural European atmospheres: estimation of secondary organic carbon concentrations[j]. Atmos

6 No. 5 Modeling of air quality with a modified two-dimensional Eulerian model: A case study in the Pearl River Delta 577 Environ, 33: Cousin F, Liousse C, Cachier H et al., Aerosol modelling and validation during ESCOMPTE 2001[J]. Atmos Environ, 39: Dentener J F, Carmichael G R, Zhang Y et al., Role of mineral aerosol as a reactive surface in the global troposphere[j]. J Geophys Res, 101: Douglas S, Kessler R, User s guide to the diagnostic wind field model (Version 1.0)[R]. Systems Applications Inc., San Rafael, CA, 48. Gery M W, Whitten G Z, Killus J P, Development and testing of CBM-IV for urban and regional modeling[s]. EPA Publication No. EPA-600/ U.S. Environmental Protection Agency, Research Triangle Park, NC. Griffin R J, Cocker III D R, Flagan R C et al., Organic aerosol formation from the oxidation of biogenic hydrocarbons[j]. J Geophys Res, 104: Griffin R J, Dabdub D, Kleeman M J et al., Secondary organic aerosol: 3. Urban/regional scale model of size and composition-resolved aerosols[j]. J Geophys Res, 107: Hu Y T, Study on regional air quality and its impact factors[d]. Peking University Doctor of Science Dissertation. Kamens R M, Jaoui M, Modeling aerosol formation from α-pinene + NOx in the presence of natural sunlight using gas-phase kinetics and gas-particle partitioning theory[j]. Environ Sci Technol, 35: Lamb B, Grosjean D, Pun B et al., Coordinating Research Council[R]. Alpharetta, GA, NTIS PB Li J L, Zhang Q S, Tang X Y et al., A mathematical model of photochemical pollution at Xigu distruct in Lanzhou[J]. Acta Scientiae Circumstantiae, 8(2): Moucheron M C, Milford J, Development and testing of a process model for secondary organic aerosols[m]. Nashville: Air and Waste Management Association. Nenes A, Pandis S N, Pilinis C, ISORROPIA: a new thermodynamic equilibrium model for multiphase multicomponent inorganic aerosols[j]. Aquatic Geochemicals, 4: Odum J R, Jungkamp T P W, Griffin R J et al., Aromatics, reformulated gasoline, and atmospheric organic aerosol formation[j]. Environ Sci Technol, 31: Pankow J F, An absorption model of gas/particle partitioning of organic compounds in the atmosphere[j]. Atmos Environ, 28: Pun B K, Griffin R J, Seigneur C et al., Secondary organic aerosol: 2. Thermodynamic model for gas/particle partitioning of molecular constituents[j]. J Geophys Res, 107: 4333, doi: /2001JD Pun B, Wu S Y, Seigneur C et al., Uncertainties in modeling secondary organic aerosols: Three-dimensional modeling studies in nashville/western tennessee[j]. Environ Sci Technol, 37: Scire J S, Robe F R, Fernau M E et al., A user s guide for the CALMET meterorological model (version 5)[R]. Earth Tech Inc. Concord, MA. Strader R, Lurmann F, Pandis S N, Evaluation of secondary organic aerosol formation in winter[j]. Atmos Environ, 33: Tabazadeh A, Jacobson M Z, Singh H B et al., Nitric acid scavenging by mineral and biomass burning aerosols[j]. Geophy Res Lett, 25: Tang X G, A simulation study on secondary pollutant in regional atmosphere[d]. Peking University Master of Science Dissertation. Turpin B J, Huntzicker J J, Identification of secondary aerosol episodes and quantification of primary and secondary organic aerosol concentrations during SCAQS[J]. Atmos Environ, 29: