The Extragalactic Reference

© 2010 Zhang et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Correction Diagnosis and phylogenetic analysis of Orf virus from goats in China: a case report


Introduction
The Pearl River Delta and adjoining Hong Kong metropolitan area of China, like virtually all other major urban-industrialized regions of the world, suffers from photochemical smog characterized by the unhealthily high concentrations of ozone (O 3 ) and fine par-Introduction vations were performed than at the other sites. It is located in a sparsely populated coastal area with light local emissions, on northwestern Lantau Island roughly in the north-south centerline of the Pearl Estuary with Hong Kong's urban center of 32 km 5 to the east and Macau/Zhuhai to the west at about the same distance. The area is surrounded by major urban centers in the PRD region, making it a good location to characterize local and regional emissions and the photochemical evolution of urban plumes. TO was selected as the location for the Supersite because available air quality data revealed that the highest ozone concentrations during air pollution episodes 10 are generally found in the western part of Hong Kong (Kok et al., 1997;Wang et al., 2001a). The measurement site is located inside an inactive barracks, at an elevation of ∼80 m above sea level. Information about the instruments can be found in  and Zhang et al. (2004). Longer term (18-month) data on various trace gases from this site were analyzed by Wang et al. (2005) and Simpson et al. (2006), 15 and fine aerosol data in December 2002 and pollution episodes observed in the earlier periods were examined by Cheung et al. (2005); Wang and Kok (2003); and Wang et al. (2003). Trace gases including O 3 , NO, NO 2 , CO, SO 2 and NO y were measured continuously at most of the sites, as were meteorological parameters. Among these species/parameters, O 3 , NO, CO, and temperature, together with organic compounds, are the critical inputs to the Observation Based Model (OBM) used to assess the sensi-5 tivity of local O 3 photochemical production to changes in the concentrations/emissions of NO x and VOC (Sect. 3.2). In addition, CO, SO 2 and NO y can be used as tracers of specific types of pollution and thus can provide valuable insights into the specific sources of pollutants that contribute to concentrations measured at the site (e.g. pollution from Hong Kong versus pollution from Guangdong). However, among the species 10 needed for the OBM, CO was not measured at CW and YL, NO was not measured at YL, and CO measurements at TC and TM were made using instrumentation that lacked sensitivity to quantify CO concentrations when these concentrations are relatively low (i.e., <1 ppmv), because the EPD CO instruments were designed to monitor for compliance with the Hong Kong ambient air quality standard for CO (8-h average 15 of 8.7 ppmv). In order to carry out the OBM analysis for these sites, the required data were extrapolated from related measurements as described below. In the case of CO, sufficiently sensitive determinations of the 24-h averaged CO concentrations at CW, YL, TC, and TM were made using whole air samples collected from these sites (see Sect. 2.4 below). These 24-h averaged CO measurements were used to set the magni-

EGU
OBM analyses were not significantly affected by these approximations. At YL, the NO concentrations required in the OBM were estimated from the NO 2 and related data obtained for YL using the chemical mechanism and iterative algorithms already contained in the OBM. Sensitivity calculations in which the NO concentrations were allowed to vary within reasonable limits at YL indicate that the conclusions 5 reached using the OBM calculations were again not substantively affected by these approximations. The NO instruments at all the sites had enough sensitivity to measure the relatively high concentrations of NO during the episode days.
In addition, aerosol optical properties including light scattering and light absorption were measured during most of the measurement period at Tai O. These measurements 10 were used to identify whether significant statistical correlations existed between ozone and particulate matter pollution.

Collection of whole air samples and analysis for VOCs and CO
The ambient concentrations of C 1 -C 10 hydrocarbons, CO, and C 1 -C 2 halocarbons at each of the sites were determined on selected days using whole air sampling canisters 15 with subsequent analysis using gas chromatography. The samples were then shipped to and analyzed at the University of California -Irvine. The detailed descriptions of the analytical techniques are given in Colman et al. (2001).
As noted above, hourly speciated VOC and CO concentrations are needed as input to the OBM for a given site. However, the VOC analyses made for the 4 EPD sites 20 were based on ambient whole air samples gathered over a 24-h period, and thus their use in the OBM required that estimates of hourly VOC concentrations at each site be derived. The derivation of diurnal VOC profile is presented in Sect. 3.2.2. In the case of the VOC concentrations at TO, while individual samples were generally collected at several times during the day, the frequency of collection was not hourly (see above). 25 For this site, hourly VOC concentrations were estimated through a simple linear interpolation. Once again, it should be noted that sensitivity calculations with the OBM (in this case using constant, 24-h concentrations throughout the day) indicated that the 8967 Introduction EGU approximations involved in estimating the diurnal variation in the VOC concentrations did not substantively affect our results.

Results and analyses
3.1 Data and diagnostic analysis 3.1.1 O 3 Pollution episodes 5 Figure 1 illustrates the time series of measurements made at TO during the field measurement period. For the purposes of the analyses discussed here, we define an O 3 episode day as a day when the peak one-hour averaged O 3 concentration at TO (the site where O 3 was generally the highest) exceeded 100 ppbv. (There were not any days during the measurement campaign when O 3 exceeded 100 ppbv at one of 10 the other four sites but did not exceed 100 ppbv at TO, so this definition is inclusive.) Inspection of Fig. 1 reveals that during the campaign, 9 O 3 episode days were encountered. Of these 9, 4 occurred during a multiple-day episode spanning 9-12 October (O 3 maximum = 149 ppbv), 3 occurred during a multiple-day episode spanning 5-7 November (O 3 maximum = 203 ppbv). The November 7th episode day was especially 15 interesting as the peak O 3 on that day of 203 ppbv is the highest O 3 concentration ever reported for the Hong Kong/PRD region to date. The two other individual O 3 episode days occurred on 25 October and 12 November. Further inspection of Fig. 1 reveals that the O 3 episode days occurred when the total UV was high, the wind speed was low, and the wind direction was generally from 20 the north/northeast/northwest, which is consistent with previous observations in the area . On some episode days SO 2 or NO y was high (e.g. 11 October and 7 November), but not on others (e.g. 9 October). CO was generally high on episode days, for example 9-11 October, 25 October, and 5-7 November. Aerosol absorption was high on 10-11 October and 7 November. On 7 Introduction

EGU
November, when O 3 reached its record high of 203 ppbv, SO 2 , NO y , CO, TUV and light absorption were all relatively high. By contrast, note that on 30 November, CO, SO 2 , NO y , and aerosol absorption were all high and the wind direction was favorable for the import of O 3 precursors into the Hong Kong area, and yet, probably because of low temperatures and overcast conditions (i.e., low total UV), an O 3 episode did not occur.
In the analyses presented here, we will focus on the aforementioned 9 O 3 episode days as well as an additional near-O 3 episode day (8 November) when O 3 peaked at 91 ppbv to see if there are any striking differences on this day as compared to the actual episode days. Relevant episode characteristics on each of these days are listed in Table 1. For example, VOC data were obtained at TO on all days but 12 October, and 7 November was the only day when VOC data was obtained at the EPD-operated sites as well as at TO. For this reason, and because the O 3 recorded on 7 November was the highest on record for the area, the data gathered on this day are subjected to the most comprehensive analysis.

Diurnal variations of O 3 and related species
15 Figure 2a shows the diurnal O 3 variations on the episode days at TO. As is typical of urban areas, O 3 is low at night and in the morning (presumably due to NO titration and dry deposition) and peaks during the afternoon (as a result of photochemical production and to a less extent downward mixing of ozone from above the boundary layer). There are some notable features of the O 3 variation at TO. First, O 3 does not 20 begin to increase until relatively late in the morning, and secondly, O 3 appears to peak relatively early and begins to decline relatively early in the afternoon, which is similar with the observations of . This behavior could be attributed to high concentrations of NO (see later discussion), a short period of intense sunlight (the episodes are occurring well after the fall equinox), and transport of air masses of  EGU field campaign, O 3 was highest on Lantau Island (TC, TO), and higher at TO than at TC. The diurnal variations of O 3 observed at the EPD-operated sites are similar to that observed at TO in that the O 3 increase begins relatively late in the morning. However, unlike TO, only the YL site also showed an O 3 peak relatively early in the afternoon; the TC, CW and TM sites showed peaks closer to 18:00. Figure 3 illustrates the diurnal variations of other primary and secondary pollutants at TO on the O 3 episode days. There are two aspects of Fig. 3 that bear noting. The first is the generally large concentrations of NO that were encountered; daytime NO concentrations of several ppbv were the norm. Similarly high NO concentrations were also encountered at the other sites. As later discussion and analyses will show, 10 these high NO concentrations directly lead to two important conclusions: (1) HONO concentrations in the early morning hours may be quite high and may significantly enhance O 3 production; and (2) photochemical O 3 production in the area during the episode days was strongly VOC-limited.
The second feature of note is the anomalously high early morning concentrations of 15 NO and VOCs (as well as CO and aerosol absorption, not shown) on 11 October and 7 November. These high concentrations suggest that TO was subject to an unusually large amount of pollution during the early morning on these two days due to unusual nighttime transport patterns that brought urban emissions to TO.

Transport characteristics of O 3 episode days 20
One of the key questions to be addressed in this study is the relative roles of local emissions and emissions transported from Guangdong Province in producing O 3 pollution in Hong Kong. To address this question, we attempted to characterize the degree to which the air masses encountered in Hong Kong on the O 3 episode days were impacted by air transported from Guangdong Province. Each day was assigned one of 25 three possible categories: "L", indicating that the episode was largely local in character; "L, r", indicating a local episode with some impact from Guangdong Province; and "R", indicated an episode with significant impact from Guangdong Province. As described 8970 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
EGU below, two independent analyses were carried out to determine the appropriate category for each day. Only one of the episode days (7 November, the most severe episode) received an "R" categorization; 4 days (10,11,12 October, and 8 November) received an "L, r" categorization; and the remaining 5 days received an "L" categorization (Table  1). Thus it would appear that both local and transported pollutants can act indepen-5 dently or in concert to bring about O 3 pollution in the Hong Kong area, although the days with the highest Nonmethane Hydrocarbon (NMHC) and NO levels (7 November and 11 October; Fig. 3) were associated with some degree of regional influence. Chemical Tracers: The ratio of the enhancement of CO over background values to the enhancement of NO y (i.e., dCO/dNO y ) in an air mass is a useful diagnostic 10 indicator of the relative influence of pollution from South Mainland China versus Hong Kong (Kok et al., 1997;Wang et al., 2001a, because the emission ratio of COto-NO x from Guangdong and Hong Kong are so different; i.e., ∼15 in Guangdong and ∼1 in Hong Kong (Streets et al., 2003). Therefore high ratios are generally indicative of air masses from Mainland China and low ratios of air masses impacted by local Hong 15 Kong emissions.
In the analysis presented here, dCO/dNO y was calculated from the 1-h averaged CO and NO y measurements recorded at TO and subtracting the background CO and NO y concentrations from the observed concentrations. The background CO and NO y concentrations were estimated in two ways. A "seasonal" dCO/dNO y was calculated 20 using constant background CO and NO y concentrations of 211 and 3.37 ppbv, respectively. These concentrations are the mean values observed for marine air advecting over Hong Kong in the fall season (Wang et al., 2001b). A "24 hourly" dCO/dNO y was calculated using the minimum CO and NO y concentrations observed on that day. The diurnal variation in the two sets of ratios for each O 3 episode day can be found in Zhang 25 et al. (2004). In summary, 7 November is the only day with consistently high ratios during the late morning and early afternoon (consistent with its "R" categorization). 10 October, 11 October, 12 October and 8 November have transient spikes of high ratios and thus have "L, r" categorizations. On the other episode days, the ratios remained Introduction EGU low throughout the photochemically important period and thus have "L" categorizations. Back trajectories: As an independent check on the results obtained from the dCO/dNO y ratios, back trajectories for air masses at TO during each of the O 3 episode days were calculated. Figure 4 shows the trajectories for 10 October, 7 November, and 12 November illustrating a case of "L, r", "R", and "L", respectively. The back trajec-5 tories were calculated using the NOAA/ARL HYSPLIT4 model (HYbrid Single-Particle Lagrangian Integrated Trajectory model, version 4.6) driven by wind data with a 3-km resolution calculated by NCAR/PSU Mesoscale Meteorological Model (MM5 version 3.6) (Ding et al., 2004). These plots generally corroborate the results from the ratio dCO/dNO y . To facilitate the analysis of the distribution and speciation of VOCs observed at the various sites during the 7 November O 3 pollution episode, we have grouped the 40+ species measured at each site into three major types: CO, anthropogenic hydrocarbons (AHC), and biogenic hydrocarbons (BHC), with BHC defined as isoprene and 15 the pinenes. AHC are further divided into five sub-types according to their structure and reactivity with the OH radical in the atmosphere: reactive aromatics (R-AROM) encompassing all aromatics except benzene; reactive olefins (R-OLE), comprising all olefins except ethylene; alkanes with four or more carbons (>=C4); ethylene (ETH); and the low reactivity hydrocarbons (LRHC) which include methane, ethane, propane, 20 acetylene and benzene.
Based on the 24-h averaged concentrations of the major VOC groups and AHC subgroups measured at the five sites on 7 November, AHC clearly dominates over BHC at all sites (Fig. 5). Note in the figure that methane is not included in AHC since its extremely low reactivity though its high concentration. The AHC levels occur in the order 25 YL>TO>TC>CW>>TM. Although TC and TO are downwind of CW and YL, the AHC levels at these two sites are higher than at CW, and the AHC level at TO is comparable to that at YL. This suggests that there must be some other significant source of VOCs EGU at TC and TO other than simply transport from the urban/industrial areas surrounding CW/YL. Within the AHC group, R-AROM is the dominant subgroup, which in turn is richest in xylenes and toluene, and to a lesser extent tri-methylbenzenes and ethylbenzene. As was the case with AHC, R-AROM is highest at YL and TO followed by TC, CW and TM. The ETH levels (not shown in Fig. 5) occur in the same order as the 5 AHC levels, but only contribute about 3% of the AHC concentrations.

VOC reactivity
Not all VOCs react at the same rate. In general the rate of reactivity among VOCs can vary as much as or even more than the concentrations of the VOCs in the atmosphere.
To take both the concentration (on a C atom basis) and OH-reactivity of each VOC into 10 account, we use a reactivity scale called propy-equivalents (Chameides et al., 1992).
where C i (propy-equiv) propy-equivalents (ppbC) of any VOC species i , C i (observed) observed concentration (ppbC) of species i , k OH (i ) reaction rate coefficient (molec/cm 3 /s) of species i withradical OH, and k OH (C 3 H 6 ) reaction rate coefficient (molec/cm 3 /s) of C 3 H 6 with radical OH.
In this formulation, the concentration (weighted by the number of carbons) of each VOC species is re-normalized by a factor that is proportional to its reactivity with OH.

15
The resulting reactivity is referred to as propy-equivalents because the normalization factor used is the rate constant for propylene with OH. The choice of propylene is arbitrary and another normalization factor could be used with the same results.
Distribution of Reactivities on 7 November: Figs. 6a-b shows the propy-equivalents of different VOC groups at the five sites on 7 November, and Figs. 7a-f shows the 20 relative contributions of the major VOC groups to the total reactivity. In general the reactivity analyses reinforce the conclusions reached earlier on the basis of the VOC EGU concentrations; i.e., AHC dominated over BHC, and R-AROM was the most important AHC subgroup. However, there are a number of additional insights that can be garnered from the propy-equivalents analysis. For example, even though the CO concentrations were greater than the total VOC concentrations at all sites (Fig. 5a), the total AHC reactivity was greater than that of CO at all sites (Fig. 6a). This suggests 5 that AHC played a greater role in generating local O 3 than CO though CO might have an impact on O 3 formation on a regional scale because of its longer lifetime. Among the AHCs, R-AROM contributed most to the total anthropogenic reactivity at all sites, accounting for over 50% at CW, YL, TC and TO and close to 40% at TM (Figure 7a). Of the species included in the R-AROM group the xylenes dominated, accounting for 10 over 25% of the total reactivity of anthropogenic VOCs (including CO) at TC and TO and over 18% of that at CW and YL. Toluene follows the xylenes in the R-AROM group (Fig. 7b). At YL, R-OLE also made a significant contribution to the total reactivity from anthropogenic compounds (about 18%, Fig. 7e). At CW, YL, TC, and TO, the contributions from >=C4 were also high (about 14%, Fig. 7f). ETH and LRHC contributed 15 much less to the total anthropogenic reactivity, with ETH contributed only 2-3% and LRHC 2-4% (not shown).
The results from TM are interesting because even though TM is a rural site, the influence of anthropogenic compounds appears to have been significant. While the contribution of BHC to the total anthropogenic reactivity was highest at TM (about 19%,  (Fig. 7d) still dominated. The fraction of CO reactivity to the total anthropogenic reactivity was highest at TM; i.e. ∼17% as compared to about 6-9% at other four sites. This might be due to the fact that CO has a much longer lifetime than most VOCs. As more reactive VOCs are consumed by photochemistry, the relative importance of CO will tend to rise. Since TM is a rural site, one might 25 expect CO would account for a higher fraction than at other sites. On the other hand, the contribution from R-OLE was relatively high at TM (about 10%, Fig. 7e), and given the relative short lifetimes of R-OLE, this would appear to suggest that TM was directly influenced by local anthropogenic emissions. Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
EGU Reactivity at Tai O: Fig. 8 shows the total propy-equivalent reactivity at Tai O on all episode days with VOC data. 11 October and 7 November are similar in that the early morning total propy-equivalent reactivity is much higher than that observed later during the day. As the morning progresses, however, the total propy-equivalent reactivity decreases rapidly and by noon the total reactivity on these two days was close to that 5 on other episode days. As noted earlier, the anomalous behavior on 11 October and 7 November may reflect the presence of unusual meteorological conditions that caused the early-morning accumulation of pollutants at the TO site. Similar to the situation for 7 November, AHC dominated the total reactivity on all the episode days, and R-AROM was the most important subgroup of AHC (not shown).

Ozone production and relative incremental reactivity
In this section we present the results of our application of the Observation Based Model (OBM, Cardelino and Chameides, 1995) to the data collected at each of the five sites on the 7 November episode and at TO on all episode days. We begin with a brief overview of the OBM.

Introduction to OBM
The OBM uses the concentrations of primary hydrocarbons, NO, CO and O 3 , as well as meteorological data measured as a function of time at a given location as input for a coupled set of photochemical box models that calculates the total amount of ozone that is photochemically produced during the daytime at that location based on Carbon- 20 Bond IV mechanism (Cardelino and Chameides, 1995). The model also calculates the sensitivity of the O 3 production to changes in the concentrations of the precursor compounds, i.e. Relative Incremental Reactivity (RIR). Since the production of O 3 is related to the concentration of O 3 at the site, and the concentration of a precursor is essentially linearly related to its emissions, RIR can be used to assess a given emission Introduction

EGU
Internal tests can be carried out to confirm that the application of the OBM to a given dataset is appropriate. One test is to assess the consistency of the RIRs across multiple sites and/or multiple days. For example, if the standard error of the mean for the area-averaged RIRs defined in the OBM is relatively small, the calculated RIRs are more likely to be robust. Another test is to consider the magnitude of the ozone 5 production calculated by the OBM at each site and compare it to the increment of O 3 observed at the site during the episode. If the calculated O 3 production is similar in magnitude to the observed increment, it suggests that the RIR explains a significant portion of the amount of O 3 that appeared at the site and is therefore relevant to policymaking decisions with regard to emission control strategies. As noted earlier, because the OBM uses observed concentrations, it is an observation-based as opposed to an emission-based model. It complements emission-based models as part of a weight-ofevidence approach to air quality control.

Deduction of hourly VOC profiles
Since the VOC data at the four EPD sites are 24-h averages, it was necessary to derive 15 time-dependent concentrations from these averages for use in the OBM. The method for accomplishing this is based on the entraining Eulerian box model (Seinfeld and Pandis, 1997). Assuming the atmosphere in the area of interest is well-mixed, the concentration of a species is determined by its emissions, chemical reactions, deposition (dry/wet), advection, and vertical entrainment. Equations 2a-b give the resulting rate 20 of change with time in the concentration of species i : Since emission rates for the different VOCs and their chemical destruction rates are 5 unknown (in the latter case because OH is unknown), we adopt an iterative approach to determine the hourly variations in the VOC species, along with their emission and destruction rates on the basis of their 24-h averaged concentrations and other inputs. We begin by assuming an initial OH profile and an emission rate for the VOCs, then calculate the time variation in the VOC species. The resulting 24-h averaged VOC 10 concentrations thus calculated are compared to the observed 24-h average concentrations, and the emission rates are appropriately scaled to bring the calculated 24-h averages in line with the observed averages. A new OH profile is calculated on the basis of the VOC concentrations, and the process is repeated until convergence is obtained. In the results presented a 2% agreement is used as the basis for establishing consistency.
In addition to calculating the diurnal VOC profiles, this approach is also used to calculate the diurnal profiles in HONO, formaldehyde and higher aldehydes -species EGU that were not observed during the field campaign but can impact the OBM results. These calculations and their implications are discussed in Sect. 3.2.3. Emission Rates: The annual emissions of speciated anthropogenic VOCs from the NASA TRACE-P project (Streets et al., 2003) were used to set the initial VOC emission rates of the VOCs. The diurnal variations in the anthropogenic VOC emissions 5 were estimated on the basis of the source type: industry and power generation were assumed to have no diurnal variation; domestic sources were kept constant during the daytime and set to 0 at night (9 p.m.-5 a.m.); and mobile emissions were assigned a diurnal variation that differed as a function of day of the week (Cardelino, 1998). The diurnal variations of biogenic VOCs (isoprene, α-pinene and β-pinene) were estimated 10 considering the temperature and solar radiation.
Mixing height profile: The mixing height was estimated based on the principle that heat transferred from the surface to the atmosphere results in convection, vigorous vertical mixing, and establishment of a dry-adiabatic lapse rate (Holzworth, 1967). Here upper air soundings and hourly temperature were used to compute morning and after-15 noon mixing heights. The upper air soundings are reported by National Climate Data Center (NCDC) and available at the NOAA website (http://raob.fsl.noaa.gov/). We followed the approach of Holzworth (1967) with the exception that 2 • C instead of 5 • C was used to account for the temperature difference between rural and urban environments and for some initial surface heating just after sunrise while calculating the morning 20 mixing height, since Hong Kong has a latitude of 22.32 • N and the daily temperature difference between different areas is probably not as much as that at mid-latitudes. The hourly mixing height profiles were obtained by interpolating between the morning and afternoon mixing heights, considering that the mixing height increases rapidly after sunrise and slowly in early afternoon until it reaches the afternoon mixing height. 25 The diurnal variations in the total VOC reactivity at the 4 EPD sites on 7 November were therefore derived through the above method. Encouraging similarity is found between the calculated profile at YL and the observed profile at TO (two sites with similar average reactivities), and the inferred VOC diurnal profiles are utilized for OBM

OBM-calculated net O 3 production
In principle, the O 3 that appears at a given site during a pollution episode is due to local photochemical production plus the transport of O 3 that has been produced elsewhere. However, the OBM only calculates the O 3 produced locally and/or in air masses 5 with similar chemical compositions as that found locally. As a result the RIR functions calculated by the OBM only pertain to the O 3 produced locally; i.e., they represent estimates of the sensitivity of the O 3 produced locally to changes in precursor concentration/emissions, and not of the O 3 that had been transported to the site. If local production represents a significant portion of the total O 3 increase (or increment) experienced in the area, then this feature of the OBM does not represent a significant limitation. On the other hand, if local production is much smaller than the O 3 increment, then the RIR functions calculated by the OBM are of limited utility. In this section we examine this issue by comparing the net photochemical production O 3 calculated by the OBM at each site with the actual O 3 increment observed at each site over the 15 course of the episode day. Figures 9a-b illustrates the model-calculated net O 3 production and the observed O 3 increment at each of the sites on 7 November and at TO for all episode days, with and without early morning HONO and aldehydes (see below). In general the O 3 increment calculated by the OBM agreed well with observations at all sites on 7 20 November, with the exception of TC (Fig. 9a). At TO, the agreement between the OBM and observations ranged from excellent during the 11 October and 7 November episodes to much poorer during the 6 and 12 November episodes (Fig. 9b). Generally between 50-100% of the O 3 increase observed in Hong Kong during the O 3 episodes can be explained by photochemical generation within the Hong Kong area. When

25
HONO and aldehydes were omitted from the OBM the agreement with observations significantly deteriorated (Figs. 9a-b). These latter results suggest that early morning HONO and aldehydes concentrations were significant (as estimated by the OBM), and 8979 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
EGU as a result of the OH produced by these compounds in the early morning, the OBM was able to account for much/most of the ozone increments observed during episode days when HONO and aldehydes were included. Exceptions to the latter conclusion are TC on 7 November and TO on 9 October, 6 November, and 12 November when the calculated O 3 production was significantly less than the observed increment even 5 with early morning HONO and aldehydes included. The role of HONO and aldehydes in early morning chemistry: Previous investigators have found that the presence of HONO and/or aldehydes in the urban atmosphere in the early morning hours can significantly enhance the amount of O 3 produced over the course of the day (Jenkin et al., 1988;Alicke et al., 2002Alicke et al., , 2003Hu et al., 2002;10 Finlayson-Pitts et al., 2003). HONO and aldehydes are fairly reactive and as a result their concentrations are usually fairly small during most of the daylight hours. However, at night, in the absence of sunlight and OH, their concentrations can accumulate if sufficient sources of these compounds are present. Then, in the early morning when the sun first appears, they can be rapidly photolyzed; one product of this photolysis is 15 OH. Because other sources of OH (e.g., O 3 photolysis) tend to be small in the early morning, HONO and aldehyde photolysis can represent a significant early-morning source of OH, and this source can jump-start the photochemical reactions that lead to O 3 production. The net result is more O 3 production over the entire course of the day.
HONO and aldehydes were not measured during the field campaign and thus could 20 not be specified in the OBM. We therefore carried out two sets of OBM calculations: in the calculations labeled "OBM w/o HONO and ALD" we assumed that no HONO, H 2 CO , and higher aldehydes were present in the morning; in the standard OBM calculations we allowed the OBM to estimate the early morning HONO, H 2 CO, and aldehyde concentrations using the iterative process described in Sect.

5
The resulting diurnal variations in HONO, formaldehyde and higher aldehydes calculated by the OBM at TO on 7 November are shown in Fig. 10a. Note the high HONO concentrations (of ∼10 ppbv) just before sunrise. As the sun comes out, however, HONO starts to photolyze and its concentration rapidly drops to sub-ppbv levels. However, as illustrated in Fig. 10b, the photolysis of HONO in the early morning represents 10 a major source of OH in the OBM calculations, and, as a result, early morning OH concentrations are significantly enhanced (Fig. 10c). Inspection of the figures reveals that although the concentrations of formaldehyde and higher aldehydes are higher than HONO, their impact on OH production is relatively small. In addition, after ∼11:00 LT, HONO concentrations have fallen sufficiently and O 3 photolysis takes over as the dom-15 inant source of OH.
Similar results are achieved at CW, YL, and TC on 7 November (not shown), suggesting that HONO played an important role in the early morning chemistry provided that HONO was present at the concentrations derived from the OBM calculations. Earlymorning HONO concentrations of ∼10 ppbv in the Hong Kong area do not seem un-20 reasonable given that Hu et al. (2002) observed peak early-morning HONO levels approaching 12 ppbv in nearby Guangzhou in June 2000. Nevertheless, because HONO was not measured during the field campaign, future field experiments will be needed to determine whether the OBM-based predictions of high HONO concentrations in the early morning in Hong Kong area and the resultant enhancement in early morning OH 25 production rates are appropriate. BHC has a significant RIR at TM, the most rural site, but even here AHC has a much larger RIR. Not surprisingly (in light of the reactivity analysis presented in Sect. 3.1.5) R-AROM has the largest RIR among the AHC sub-groups. The RIR for NO x is negative at all sites except TO where it is essentially zero (Fig. 11a), and the area-averaged RIR for NO x is also negative (Fig. 12a). This nega-10 tive RIR for NO x is the result of the relatively high NO concentrations that were typically encountered at the sites during the field campaign. The results suggest that O 3 photochemistry is strongly VOC-limited throughout the Hong Kong area, and that initial reductions in NO x emissions in the area would actually lead to local O 3 enhancements.
Finally it should be noted that the standard error σ of the means in the area-averaged 15 RIRs tend to be relatively small when compared to the RIRs themselves (Fig. 12). This suggests that, at least from a statistical point of view, the RIR functions calculated here are robust. RIRs at TO: The calculated RIRs for the major O 3 precursor groups at TO on the episode days with VOC data generally confirm the results obtained from the other sites 20 for the 7 November episode. AHC is the dominant precursor group with largest RIRs, and, among the AHCs, R-AROM dominate (not shown). On all days except 9 October the RIR for NO x was negative, and thus NO x tended to suppress O 3 production on most of the episode days. The anomalous result for 9 October can be attributed to the fact that the NO concentrations were significantly lower than on other episode days; 25 the maximum NO concentration on 9 October was only 5.5 ppbv, compared to maxima of 7.6-95.4 ppbv on other days (Fig. 1).
Sensitivity of RIRs to model uncertainties and model inputs: There are a large num-Introduction EGU ber of input parameters and model formulations in the OBM that are uncertain and add uncertainty to the RIR-functions calculated by the OBM. However, we have carried out a wide variety of sensitivity studies where we allowed input parameters and formulations to vary over reasonable ranges. In all cases, the major results were essentially the same. For example, we compared the RIRs obtained with the standard version 5 of the OBM using diurnal profiles for VOC species (Sect. 3.2.2) with OBM results obtained with constant VOC concentrations as a function of time of day. While differences between the two calculations were obtained, the essential results are the same; i.e., the RIRs for AHC are large and positive and the RIRs for NO x are negative at the four EPD sites and small at TO. 10 We also examined the sensitivity of the OBM results to the methods used to calculate both the inputs into the OBM and the RIR functions. In the standard model, the hourly-averaged concentrations input into the model are arithmetic means of the highresolution data collected at the sites. The RIRs were calculated from the difference in the net O 3 production for the observed concentrations and for a -10% of change in a 15 precursor's concentrations. Here again, the essential results of the OBM calculations were unchanged.
Applicability of the autumn results: The findings that the O 3 photochemistry in the Hong Kong area is strongly VOC-limited and dominated by anthropogenic VOCs, especially reactive aromatics, are robustly supported by data that were gathered during 20 one autumn. A comparison with the results from a similar analysis of summer episodes observed at Tai O and at an inland PRD site (Wan Qing Sha) suggests that these conclusions generally hold for inland PRD sites during a majority of summer episodes, although natural HCs could play an important role during summer at Tai O (So and Wang, manuscript in preparation

Conclusions
Various observation-based approaches were used to analyze the relationships between ozone, ozone precursors, and cross-border transport during ozone episodes observed in the Hong Kong area. Local emissions from Hong Kong as well as emissions from Guangdong Province 5 appear to act in concert or individually in fostering O 3 pollution episodes in Hong Kong.
Between 50-100% of the O 3 increase observed in Hong Kong during O 3 -pollution episodes can be explained by photochemical generation within the Hong Kong area. In addition, HONO (and to a lesser extent aldehydes) was calculated to play a critical role in local O 3 generation, provided that HONO was present at the concentrations derived 10 from OBM calculations. The reactivity of the VOCs is dominated by anthropogenic VOCs, with VOCs from natural or biogenic sources making a minor contribution during fall, when this field study was conducted. Of the anthropogenic VOCs, reactive aromatics dominate, of which xylenes and toluene are the most important.

15
The formation of O 3 throughout much of the Hong Kong area is limited by VOCs, and high NO concentrations suppress O 3 production in much of the Hong Kong area. Ozone sensitivity to VOCs is dominated by the contribution from reactive aromatics.
Further studies are needed to confirm and quantify the role of HONO in early morning chemistry. While a preliminary comparison with the results obtained during summer 20 both in Hong Kong and at a PRD inland site suggests that the conclusions obtained in the present study generally hold, analysis of data collected from more areas in the PRD would be valuable to characterize additional areas of significant O 3 pollution and to determine the relative roles of anthropogenic and natural VOCs and the VOC-versus NO x -limitations of the O 3 photochemistry. (2) L = local episode L, r = Local episode with some impact from the mainland R = Regional episode with significant impact from the mainland.