Air quality diagnosis from comprehensive observations of total OH reactivity and reactive trace species in urban central Tokyo

We have conducted a comprehensive observational study to determine the mixing ratios of atmospheric chemical species and total OH reactivity in central Tokyo, in order to diagnose the air quality during summer and winter 2007 and autumn 2009. Concentrations of over 70 reactive trace species were continuously measured throughout each season. The total OH reactivity was measured directly using a laser-induced pump and probe technique. The observed chemical species exhibited seasonal variations. There was a good correlation between NO and CO in winter, but not in summer. This indicates that during winter, vehicle exhaust provided a signi ﬁ cant source of NO in central Tokyo, while photolysis of NO 2 was dominant in summer. Similar values (approximately 30 s (cid:1) 1 ) for the averaged total OH reactivity were observed in both summer and autumn during the daytime. However, VOCs accounted for a larger percentage of the OH reactivity in summer, while NO x was more dominant in autumn. We ﬁ nd that photochemical processes leading to oxidant production via VOCs dominated OH removal in summer, while the production of nitric acid from the reaction between OH and NO 2 was dominant in autumn, by-passing the oxidant production pathways and resulting in reduced oxidant formation. (cid:1) 2011 Elsevier Ltd. All rights reserved.


Introduction
In the past three decades, the increase in tropospheric ozone (O 3 ) concentration has remained a significant problem in most cities in Japan, even though the levels of ozone precursors such as nitrogen oxides (NO x ) and volatile organic compounds (VOCs) have decreased following the successful implementation of emission controls on vehicle exhaust (Bureau of Environment, Tokyo Metropolitan Government, 2010;Tajima et al., 2010).The concentration of NO x (NO x ¼ NO þ NO 2 ) and non-methane hydrocarbons (NMHCs) has decreased by 10%e20%, while in contrast, O 3 has increased by 2% per year between 1992 and 2002 in Japan (Akimoto, 2006).Tropospheric O 3 makes a significant contribution to global warming because of its large radiative forcing, which is nearly equal to that of methane (CH 4 ).Thus there is a need within the atmospheric science community to investigate the cause of this increase in tropospheric O 3 .Recent research has suggested that factors responsible for this increase in O 3 may include the presence of unknown VOCs (Sadanaga et al., 2005;Kajii et al., 2006) which facilitate highly efficient O 3 production following the decrease in NO x , the acceleration of O 3 production by the heat island phenomenon, and the transport of O 3 and its precursors from the Asian continent (Ohara and Sakata, 2003).However, the most significant causative factors of this increase in O 3 in the urban atmosphere have yet to be identified.In order to completely understand the mechanism of tropospheric O 3 production, simultaneous observations of all the precursors of O 3 and an understanding of their reaction pathways are necessary, but this is not a practical approach.A great number of reactions contribute to the formation of tropospheric O 3 .Lewis et al. (2000) reported that there are more than several hundred kinds of VOCs in the troposphere, and Goldstein and Galbally (2007) reported that there are probably more than 10 5 different VOCs in the troposphere, which underscores the difficulty in fully characterizing all tropospheric O 3 reaction pathways.
The measurement of total OH reactivity (k OH ), however, provides comprehensive information regarding OH loss processes, which include the reactions of almost all chemical species with OH in the atmosphere.These reactions constitute the first step in the process of photochemical O 3 production in the atmosphere.Total OH reactivity measurements in the urban atmosphere have been conducted in New York City, USA (Ren et al., 2003a(Ren et al., , 2003b(Ren et al., , 2006)), Mexico City, Mexico (Shirley et al., 2006), Mainz, Germany (Sinha et al., 2008), Pearl River Delta, China (Hofzumahaus et al., 2009;Lou et al., 2010), and Houston, USA (Mao et al., 2010).The values of the OH reactivity observed during these measurements vary widely, between 10 and 200 s À1 , which reflects considerable regional diversity in the urban atmosphere.Consequently further observations in many different locations together with an improved understanding of photochemical reaction processes in the atmosphere are required.We have previously conducted direct measurements of total OH reactivity using a laser-induced pump and probe technique at a semi-urban site in Tokyo, at Tokyo Metropolitan University (TMU) in 2003e2004 (Sadanaga et al., 2004b;Yoshino et al., 2006).Here the measured OH reactivity was compared with the OH reactivity calculated from the sum of all individual species measured that contribute to the OH reactivity.The calculated OH reactivity and the directly measured OH reactivity would be in agreement if all the chemical species that react with OH in the atmosphere were measured.The results from these studies, however, showed differences between the directly measured and the calculated values, demonstrating the presence of OH reactions with unmeasured chemical species.This phenomenon is known as the "missing OH reactivity".Unmeasured secondary species produced by photochemical oxidation in the atmosphere likely contributed to these processes since the missing OH reactivity was more significant in the summer and spring.We have also considered the potential O 3 production based upon the directly measured and the calculated OH reactivities.The O 3 production potential obtained from the directly measured OH reactivity was significantly higher than that obtained from calculated OH reactivity, particularly under lower NO x conditions (<20 ppbv) around TMU (Sadanaga et al., 2005;Kajii et al., 2006).These results indicate that unmeasured species, including secondary products caused by photochemical reactions in the atmosphere, made a significant contribution to the O 3 production.In addition, the missing OH reactivity was also observed in OH reactivity measurements of exhaust from gasoline-fueled vehicles, but the values were around 17e18% of the total OH reactivity, which is smaller than the value of approximately 30% that has been reported based on ambient air measurements (Nakashima et al., 2010).This indicates the presence of unknown species that come from primary emissions as well as secondary productions from photochemical reactions in the atmosphere.
In this study, we conducted intensive observations of reactive trace species in the atmosphere of urban central Tokyo, and have diagnosed the air quality using trace species measurements and direct measurements of the total OH reactivity.

Site description
All observations were conducted at the Tokyo Metropolitan Research Institute for Environmental Protection (TMRIEP,35.7 N,139.8 E), which is in Toyocho, Koto-ku, in urban central Tokyo.Fig. 1 shows the location of the site.Our observations were conducted over three seasonsdsummer (August 21e27, 2007), winter (December 20e21, 2007), andautumn (October 27e30, 2009)dwith several days of observation in each season.The sample inlets were situated in the windows of laboratories on the fifth and sixth floors of the TMRIEP main building and were set approximately 20 m above the ground (mean sea level).The observation site is surrounded by roads which experience heavy traffic, as well as many factories and buildings, and thus it is directly affected by primary emissions such as anthropogenic VOCs and NO x .Furthermore, Tokyo Bay is located approximately 2 km south of the site; hence, the site is also influenced by VOCs from the ocean and SO 2 from ship emissions.The composition of the atmosphere at the site is considered very complex and provides a medium for many chemical reactions.We therefore consider this location representative of urban central areas in Japan.

Details of measurements of chemical species in the atmosphere
Trace gas concentrations and the total OH reactivity were measured using the following techniques during each observation period.Meteorological data were also measured.
2.2.1.Measurement of NO x , CO, O 3 , and SO 2 NO x , CO, O 3 , and SO 2 were continuously measured using automated equipment 24 h a day during the observation periods.O 3 was measured by UV absorption (Model 1150, Dylec), CO by nondispersive infrared spectroscopy (Model 48C, Thermo Electron), and SO 2 by pulsed fluorescence (Model 43CTL, Thermo Electron).NO was measured by O 3 chemiluminescence (Model 42i-TL, Thermo Electron).NO 2 was measured by a laser-induced fluorescence (LIF) technique (Matsumoto et al., 2001;Matsumoto and Kajii, 2003).Sample air was introduced into each instrument through PTFE tubes with a 6.35 mm outside diameter.These measurements were conducted at 1-min averaging intervals.Zero air was introduced into all instruments every hour to measure the zero level of the instrument.The zero air was generated by a zero air generator (Model 111, Thermo Electron), in which chemicals were removed by purafil and charcoal filters and a heated Pt catalyst.

Measurement of VOCs
NMHCs were measured using two instruments.C 2 eC 5 NMHCs were measured in situ approximately every half hour during the summer observation period.Samples were analyzed using a gas chromatograph with a flame ionization detector (GCeFID, HP 5890, Hewlett Packard).NMHCs were separated on PLOT (Al 2 O 3 ) capillary columns in a system previously described (Goldan et al., 2000).Concentrations were calibrated with a reference standard (#1660a, 1 ppm propane, National Institute of Standards and Technology, USA).Sample volumes were 250 standard milliliters.Approximately 60 species of C 2 eC 10 NMHCs were also measured using a GCeFID (HP 6890, Hewlett Packard) during all periods.Details of this system were previously described by Kato et al. (2001).The ambient air samples for NMHCs analysis were introduced into stainless steel canisters (3 or 6 L) at 1-h intervals in the daytime.0.5 L of the sample air was introduced into a three-stage preconcentrator (Model 7000, Entech) to concentrate volatile components prior to injection into the GCeFID and separation using an HP-1 Column (60 m length, 0.32 mm i.d., 1-mm film thickness, Agilent J&W).The temperature of the GC oven was initially kept at À50 C (for 8 min) and then increased to 40 C at 5 C min À1 , and subsequently further raised to 150 C at 15 C min À1 .NMHC concentrations were calibrated with standard gas (Matheson, Enviro-MAT).Detection limits of the NMHCs measured from the canisters by GCeFID are in the range of 1e3 pptv with accuracy of 2e13% and precision of 2e15%.Benzene, toluene, isoprene, and terpenes were measured by means of proton transfer reaction mass spectrometry (PTR/MS) at 1-min averaging intervals to increase the time resolution for the measurement of NMHCs (Kato et al., 2004).Formaldehyde was measured by the Hantzsch reaction using a fluorometric detector (AL4021, AERO-LASER) at 1-min averaging intervals.Oxygenated volatile organic compounds (OVOCs) other than formaldehyde, such as acetaldehyde, acetone, and methanol, were also measured by PTR/MS at 1-min averaging intervals.PTFE tubes with a 6.35 mm o.d. were used for the sample inlets.The zero level of the PTR/MS was also measured every hour using zero air.

Measurement of OH reactivity
OH reactivity was measured every few minutes in the daytime using a laser-induced pump and probe technique (Sadanaga et al., 2004a).Sample air was introduced through a 12.7 mm o.d.PTFE tube into a reaction tube by a diaphragm pump (DA-60S, ULVAC).The flow rate of the sample air was set to 20 SLM using a mass flow sensor (Model 3810S, Kofloc) and needle valve, and the pressure in the reaction tube was maintained at atmospheric pressure.A pulsed 266-nm laser beam from the fourth harmonic of a Nd:YAG laser (Quanta-Ray INDI-40, Spectra physics, or Tempest 300, New Wave Research) was used to irradiate the reaction tube, where it generated OH radicals artificially through the photolysis of O 3 and the reaction with water vapor in the sample air, as follows: The repetition rate and the laser energy of the laser beam were set to 2 Hz and around 10 mJ, respectively.The generated OH radicals reacted with the trace species in the sample air, resulting in a decrease in the radical concentration.The changes in the OH concentrations were detected in a fluorescence detection cell by a laser-induced fluorescence technique.A portion of the sample air in the reaction tube was introduced into an LIF cell using a rotary pump (D-950K, ULVAC), and the pressure was maintained at around 2 Torr.An A 2 S þ À X 2 P(0,0) Q 1 (2) rotation line of OH in the sample air was excited by a second harmonic pulsed dye laser (Scanmate, Lambda Physik, or Precision Scan, Sirah) pumped by a second harmonic pulsed Nd:YVO 4 laser (YHP-40-532Q, Spectra Physics) into the cell to detect the fluorescence of OH.The repetition rate of the laser was set to 10 kHz, and the power was approximately 2e3 mW.The fluorescence signal was detected by a photomultiplier tube (R2256P, Hamamatsu) equipped with a dynode gating system, and the signal was recorded using a photon counting method.
In the autumn period, the sample air was diluted with zero air to remove the influence of OH recycling via the reaction of NO with HO 2 which takes place under high NO conditions.Before and after the OH reactivity measurement, zero air was introduced into the reaction tube and its decay rate was measured each day.The zero air decay rate was measured 8 times each day, and the averaged value of these measurements was applied.

Results and discussion
3.1.Diurnal profiles of NO x , CO, O 3 , and SO 2 Fig. 2 shows the diurnal variations in the NO x (NO x ¼ NO þ NO 2 ), CO, O 3 , and SO 2 concentrations during all study periods.The averaged daytime concentrations of each species from all observation periods are summarized in Table 1.
O 3 usually exhibits a high concentration during the daytime and a low concentration in the early morning and evening.In this study, O 3 had a daytime peak at around 14:00 and decreased to near zero at night, since the O 3 reaction converting NO to NO 2 results in regeneration of O 3 following NO 2 photolysis during the day, while this pathway does not occur at night.This is a typical diurnal variation regularly observed in the urban atmosphere.The highest mixing ratios of O 3 were above 120 ppbv, on August 25e26 in the summer observation period, during which time there was strong solar radiation.In the winter and autumn, there was a reduced amount of sunlight and the maximum concentrations of O 3 were in the range of 20e40 ppbv.
NO and NO 2 also showed typical diurnal variations for an urban area.The NO concentration was almost zero during the nighttime and increased rapidly in early morning.It then decreased around noon, and increased again in the evening rush hour.This diurnal variation is representative of conditions where vehicle exhaust is the main emission source.NO 2 never decreased to a zero value during the nighttime.In the summer, NO 2 essentially showed a similar variation to NO, but at a consistently higher concentration than NO.However, in the winter, NO 2 showed different behavior to NO, with a relatively constant concentration of around 50 ppbv Fig. 3 shows scatter plots for CO and NO in all seasons.There were good correlations between NO and CO in winter and autumn, but not in the summer.In summer and autumn, the plot shows shallower slopes between CO and NO in the daytime compared to nighttime, with no significant difference found in winter.The good correlations between NO and CO in winter and during nighttime in the autumn reflect air dominated by fresh emissions from vehicle exhaust.In contrast, in summer and the autumn daytime, photolysis of NO 2 provided a significant source of NO in addition to automobile emission sources.The DCO/DNO emission ratio in all seasons was approximately 3e5, as can be seen in Fig. 3.These values were much lower than the emission ratios of 10e150 obtained from the direct measurement of gasoline-fueled vehicles that were compiled with the 2005 Japanese emission control standard (Nakashima et al., 2010).These high emissions of NO were therefore presumably produced by diesel-fueled vehicles and/or gasoline-fueled vehicles that did not comply with the new regulations.
The CO mixing ratio ranged from 200 to 500 ppbv in summer; in winter and autumn, the range of CO values was approximately 500e600 ppbv and often exceeded 1 ppmv.Our measurements suggest that the combustion of fossil fuel contributed significantly as an emission source of CO, since similar trends between CO and NO 2 were observed.SO 2 was observed at mixing ratios of several ppbv, sometimes exceeding 10 ppbv in summer and autumn.A high level of SO 2 was observed when the wind was from the south, as shown in Fig. 4. Ship emissions from Tokyo Bay would also contribute to the high level of SO 2 .In winter, no large fluctuation in SO 2 concentration was observed, and the range was 2e5 ppbv.The source of SO 2 in winter remains unexplained, as there was no correlation between SO 2 and CO.

Diurnal profiles of VOCs
Fig. 5 shows the diurnal variations of NMHCs in all observation periods.Note that canister samples for NMHCs were not collected  1.The averaged concentrations of anthropogenic NMHCs such as alkanes, alkenes, and alkynes decreased in order of increasing carbon number in all seasons, because reactivity with OH and/or O 3 and boiling point increase with increasing molecular weight.Our findings are similar to those from previous observations in Hachioji, a suburban area of Tokyo (Tajima et al., 2010).Alkanes accounted for approximately 60% of the total NMHCs in all observations, alkenes 20%, aromatics 10%, and biogenics 2%.Most NMHCs (e.g., ethane, propane, ethylene, and toluene) showed higher mixing ratios in winter than in summer, with autumn mixing ratios in between the winter and summer seasonal averages.Ethane, propane, and ethylene all have a comparatively low reactivity with OH, and their loss processes by OH were limited in winter because of the diminished photoproduction of OH.In summer, isopentane, n-pentane, and isoprene concentrations were higher than in winter.Isoprene is emitted from vegetation, especially broadleaved plants, and thus its emissions increase in summer as a result of higher biological activity.Isoprene is also emitted from automobiles (Sharma et al., 2000;Nakashima et al., 2010) but the much higher concentrations observed in summer at this site demonstrated that biogenic sources of isoprene were dominant.Isopentane and n-pentane are emitted from the volatilization of gasoline with increasing atmospheric temperature.Species with short atmospheric lifetimes did not show obvious seasonal variations.Aromatics, including benzene and toluene, showed higher values in winter than in autumn and summer, though the reason for this remains unexplained.The source of these species could be vehicle exhaust, since they showed significant diurnal variation that increased in the morning and evening, and decreased at noon.Fig. 6 shows the diurnal variation in OVOCs such as methanol, formaldehyde, acetaldehyde, and acetone during all periods.In contrast to NMHCs, most OVOCs showed larger values in summer than in winter and autumn.The formaldehyde, acetaldehyde, and acetone levels in winter and autumn increased in the morning, decreased at noon, and then increased again in the evening.The morning and the evening increments in concentration were due to vehicle exhaust emissions.In summer, these species showed variations similar to those in winter and autumn during the morning, but maintained a relatively large value from noon to evening.This indicates the presence of additional sources of formaldehyde, acetaldehyde, and acetone, and photochemical production processes in the atmosphere which contributed to the increased concentrations from noon to evening.Regarding methanol, the averaged summer concentration was several times larger than the winter value.It should be noted that the elevated concentrations on the evenings of August 22 and 25 were partially due to a strong local source within the laboratory (as a result of changing the dye solution in the laser).

Diurnal profiles of total OH reactivity
Fig. 7 shows a time series of the observed and calculated OH reactivity for all periods.OH reactivity was not measured on August 24 and 25, 2007.OH reactivity was predicted based upon the observed concentrations of trace gas species and their rate coefficients of reaction with OH (Atkinson, 1994, 1997, 2003, Atkinson et al., 1997;Calvert et al., 2000;Sander et al., 2002).The rate coefficient k NO2 of the OH þ NO 2 reaction was obtained from direct measurement in the reaction tube because k NO2 is humidity and pressure dependent (Sadanaga et al., 2004b(Sadanaga et al., , 2006)).The calculated  OH reactivity, k OH, is defined as the sum of the products of the observed concentrations and rate coefficients for each species, according to the following formula: The calculated and observed values for OH reactivity were compared.The observed OH reactivity showed a strong fluctuation because trace species change rapidly in an urban atmosphere; this was quite a different result from the observations made in Hachioji, which is 30 km west of Shinjuku, central Tokyo (Yoshino et al., 2006).The observed and calculated values showed a similar variation, which supports this method of measurement of total OH reactivity.In addition, the differences between the observed and the calculated OH reactivitydthe missing OH reactivitydwere similar to the observations in Hachioji, which indicates the presence of unknown reaction processes of OH radicals.The missing OH reactivity did not show diurnal variations.
In summer, the OH reactivity fluctuated substantially, between 15 and 70 s À1 .On August 26, however, the OH reactivity was quite low, in the range 15e30 s À1 , because that was a weekend and the air was relatively clean.Differences between the observed and calculated OH reactivity were observed on all days during the summer measurement period.We have previously performed a three-dimensional air quality study which consists of the Weather Research and Forecasting (WRF) model for meteorology, the Community Multi-scale Air Quality (CMAQ) modeling system for pollutant concentrations, and emission inventory models.The model study of OH reactivity in the summer period suggested that unknown secondary products and unidentified primary emitted species contribute to the missing OH reactivity (Chatani et al., 2009).In winter, the observed OH reactivity was in the range of 10e80 s À1 and contained large uncertainties.The large uncertainties in OH reactivity were particularly prominent in the morning, and there were different variations in the observed and the calculated values.This is believed to be due to the OH recycling from the HO 2 þ NO reaction, which causes an underestimation of OH reactivity when the NO level is high during the morning rush hour (Sadanaga et al., 2004a;Yoshino et al., 2006).We have excluded these data with large uncertainties from Fig. 7 and later discussions.In autumn, the sample air with high NO x was diluted to inhibit OH recycling, although the air was not diluted on October 28, thus no data is available for that day.The observed OH reactivity in autumn was in the range 10e80 s À1 , relatively similar to winter values.On October 27, which was a very clear and comparatively clean day, the OH reactivity ranged from 10 to 20 s À1 .On October 29 and 30, OH reactivity ranged from 30 to 80 s À1 , with the values increasing in the morning, decreasing around noon, and increasing again in the evening.The high values in the morning and evening reflect the primary emissions from vehicle exhaust, NO and NO 2 in particular.The inconsistency between the observed and the calculated OH reactivity was small on October 27, a bright, clear day, while it was relatively large on October 29 and 30, which were highly polluted days.During the observations in suburban TMU, the secondary products of atmospheric photochemical reactions contributed to the unknown OH loss processes.However, despite a clear day on October 27, the missing OH reactivity was smaller compared the other days in the autumn observations.The comparatively larger missing OH reactivity observed during autumn compared to summer, especially during the polluted days, suggests that the missing sink in autumn is most likely owing to unmeasured primary emitted species.

Contribution of each species to total OH reactivity
Fig. 8 shows the percentage contribution of each trace species to the total OH reactivity in all summer and autumn periods.In the figure, the entire 360 circle indicates the observed OH reactivity, and the ratios of each species indicate the calculated OH reactivity of each species.In summer, NO and NO 2 accounted for a quarter of all OH reactivity, at 26.0%.Anthropogenic NMHCs accounted for 22.1%, and OVOCs for 12.4%.Biogenic NMHCs accounted for 5.82% despite their concentration ratio of only 2% of the total VOCs, due to the high reactivity of terpenoid compounds with OH.The total ratios of VOCs accounted for the largest percentage, which exceeded 40% of the total OH reactivity.In autumn, the percentage of NO 2 was the highest, at 31.1%, which combined with NO at 6.19% to account for 37.3% of the total OH reactivity.Anthropogenic NMHCs accounted for 18.0%, OVOCs for 4.88%, and biogenic NMHCs for 0.92%.The total contribution of VOCs was 23.8%.The averaged values of the observed OH reactivity were similar, at 33.4 s À1 for the summer period, and 32.3 s À1 for the autumn period.VOCs, however, accounted for the highest percentage in summer; in contrast, NO x made the highest contribution in autumn.These results indicate that the processes of photochemical production of oxidants from VOC and OH reactions are dominant in summer, while the process of nitric acid production from the NO 2 reaction with OH is dominant in autumn.

Contribution of the missing OH reactivity to photochemical ozone production
If the missing OH reactivity originated from unknown VOCs, the photochemical production of oxidants would be expected to increase with increasing OH reactivity.The ozone production  potential may be used to investigate the impact of unknown species on production of O 3 and can be calculated using a simple box model.The ozone production potential is defined as the total of HO 2 and RO 2 radicals via reaction of VOC and CO with OH while the concentration of OH is controlled by the reaction between OH and NO 2 .Detailed descriptions of the model can be found in Sadanaga et al. (2005).The chain reactions involved are as follows. OH In the urban atmosphere, we can assume that the produced HO 2 and RO 2 radicals react with NO to produce NO 2 because of high level NO x , and that O 3 is subsequently produced by photolysis of NO 2 .Therefore, the production rates of HO 2 and RO 2 ðP RO2HO2 Þ play a substantial role in determining the production rate of O 3 .The ozone production potential is proportional to the number of accumulated HO 2 and RO 2 radicals through the radical propagation cycle from a single OH radical.The production rate of RO 2 and HO 2 radicals P RO2HO2 and ozone production potential F are defined as follows.
Here, [OH] 0 is the initial OH concentration, which is OH generated by primary sources such as O 3 photolysis.[OH] 0 was set to the appropriate concentration (typically 10 6 molecules cm À3 ).Sk 1i [VOC i ][OH] is the summation of reactivity of OH þ VOC i .The VOC i reactivity was obtained from both observed (combination of missing OH reactivity and known VOCs reactivity) and calculated OH reactivity in this study.Fig. 9 shows the ozone production potential in the summer period.The ozone production potential obtained from directly measured OH reactivity was significantly higher than that obtained from calculated OH reactivity, particularly under lower NO x conditions (<30 ppbv).This is also similar to the results obtained during autumn, although the ozone production potential was lower than that in summer.When the NO x concentration is higher than 30 ppbv, the NO 2 þ OH reaction is dominant and inhibited O 3 production.On the other hand, under low NO x conditions, the NO 2 þ OH reaction is inhibited while ozone production efficiency is nonlinearly enhanced because VOCs þ OH reactions are facilitated.This suggests that the missing OH reactivity owing to unknown VOCs contributed significantly to photochemical ozone production in the urban atmosphere.

Conclusion
We conducted comprehensive observations in Toyocho, Kotoku, Tokyo, in the summer and winter of 2007 and in the autumn of 2009.The observation site was situated in central Tokyo, where there were primary anthropogenic emissions such as vehicle exhaust and emissions from factories.The observed trace species had large variations because the site was close to these emission sources and directly reflected the primary emissions.A large variation in OH reactivity was also observed.
O 3 variations were typical of an urban atmosphere, in that concentrations were high in the daytime and low in the morning and evening.O 3 concentrations higher than 120 ppbv were observed in summer, while in winter and in autumn, O 3 remained at 20e40 ppbv.NO and NO 2 also showed variations that were typical of an urban atmosphere, in that they increased in the morning and in the evening rush hour, and decreased around noon.A comparison of NO and CO indicates that in winter and in autumn, they exhibited a good correlation, while in summer, they did not.These results show that primary vehicle emissions were the dominant NO source in winter and in autumn, while photolysis of NO 2 was significant in summer, along with vehicle emissions.
The total NMHC concentration was higher in winter than in summer and in autumn.This reflects a reduction in the loss processes of NMHCs because of the low OH levels in winter.OVOCs were present in higher concentrations in summer than in winter and autumn.This reflects the importance of the photochemical production of OVOCs.
The level of OH reactivity ranged from 10 to 80 s À1 in all periods, with values that were higher than those obtained in Hachioji, a suburban area.OH reactivity was high in the morning and the evening rush hour, which reflects primary emissions from vehicle exhaust.The missing OH reactivity was observed on all days.The missing OH reactivity may be at least partly due to unknown VOC species produced from photochemical reactions in the atmosphere and/or primary emissions sources in the urban environment.The missing OH reactivity significantly contributed to the ozone production potential, particularly under low NO x conditions during the summer period.With the breakdown of OH reactivity, NO x accounted for a large contribution to OH reactivity in autumn, while VOCs accounted for a large value in summer, despite showing similar values for total OH reactivity.These results indicate the presence of different dominant processes in atmospheric reactions that are dependent upon the season.financially supported by Ministry of Educations and Science, Sports, and Culture of Japan (Grant-in-Aid No. 21221001).A part of this research has been performed commissioned by Ministry of Economy, Trade, and Industry of Japan, and as part of a collaborative research program between auto and oil industries called Japan Auto-Oil Program (JATOP) run by Japan Petroleum Energy Center.

Fig. 1 .
Fig. 1.Location of the observation site at Tokyo Metropolitan Research Institute for Environmental Protection in Toyocho, Koto-ku, Tokyo, in Japan.

Fig. 2 .
Fig. 2. Diurnal profiles of concentrations of CO, O 3 , SO 2 , NO and NO 2 for all observations.

Fig. 3 .
Fig. 3. Scatter plots of CO and NO for all observations.Upper tiles represent daytime and lower tiles represent nighttime for each season.

Fig. 4 .
Fig. 4. Averaged concentration of SO 2 for summer and autumn obtained for different wind directions.

Fig. 5 .
Fig. 5.Diurnal profiles of concentrations of NMHCs from GCeFID analysis for all observations.The NMHC measurements were conducted only in the daytime.

Fig. 7 .
Fig. 7. Time series of observed (open circles) and calculated (lines) total OH reactivity, and their differences (dots) for all observations.

Fig. 8 .
Fig.8.Contribution of different trace species to total OH reactivity for summer and autumn observations in the daytime (9:00e18:00).Averaged total OH reactivity for each observation period was 33.4 s À1 in summer and 32.3 s À1 in autumn.

Fig. 9 .
Fig. 9. Ozone production potentials (F) with OH reactivity versus concentration of NO x in the daytime during the summer period.F corresponds to the number of accumulated HO 2 and RO 2 radicals through the radical propagation cycle from a single OH radical.The ozone production potentials obtained for observed and calculated OH reactivity are shown as annotated.

Table 1
Averaged concentrations and standard deviations of observed trace species in daytime from 9:00 to 18:00 during the all periods.Units are all in ppbv.
on August 24 and 25, 2007.The averaged daytime concentrations of each species for different seasons are summarized in Table