Atmospheric CO , O 3 , and SO 2 Measurements at the Summit of Mt . Fuji during the Summer of 2013

Atmospheric trace gases CO, O3, and SO2 were observed at the summit of Mt. Fuji (3776 m a.s.l.) during the summer of 2013. Considerable variations were observed in the concentrations of CO and O3; however, they were correlated in most cases. Trends analyzed through backward trajectory calculations showed lower concentrations of CO and O3 transported from the Pacific Ocean and South East Asia directions, while higher concentrations were detected from the direction of the Asian continent. High O3 and low CO concentrations were observed during some periods; in these air masses, water content of the air was low indicating that the air originated from high altitudes and was influenced by the stratosphere. Gaseous SO2 was mostly lower than the detection limit of the instrument used for measurement (0.06 ppbv), but on August 20–21, high SO2 spikes of about 5 ppbv were observed. Backward and forward trajectory calculations confirmed that volcanic smoke from the Sakurajima volcano was transported to the summit of Mt. Fuji.


INTRODUCTION
It is important to be aware of air quality on a regional scale.Polluted air originating from high emission areas can be transported to remote locations and can influence the air quality there.Also on site observation data at remote places are necessary for the validation of global or regional calculations and remote sensing observations.O 3 is a significant short-lived trace gas.O 3 becomes high concentration in polluted urban areas through photochemical reactions, causing health risks for humans and vegetation.In addition, O 3 can be transported to remote rural areas.The concentrations of O 3 , a short-lived tropospheric greenhouse gas, has been increasing in East Asia at a regional scale according to some reports (Akimoto, 2003;Tanimoto, 2009;Tanimoto et al., 2009); high levels of O 3 can affect plant growth and decrease crop production.A regional and hemispheric increase of tropospheric O 3 causes a positive shift in radiative forcing.In recent years, regional transport has led to high concentrations of O 3 in Japan.For example, quite high photochemical oxidant (more than 120 ppbv) was observed in the western coastal areas where there is minimal human activity (Ohara et al., 2008).On the other hand, there is a report that O 3 concentration have been decreased from 2008 at mountain site (Happo) in Japan (Okamoto et al., 2015).More monitoring of O 3 at remotes will be required.
Monitoring CO at remote sites is also important to understand because this gas is a strong indicator of pollution.O 3 can be produced by photochemical reactions in the presence of NO x and volatile organic compounds (VOCs), which work as peroxy radical sources, especially in polluted areas.NO x is oxidized to HNO 3 and is removed from the atmosphere within a few days.In remote places, NO x concentrations become quite low after transport.On the other hand, CO is emitted along with NO x and VOCs during the process of combustion.The lifetime of CO is longer than one month so its concentration generally remains high even after being transported to remote locations.Therefore, monitoring of O 3 and CO concentrations provide useful information in understanding the long-range transport of the polluted air to remote areas.
SO 2 is another atmospheric pollutant emitted from the combustion of fossil fuel.The main sources of SO 2 is the combustion of coal, a process highly common in China (Kurokawa et al., 2013).Therefore measurement of SO 2 at remote sites can be useful to detect long range transport of polluted air (Igarashi et al., 2004(Igarashi et al., , 2006)).In addition to anthropogenic emission, SO 2 is also emitted through volcanic activity.There are several volcanoes in Japan and some of them become active recently.SO 2 measurement would be important also to monitor volcanic activities.
Atmospheric pollutants are emitted near ground mostly and transported in the boundary layer along the ground and ocean.In addition, polluted air can also be lifted by convection and transported in long distance through the free troposphere.Most ground-based observatories are located at low altitudes, limiting the availability of onsite measurements in the free troposphere.Therefore, aircraft measurements are beneficial when measuring the transport of polluted air.Some intensive studies have been performed to observe the continental outflow of atmospheric pollutants to the East China Sea and the Pacific Ocean (Hoell et al., 1997;Huebert et al., 2003;Jacob et al., 2003;Hatakeyama et al., 2014).However, aircraft measurements are quite expensive and are only suitable for short-term observations.On the contrary, ground-based measurements at mountain sites are valuable for long-term observation at relatively low costs.
Mt. Fuji is a volcanic mountain in Japan, well known for its beauty, and was registered as a UNESCO (United Nations Educational, Scientific and Cultural Organization) World Heritage site in 2013.The Mt. Fuji weather station is located at the summit of Mt.Fuji and has been used as an important meteorological observatory.Because of its high elevation (3776 m a.s.l.) and absence of local pollution sources, the Mt.Fuji weather station is ideal for regional air quality measurements in the free troposphere.
Since East Asia is one of the most populated area, the regional atmospheric pollution would be reflected by economical and technological situations.Therefore update observed information of atmospheric pollutants at this region would be meaningful.This paper reports the observations of CO, O 3 , and SO 2 concentrations carried out at Mt. Fuji during the summer of 2013.

Observation Site and Period
The Mt. Fuji weather station is located at the summit of Mt.Fuji (3776 m a.s.l., 35.4°N, 138.7°E), as shown in Fig. 1, and it is operated by the Japan Meteorological Agency.Especially, the mount Fuji radar system was famous and it worked as an important weather forecast and typhoon warning system in previous times.However, most modern meteorological observations shifted to be made through satellite remote sensing.In 2004, the Mt.Fuji weather station reduced its operations leaving only automated tools monitoring fundamental meteorological measurements.Since 2007, the non-profit organization "Valid Utilization of Mt.Fuji Weather Station the station" attempted to support activities at the Mt.Fuji weather station and it is available for scientific and educational purposes only during summer months.Some atmospheric measurements (CO, O 3 , SO 2 , CO 2 , Hg, aerosol, fog/rain, radon etc.) have been conducted with their support.
For our study, instruments to monitor CO, O 3 , and SO 2 were installed from July 18 to August 22, 2013.With the exception of electric power outage on July 23 and 24 caused by a thunderstorm approach, data were successfully collected during the entire measurement period.

Instruments
CO (Thermo Electron, Model 48C) and O 3 analyzers (Thermo Electron, Model 49i) were set up on the second floor of the main building at the Mt.Fuji weather station.Ambient air was collected through a PFA tube (O.D. 6.35 mm, I.D. 4 mm, about 10 m length) and transferred into these instruments.A Teflon filter and a PFA water trap were installed in the inlet tube because fog condensation (cloud water droplets) on the inlet tube could potentially be a serious problem at this observation site.
The CO analyzer employs non-dispersive infrared absorption for the assessment.The background signal of the CO monitor is influenced by temperature (CO signal decreases as temperature increases) and water vapor, and also increases gradually as the correlation wheel accumulates dirt.Therefore, frequent background signal checks are necessary for ambient air measurements at clean remote sites using zero air (CO removed from air), which is generated by a heated Pt catalyst (Thermo Electron Model 96).CO is oxidized to CO 2 , but because the Pt converter does not remove water, the amount of water in the ambient air and zero air should be the same.Every hour, the zero air was measured from 0 to 15 min while ambient air was measured from 16 to 59 min using an automatic switch of the solenoid valve in the CO analyzer.CO concentrations in the ambient air can be obtained by subtracting data during zero air measurements from the data during ambient air measurements.Noting that the air temperature of the observatory was not controlled to be constant.During the measurement period the CO instrument reading of zero air was about 100 ppb initially and increased to about 1500 ppb, with roughly 200 ppb diurnal variation.The CO analyzer was calibrated by standard CO gas (Taiyo Nissan, Japan, 1.96 ppm) in the laboratory before and after field measurements without onsite calibration.Since the measurement principal is based on infrared absorption, decrease of the density of the molecules in high altitude (low pressure) influences the observed signal.However, we can assume ideal gas in ambient CO measurement and the instrument has a function to correct the pressure and temperature differences in the absorption cell.These corrections convert the observed mixing ratio to standard pressure and temperature conditions.Using the standard deviation during zero gas measurement on site, the detection limit of the CO analyzer in this measurement campaign was estimated to be 22.6 ppb.
The working principle of the O 3 analyzer is based on UV absorption.The O 3 analyzer was calibrated by an O 3 primary standard (Model 49i-PS) before field measurements (about 2 month before).Similar to the CO instrument, the observed O 3 mixing ratio was corrected to standard pressure and temperature conditions.The detection limit and uncertainty of the O 3 analyzer were 1 ppb and 1%, respectively, according to the company information.
The SO 2 analyzer (Thermo Electron, Model 43C-TL) was set up at another building of the Mt.Fuji weather station.Ambient air was drawn in from an air inlet chimney (I.D. 100 mm) at a high flow rate (143 L min -1 ) and separated in a PFA tube (O.D. 6.35 mm, I.D. 4 mm) leading to the SO 2 analyzer.Similar to the CO measurement process, zero air produced by catalyst (SP Blend Media, Purafil) and SO 2 in ambient air were measured periodically.The analyzer was calibrated before and after the field measurements using standard SO 2 gas (Takachiho, Japan, 0.98 ppm), diluted by mass flow controllers.The detection limit of the SO 2 monitor was documented as 0.5 ppbv (for 300 sec) in the manual from the company, but from the actual measurements in our system, the detection limit was estimated to be an hourly average of 0.06 ppbv.Igarashi et al. (2004) used the same commercial SO 2 analyzer (Model 43C-TL), but their inlet system was different from ours.They carefully tested the performance of the SO 2 measurement system and concluded that the detection capability was 0.05 ppbv.
The output analog signal of these instruments was converted to digital data by U3-HV (Labjack) and was logged on a personal computer at intervals of one minute.The data logging computers were attached to the network at the Mt.Fuji weather station connected to the Internet with the wireless LAN provided by the National Institute of Radiological Sciences.The real-time data could then be verified by a remote control system from any location, which was beneficial considering the observatory was not easily accessible.The electric power supply to the Mt.Fuji weather station needed to be shut down during an expected thunderstorm to avoid fatal damages, and we ensured the instruments successfully restarted after electric power came back.

Backward Trajectory Calculation
Backward trajectories were calculated using the NOAA HYSPLIT 4 model (Draxler and Rolph, 2012;Rolph, 2012).The air mass was traced back 5 days (120 h) and calculated at intervals of 4 h.A meteorological data set was used for GDAS along with model vertical velocity data for vertical motion.The initial elevation reading by the backward trajectory calculation was 3776 m.These readings are categorized based on different origins, as: North (N), North West (NW), China Korea (CK), China (C), South East Asia (SE), and Pacific Ocean (P) as shown in Fig. 1.Trajectories that were difficult to categorize are named unclassified (U).Because of the problems with the power supply, no CO, O 3 and SO 2 data were available during thunderstorm approach; a total of 194 backward trajectory calculations were thus obtained.

CO and O 3 Results
Hourly mean CO and O 3 concentrations are shown in Fig. 2.Even in a remote location at a high altitude, there were considerable variations in the concentrations of both CO and O 3 .Average CO and O 3 concentrations and standard deviations were 105.9 ± 44.3 ppbv and 29.3 ± 13.2 ppbv for CO and O 3 , respectively.The variation of CO and O 3 concentrations during the summer of 2013 is consistent with previous recordings by Tsutsumi and Matsueda (2000).Tsutumi et al. (1994) reported the standard deviation of O 3 concentration to be low during the winter but high during the summer.
The top of Mt.Fuji reaches the free troposphere where local sources of emissions are not expected.Therefore, generally, it is safe to assume that the large variations in CO and O 3 concentrations originate from transport from a distant air mass.However, Tsutsumi and Matsueda (2000) described that the air at the top of the Mt.Fuji was influenced in some cases by emissions from Nagoya, a large city in Japan.Fig. 2 shows the origin of air masses plotted through backward trajectories.Usually during summer months air from the Pacific Ocean is dominant.However, during the observation period in the summer of 2013 only about 10% of transported air originated from the Pacific Ocean.It can be observed that CO and O 3 concentrations are low when the air masses originate from P or SE.Average concentrations and standard deviations of pollutants in the air masses of different origins are summarized in Table 1.Air masses originating from CK, NW, and N show the highest average concentrations of CO and O 3 , while air masses from SE, and P show lower concentrations.The category of C was observed only 1.5% frequency and was considered statistically unreliable.A close analysis of the results showed that CK and NW had almost the same average O 3 concentrations, but CO was higher for NW.SE and P had similar average O 3 concentrations, while SE showed a higher CO concentration.However their differences were within standard deviation, it is interesting to note that SE and P show differences in CO, which is emitted primary from combustion sources, yet there is almost the same concentration of O 3 which is   In Fig. 3, a scatter plot of CO and O 3 is shown by different symbols for air mass origins categorized by backward trajectory.Relatively polluted air masses (N, NW, CK, and C) are shown in left panel and relatively cleaner air masses (P and SE) are shown in right panel.It can be seen that low CO and O 3 concentrations mostly originated from P and SE, while high CO and O 3 concentrations mostly originated from CK, NW, and N.In the P and SE plots, there is a rough linear correlation between CO and O 3 (shown in Fig. 3) To support the comparison of different figures, "guide line" was shown in Fig. 3.The guide line passes on typical oceanic clean air (50 ppb CO and 10 ppb O 3 ) and continental air (assumed to 150 ppb CO and 55 ppb O 3 ).In relatively clean air masses, influence of polluted continental air to the clean air would make this linear correlation.In the N, NW, CK, and C plots, there is greater variation and it is difficult to see a linear correlation as observed in P and SE (guide line).Higher CO with lower O 3 concentrations were observed in NW.These deviated data would be related to less photochemical production of O 3 .
Some data from of CK, NW, and N were clearly deviated from a linear correlation between CO and O 3 (low CO but high O 3 ).Because the observatory is at high elevation, the air from high altitudes (stratosphere) is more easily transported.Tsutsumi et al. (1998) reported a stratospheric O 3 intrusion event at the summit of Mt.Fuji.Since CO is mainly emitted from the sources on ground (fossil fuel combustion, biomass burning etc.), CO concentrations are generally higher near ground and lower at high altitudes.On the contrary, O 3 is higher near the stratosphere where the ozone layer exists.Most of the low CO with high O 3 data observed here were originated from high altitudes in the backward trajectory calculations.However, the air masses originating from high altitude were not always clean and some contain generally polluted air (high CO and high O 3 ).Since air temperature decreases with altitude and water is removed as cloud droplets, water vapor decreases with altitude; therefore, water vapor content is important when confirming the influence of high altitude air.In Fig. 4, a time series of CO, O 3 , and water vapor content (observed by the Japan Meteorological Agency) are shown together.There are periods when the water content is quite low, and O 3 becomes higher while CO concentrations decrease.In Fig. 5, the correlation of CO and O 3 is plotted with different symbols for water content (low water vapor plotted in warm colors and high water vapor plotted in cold colors).It is clear that most of the low CO and high O 3 data contain low water and they would be transported from high altitudes.These data verify that influence of stratospheric air can be observed at high- (5-6 g/kg) (6-7 g/kg) (7-8 g/kg) (8-9 g/kg) (10-g/kg) guide line elevation observatories such as Mt.Fuji, even during the summer.These high O 3 , low CO, low water vapor data are named as "event" here.From Fig. 5, the "event" was defined as O 3 > 50 ppb, CO < 90 ppb, water < 4 g kg -1 .The "event" was found 67 hours during total measurement period (864 hours).The period of the "event" was shown in Fig. 4. From Fig. 5, it can be also seen that relatively higher CO data tend to appear with higher water vapor content.High CO with low O 3 data would be influenced by local pollution sources.But it can also be explained as polluted air without enough photochemical reaction for O 3 production.High water content would be related to rainy conditions with less sunlight, and the observed high CO but low O 3 air would be tended to have higher water content.

SO 2 Results
SO 2 concentrations were mostly below the detection limit of the SO 2 analyzer (0.06 ppbv) during the measurement period (not shown).This is consistent with Igarashi et al. (2004), when they reported SO 2 concentrations were lower than the detection limit of their instrument during the summer.On August 20-21, however, an increase in the concentration of SO 2 .was observed.Fig. 6 shows oneminute averages during this high SO 2 event.The combustion of fossil fuels, especially coal, is the main source of SO 2 emissions.Igarashi et al. (2004Igarashi et al. ( , 2006) ) observed episodic SO 2 spikes, especially from January to April.In their high SO 2 events, CO and 222 Rn (indicator of ground surface sources) increased, and they identified episodic high SO 2 through long-range transport of pollutants from the Asian continent where coal consumption is high.In Japan, anthropogenic SO 2 emissions are carefully controlled, except from the emissions of sea vessels.If polluted air from fossil fuel combustion sources is transported to the top of Mt.Fuji, the concentration of CO is expected to be high.During the high SO 2 event on August 20-21, however, CO concentration (9 min average data) did not increase as shown in Fig. 6.Also O 3 concentration did not fluctuate during the high SO 2 event Therefore, SO 2 transported from fossil fuel combustion is not a suitable explanation for this high SO 2 event.
Volcanoes are another source of SO 2 emissions and there are several active volcanoes in Japan.Miyake-jima, which is 170 km south of Mt.Fuji, erupted in July 2000 when aerosol (Naoe et al., 2003) and SO 2 (Igarashi et al., 2004) were observed from its volcanic plume at the summit of Mt.Fuji.Recently, Sakurajima volcano (Fig. 1, about 820 km away from Mt. Fuji) has become active and its plume was detected at a distant ground site (Noto peninsula) in Japan (Watanabe et al. 2015).Especially, there was a massive eruption and the volcanic ash rose to around 5000 m in altitude at 16:31 on August 18, 2013.Fig. 7 shows backward trajectories (a) and forward trajectories (b) during the high SO 2 event.For the backward trajectories, the starting altitudes were 4276, 4026, and 3776 m a.s.l., considering the uncertainty in backward trajectory calculations.The highest altitude trajectory passed over the Sakurajima volcano.For the forward trajectories, the starting altitude was selected as 3000 m and the starting time was at 10:00, 13:00, 15:00 and 18:00 on August 18.These air masses from Sakurajima arrived near Mt.Fuji around August 20-21, and this is consistent with the observation.Because Sakurajima volcano was quite active and erupted several times at that time (144 times eruptions were observed in one month on August in 2013), slight difference of the starting time and the assumption using the starting altitude at 3000 m for the forward trajectories will be not critical issues.At the same time of the high gaseous SO 2 peak, aerosol compositions indicated the influence of a volcanic plume (Ogata et al., paper in preparation).Therefore, it is reasonable to conclude that the plume from the Sakurajima volcano is related to observations at the top of Mt.Fuji during the high SO 2 event on August 20-21.
In Japan, there are several active volcanoes contributing to air pollution.In addition to these volcanoes there are various other natural sources of atmospheric trace species that can be observed at high altitudes.For example, smoke emissions from fires in the Siberian Forest during 2003 were observed from the summit of Mt.Fuji (Kaneyasu et al., 2007).Indeed, atmospheric monitoring at high-altitude mountain sites is useful in observing air quality.

CONCLUSION
We observed CO, O 3 and SO 2 concentrations at the summit of Mt.Fuji during the summer of 2013.CO and O 3 showed roughly similar variation in their concentrations, which can be explained by the air mass exchange between clean oceanic air and polluted continental air.Low CO and high O 3 were also observed in some cases when water vapor content in the air was low, indicating the influence of   stratospheric air.High SO 2 peaks were observed, which were identified as the volcanic plume of the Sakurajima volcano, located 820 km away.It is indeed evident that atmospheric observations taken at high-altitude mountain sites will be useful in monitoring regional ambient air in the free troposphere, including transport and intrusion events.

Fig. 2 .
Fig. 2. Time series of observed CO and O 3 (hourly mean data) during the summer of 2013.Air mass origin determined by backward trajectory is also plotted (right axis).N: North, NW: North West, CK: China Korea, C: China, SE: South East Asia, P: Pacific Ocean, U: unclassified.

Fig. 3 .
Fig. 3. Correlation plots of hourly mean CO and O 3 data for N, NW, CK, and C air masses (left) and for P and SE air masses (right).Guide lines assumed mixing of clean oceanic and polluted continental air are shown in both panels.

Fig. 4 .
Fig. 4. Time series of water vapor (right axis) with hourly mean CO and O 3 (left axis).Very low water vapor were observed.The period of the "event", high O 3 , low CO, low water vapor event (O 3 > 50 ppb, CO < 90 ppb, water < 4 g kg -1 ), was shown on x-axis.

Fig. 5 .
Fig. 5. Correlation between hourly mean CO and O 3 plotted by different symbols for water vapor content.The data with low CO and high O 3 had lower water vapor, indicating influenced by tropospheric air.The guide line in Fig. 3 is also shown by a solid line.

Fig. 6 .
Fig. 6.Time series of SO 2 data (1 min average, left axis), CO and O 3 data (9 min average, right axis) around the high SO 2 period on August 20-21.The initial time of backward trajectory in Fig. 7 is indicated by a bold arrow.The time format is local time (JST).

Table 1 .
Average and standard deviation of CO and O 3 for each backward trajectory category.