ANALYSIS OF A PERIOD OF ELEVATED OZONE CONCENTRATION REPORTED OVER THE VAAL TRIANGLE ON 2 JUNE 2013

A peak in the ambient ozone concentration was observed at three of the six ambient air quality monitoring stations in the Vaal Triangle Airshed Priority Area on the second of June 2013. The ozone peak was associated with elevated concentrations of particulate matter, including PM , PM and black carbon, but not sulphur dioxide, oxides of 10 2.5 nitrogen, carbon monoxide nor benzene. Back trajectory analysis using Hysplit showed that the air masses containing the high levels of ozone passed over areas influenced by coal fired power stations in the Waterberg, metal processing in the Bojanala region and high intensity fires 30km to 50km upwind of the stations.


Ambient Air Quality Monitoring Network
Tropospheric ozone is an important pollutant. It is involved in the production of photochemical smog, particularly over metropolitan areas and has a negative impact on crop productivity (Chameides et al., 1992). It plays an important role in atmospheric chemistry, influencing the production of the OH radical and thereby influences the oxidative capacity of the atmosphere. Tropospheric ozone is considered to have two major sources, firstly the incursion of ozone rich stratospheric air into the troposphere and secondly the de novo synthesis. In the presence of nitrogen oxides (NOx) ozone may be produced through the photochemical oxidation of hydrocarbons or carbon monoxide (CO) (Chameides et al., 1992;Seinfeld & Pandis, 2006). For this reaction to occur the presence of NOx, Volatile Organic Compounds (VOC) or CO and solar radiation is required (Chameides et al., 1992;Crutzen & Lelieveld, 2001).
Exposure to high concentrations of ozone has been implicated in increased premature mortality and reduced respiratory function (Silva et al., 2013;WHO, 2013). Due to the importance of ozone on human health, ozone is considered a criteria pollutant in South Africa (DEA, 2009). National Standards have been set with a limit value of 61 ppb as an eight hour running average (with 11 exceedances allowed annually). Many monitoring stations in South Africa report exceedances of the National Ambient Air Quality Standard for Ozone. nd On the 2 of June 2013 a peak in the ambient ozone concentrations which substantially exceeded the National Ambient Air Quality Standard was observed at three of the ambient air quality monitoring stations in the Vaal Triangle Ambient Air Quality Monitoring Network (Figure 1). Due to the extremely high levels of ozone that were observed it was decided that further investigation in the origins of this ozone episode was required.

Ambient Air Quality Monitoring Network
The Vaal Triangle is an urban and an industrial heartland of South Africa. The combination of industrial, domestic, transport, biomass burning, agricultural and other emission sources have led to degraded air quality over the area. This in turn directly impacts on the health and well-being of people residing in the Vaal Triangle (Trade and Industry Chamber, 2004). In 2007 a network of 6 ambient air quality monitoring stations were set up in the Vaal Triangle with stations located at Diepkloof, Kliprivier, Sebokeng, Sharpeville, Three Rivers and Zamdela (Sasolburg) Figure 2. on next page.

Atmospheric Conditions and Trajectory analysis
Due to the importance of atmospheric conditions, on the formation and movement of ozone and ozone precursors, an analysis of the atmospheric conditions at the time of the ozone exceedance was conducted. The synoptic charts for the central regions of South Africa were obtained from the South African Weather Service and information relating to the upper air was nd obtained for the balloon assent of the 2 of June 2013 at the Irene Weather Station located approximately 100km to the north east of the Vaal Triangle.
To assess the possible source locations of ozone precursors back trajectory analysis was performed using the Hysplit model for a 24 hour back trajectory. An ending height of 100m above ground level was chosen.

Remote Sensing of fire locations
The fire map was downloaded from Advanced Fire

Ozone Concentration 2 June 2013
The analysis of the 5 minute averaged ozone data from the stations at Sharpeville, Kliprivier, Sebokeng, Zamdela and Three Rivers stations (Figure 3) shows that a peak in the recorded ozone concentration occurred between ~9:00 and ~16:00 at the Sebokeng, Sharpeville and Zamdela Stations. At the Three Rivers and Kliprivier Stations this period of elevated ozone concentration was not observed. The maximum 5 minute ozone concentration recorded was 957 ppb for Sharpeville at 12:45, 784 ppb at Sebokeng at 13:05, 336 ppb at Zamdela at 12:40. Three Rivers and Kliprivier, which did not experience the period of elevated ozone concentration, had maximum concentrations of 51 ppb and 32 ppb at 14:00 and 14:30, respectively ( Figure 3). The timing of the maximum ozone concentration at the stations showing the incident of high ozone is slightly earlier than at the other two stations. The ozone concentration at the Eskom owned Waterworks Station, located in the Vaal Triangle Region showed a I n f o r m a t i o n S y s t e m ( A F I S ) a t http://southernafrica.afis.co.za. The system gives the near real-time and historical satellite-based fire monitoring information, covering multiple regions across the globe. The fires are detected by sensors on-board the Earth observation satellites, NASA MODIS and EUMETSAT MSG. A query was made on fires that occurred in southern Africa over a period of 24 hours on 2 June 2013, (i.e. fires that occurred during 2 June 2013 00:00 and 3 June 00:00) and the study area was zoomed into. AFIS then gives geographic location of fires and their intensity represented by different colours, yellow representing low, orange medium, red high and deep red extreme intensity (Van Rooi, 2012).
Each of the stations is equipped with instrumentation for the measurement of meteorological parameters and for the measurement of atmospheric chemical constituents ( Table 1). In addition, the Zamdela station is equipped with a Thermo Electron Multi-Angle Absorption Photometer instrument (MAAP) for the measurement of atmospheric black carbon. All instrumentation is regularly calibrated according to the SANAS TR07-03 requirements (SANAS, 2012) using certified calibration gas traceable to the national reference standard (Air Liquid) and a Thermo Electron 146i Multigas Calibrator. At the time of the nd period of high ozone concentration on the 2 of June the instrument at Diepkloof was offline for repairs, therefore the Diepkloof station will not be considered in this analysis.

Other Chemical Parameters
The ambient concentration of PM at the three 10 stations where the elevated ozone occurred showed an approximate 5 fold elevation during the period from 10:00 to 15:00 ( Figure 4). A peak 5 minute PM 10 concentration of 464 µg/m³ was observed at Sharpeville at 13:20, 35 minutes after the maximum recorded ozone concentration, but well within the period of elevated ozone. The elevation of the PM 10 concentrations was much less pronounced at Three Rivers and Kliprivier which did not display the elevation in ozone concentration.  Black carbon is a strong indicator for combustion. During the period of elevated ozone, the black carbon concentration increased from approximately 1 µg/m³ to approximately 3 µg/m³, this is during the part of the day when the concentration of black carbon is expected to decrease as the atmosphere is more thoroughly mixed and localized sources of combustion such as domestic burning are not as prevalent as they would be during the evening and early morning hours. The elevation of the black carbon concentration that would typically be expected in the early evenings and that is associated with domestic burning is seen from approximately 18:00. During the period of the peak in ozone there is a strong increase in the ratio of PM to black carbon, 10 which is not seen for the ratio of PM to black carbon.

2.5
This implies that during this period there was a change in the composition of the particulate matter with an increase in the large size fractions (Figure 7).   The concentrations of CO were not elevated at any of the sites during the period of elevated ozone concentration. Increases in the CO concentration in the early evening are typically associated with domestic combustion activities in the low income residential areas surrounding the stations. Figure 9 on next column.
The concentration of NO does not show a significant x elevation during the period of high ozone concentration, while there is an increase between 12:00 and 14:00 this does not discriminate between the stations that show higher levels of ozone and those that do not. Figure 10 on next column.

Atmospheric Conditions over the Vaal
Triangle on 2 June 2013

Back trajectory and wind analysis
A back trajectory analysis of the air mass carrying the elevated concentrations of ozone was conducted using Hysplit trajectory model where the back trajectory for a period of 24 hours was analysed for all stations. The back trajectory analysis ( Figure 14) shows that the air mass originated in the Waterberg region directly over the town of Lephalale, which is the location of a large coal fired power station, it then moved due south, passing over the Bojanala District and the Platinum mining activities in the region before passing over the gold mining region of Randfontein, located to the west of Johannesburg. During the time of the elevated ozone concentration the wind at the monitoring stations was from the north-west and north-north-west with a wind speed of between 5.7m/s to 11m/s ( Figure 15). During the time of the reported period of elevated ozone the synoptic conditions in South Africa were represented by a strong frontal system that was crossing the country from the south west, preceded by a surface trough extending from the north west to the south east over the interior of the country. The area ahead of the front typically experiences stable conditions with the occurrence of dry descending air. The synoptic chart for the 2 June 2013 is presented in Figure 12. The atmospheric profile was obtained for the upper air balloon assent on the second of June. The vertical profile taken at the Irene Station shows that there is a strong inversion layer between 2133m and 2308m above ground level ( Figure 13). The tropopause is situated at 13000m above ground level. Examination of the radiosonde measurements from the previous week indicate that the inversion was in place for a number of days prior to the reported incident. Figure 13 atmospheric temperature profile taken at the Irene weather Station

Fire Occurrence
Satellite Information from EUMETSAT MSG was used to identify the location and intensity of veld fires in the region which may have impacted on the production of ozone precursors and ultimately resulted in the extremely high concentrations of tropospheric ozone that were observed over some of the stations in the Vaal Triangle on the 2 June 2013. During the second of June there was considerable biomass burning in the central regions of South Africa. Figure 13 shows the EUMETSAT MSG output over the Vaal Triangle region. Of those fires, six were within the direct path of the air mass that carried the high ozone concentration those being the fires located to the north west of the stations as shown in Figure 16. show significant changes associated with the elevation in ozone concentration. Back trajectory analysis indicates three potential sources for the ozone precursors that fell within the path of the air mass arriving at the monitoring stations. Those include; 6 wild fires that were active 30 km to 40 km to the north west of the monitoring stations, the mineral processing activity in the Bojanala district surrounding Rustenburg and the coal fired power station at Lephalale approximately 300km to the north of the monitoring stations where the air mass originated 24 hours previously.
Analysis of the corresponding concentrations of other pollutants provides some indication as to which of the potential sources of the precursors is responsible for the high ozone concentrations reported on the day. It is expected that if the source of the ozone precursors responsible for the formation of the ozone peak observed at the Sharpeville, Sebokeng and Zamdela stations was due to industrial activity such as the combustion of coal or the processing of sulphate rich minerals, the peak in ozone concentration would be associated with increases in the concentrations of SO and NO . This however did not occur. It is known 2 x that the NO and SO in power station plumes can be x 2 removed rapidly through chemical processing (Zhou et al., 2012). The SO and NO loss is reported to be 2 x greater under cloudy conditions.
However if the production of ozone is associated with a biomass combustion source the ozone peak would associated with increases in the concentrations of PM and specifically black carbon as a product of incomplete combustion. In this case the temporal changes in the concentration of the other monitored pollutant indicate that the source of the ozone precursors is likely to be biomass burning. The wind direction and location of fires in the Vaal Triangle region strengthen this explanation. Analysis of ozone production in African biomass burning plumes, has reported elevations of the ozone concentration of up to 98ppb 18km downwind of savanna fires in South Africa (Yokelson et al., 2003).

Conclusion
The management of ambient air quality in complex environments such as the Vaal Triangle can be difficult; this is especially true for secondary pollutants such as ozone, which are formed through the interaction of numerous chemical precursors and the environmental conditions. This study presents an analysis of the chemical and environmental conditions during a period of extremely high ozone concentration. Evidence suggests that biomass burning played an important role in the period of elevated ozone, but other factors such as the long range transport of precursors from power generation and mineral processing activities may have played a role in allowing for such high concentrations of ozone to be formed. Further modelling to better understand the chemical processes and the precursor concentration requirements would help in understanding the processes that occurred and identify the precursor sources, thus enabling focused management interventions on the local or regional scales.