The value of intensive measurement campaigns in helping to understand and characterise local atmospheric composition and air quality has been recognised since as early as 1969, when the Los Angeles Smog Project took place (Whitby et al., 1972b). Since then, many such campaigns have focused on understanding the formation of photochemical smog in the most polluted cities worldwide, with early efforts concentrated in the USA (e.g. Gray et al., 1986; Husar et al., 1972; Whitby et al., 1972a). The formation of secondary organic aerosol has also been of particular interest, with many studies using elemental carbon (black carbon) as an indicator of primary emissions; when the ratio of organic carbon to elemental carbon in the sampled air is higher than expected from the ratio of the primary emissions, secondary organic aerosol formation is indicated (Castro et al., 1999; Gray et al., 1986; Turpin and Huntzicker, 1995).

In Australia, there have been a number of studies aimed at improving our understanding of ozone chemistry in the cleaner Southern Hemisphere atmosphere (Galbally et al., 2000; Monks et al., 1998), secondary aerosol formation (Cainey et al., 2007) and other air quality issues, such as air toxins and smoke (Hinwood et al., 2007; Keywood et al., 2015). There have also been some air quality studies specifically aimed at testing the Australian Air Quality Forecasting System (Cope et al., 2004) in Sydney (Hess et al., 2004) and Melbourne (Tory et al., 2004). The primary focus of these studies was to test the prediction of ozone levels in the urban environment (Cope et al., 2005). More recent studies have examined regional air quality in Wollongong (Buchholz et al., 2016) and the effect of a major fire event on air quality in Sydney and Wollongong (Rea et al., 2016). There have also been Australian campaigns focused on understanding aerosol formation and composition in the urban environment, e.g. (Cheung et al., 2011, 2012) coastal environments (Cainey et al., 2007; Fletcher et al., 2007; Modini et al., 2009) and within eucalypt forests (Ristovski et al., 2010; Suni et al., 2008). In addition, there have been some detailed studies to characterise the concentrations of VOCs in the clean background atmosphere in the Australasian region (Colomb et al., 2009; Galbally et al., 2007; Lawson et al., 2015).

In this overview paper, we describe a measurement campaign in Wollongong, a small Australian coastal city with approximately 292 000 residents. The Wollongong region is bounded by ocean to the east and by a steep escarpment covered in eucalypt forest to the west. The coastal plain is roughly triangular in shape, being very narrow in the north where the escarpment meets the sea and roughly 20 km wide in the south. The region spans about 50 km of coastline.

The MUMBA campaign involved collaboration between three Australian research groups: the University of Wollongong, the Commonwealth Scientific and Industrial Research Organisation (CSIRO) and the Australian Nuclear Science and Technology Organisation (ANSTO), as well as one research organisation from New Zealand (GNS Science). MUMBA was designed to provide a comprehensive characterisation of the local atmosphere that could test the capabilities of air quality models to forecast atmospheric composition. Influences from nearby ocean sources, urban emissions and the biogenic emissions from the surrounding eucalypt forests were expected to impact the site. This campaign aimed to make detailed measurements of atmospheric composition under the combined influence of these different sources, all of which typically affect the populated regions on the eastern coast of Australia.

The MUMBA campaign included instruments that were run at several different nearby sites. The main measurement site (34.397∘ S, 150.900∘ E) of the MUMBA campaign was located in a suburban area of Wollongong approximately 0.5 km from the ocean. The instruments were located in and adjacent to an unused hut located at the University of Wollongong's campus east (see Fig. 1a). Most of the instruments sampled from a mast at a height of ∼ 10 m above the surrounding ground level (also shown in Fig. 1a). Immediately surrounding the measurement site is a grassy plain with a suburban road to the east and a strip of forested parkland beyond, before the sand dunes and ocean. Prevailing easterly sea breezes brought air masses from the ocean to the site during the day. Urban influences from the local metropolitan area and a large industrial area, including a steelworks, typically occurred in still conditions or with southerly winds. The steelworks and surrounding industry is a large source of PM2.5 and CO, whilst traffic dominates the remainder of the urban pollution sources. The steep forested escarpment is about 3 km directly to the west of the site and approximately 400 m high with the area beyond dominated by eucalypt forest, such that westerly winds brought strong biogenic signals. The population density within the surrounding area of New South Wales (NSW), including Wollongong and Sydney, is shown in Fig. 1b.

The locations of the different measurement sites are shown in Fig. 1c.

In addition, ANSTO provided measurements of atmospheric radon concentrations from Warrawong (34.48∘ S, 150.89∘ E), an industrial suburban site located approximately 10 km to the south of the main MUMBA site. The use of radon to characterise boundary layer mixing (Chambers et al., 2011) is likely to be especially useful for testing air quality models due to the challenges of modelling within the complex topography of coastal areas. The locations of all of the sites used in the MUMBA campaign are marked on the satellite view of the region shown in Fig. 1c.

A large range of instrumentation was deployed to enable a detailed characterisation of atmospheric composition during the campaign. All measurements made during the campaign are listed in Table 1, along with the dates of operation for each instrument. MUMBA operated in two distinct stages with most gas-phase and meteorological measurements running throughout the 8 week campaign and aerosol-phase measurements added in the second half of the campaign. A few instruments operated for different time periods, and these are distinguished in Table 1. All data are available from PANGAEA (https://doi.pangaea.de/10.1594/PANGAEA.871982) as hourly averages unless otherwise specified. Further details of the instruments are given in the Appendix along with a second table that lists the specific VOCs measured during the campaign and their limits of detection. Note that some instruments can produce negative values when the concentrations are close to the detection limit. Negative concentration values (although non-physical) have not been removed from the MUMBA dataset because they are indicative of the instruments' true performance; removing negative values will produce small positive biases in the calculations of longer-term average concentrations. Also available from PANGAEA are the radon measurements made at Warrawong by ANSTO and the air quality data from the Wollongong Office of Environment and Heritage (OEH) station. The FTIR spectrometer uses a drier on the inlet, and it measured the mole fraction in dry air; the other gas-phase instruments measured in ambient air. All measurements are reported in local standard time (UTC +10).

Measurements were also available from the OEH air quality station at Wollongong (34.419∘ S, 150.886∘ E). Additional instruments were operated nearby at the University of Wollongong's main campus (at 34.406∘ S, 150.897∘ E) (Buchholz et al., 2016) and at the nearby Science Centre (34.401∘ S, 150.900∘ E), but the observations are not included here. Data from the University of Wollongong include retrievals of total column amounts of trace gases from the Total Carbon Column Observing Network (http://www.tccon.caltech.edu/) and the Network for the Detection of Atmospheric Composition Change (http://www.ndsc.ncep.noaa.gov/) as well as in situ greenhouse gas measurements (http://doi.pangaea.de/10.1594/PANGAEA.848263). The instrument installed at the Science Centre was a multi-axis differential optical absorption spectrometer, and the data are available from the authors upon request. The Australian Bureau of Meteorology (BOM) operates an automatic weather station (AWS) at Bellambi (34.37∘ S, 150.93∘ E). Again, the data are not included here but can be requested from the BOM if needed.

The summer of 2012–2013 was the hottest summer on record for Australia at the time (White and Fox-Hughes, 2013). There were two extremely hot days in the Wollongong region during MUMBA, with maximum temperatures of 40.4 ∘C on 8 January and 42.4 ∘C on 18 January 2013 recorded at Bellambi AWS (both below the record of 43.7 ∘C set on 1 January 2006). The campaign encompassed the wettest January day on record for the region, with 139 mm of rain falling at Bellambi AWS between 08:00 on 28 January and 08:00 on 29 January 2013 (see the top panel of Fig. 2).

The lower panel of Fig. 2 shows the mean hourly temperature recorded from the 10 m mast at the MUMBA site over the campaign. The two extremely hot days can be clearly seen in this figure. The mean daily maximum temperature during January 2013 was 25.7 ∘C, which is 0.9 ∘C above the long-term average of 24.8 ∘C and in the 95th percentile of monthly mean maximum temperatures for January at Bellambi AWS (using data from 1988 to the present).

The average wind speed recorded at the MUMBA site during the campaign was 2.8 ms-1, and the maximum hourly averaged wind speed recorded was 9.2 ms-1. The 1st, 2nd (median) and 3rd quartiles of the wind speed were 1.4, 2.6 and 3.9 ms-1 respectively. Figure 3 shows the composite diurnal cycles of the wind speed and wind direction as measured at the main MUMBA site. The general pattern was a relatively strong sea breeze during the day (∼ easterly winds of 3–4 ms-1) and calmer conditions overnight. Westerly winds were more frequent during nighttime (although northeasterly winds sometimes persisted into the night). This pattern was repeated all over the local region (as shown in data from OEH air quality sites and from the University of Wollongong) (Guérette, 2016).

The third panel in Fig. 3 shows the composite diurnal cycle of radon measured at the ANSTO site in Warrawong. The radon plot shows a build-up at night with a peak in the early hours of the morning, indicating a shallower and more stable boundary layer at night than during the day, with the boundary layer at its shallowest around 05:00 or 06:00. During the day, due to heating at the surface and other processes, the boundary layer grows deeper and more turbulent; this is reflected in the lower radon values observed during the day. Minimum radon levels in the afternoon are also influenced by the fetch of the air reaching the site, with air that has travelled over the ocean containing less radon than air that has travelled over land (Chambers et al., 2015). In the Wollongong region, an increased boundary layer height and strong sea breezes combine to produce the low radon levels observed in the afternoon.

Comparisons of the winds measured at the MUMBA site during the campaign to simultaneous measurements at the three air quality sites operated by the Office of Environment and Heritage in the area (at Wollongong, Kembla Grange and Albion Park) indicated that the wind patterns observed at the MUMBA site were generally representative of the region as a whole (Guérette, 2016). The long-term average wind data at 15:00 each day are publicly available from the Bellambi AWS from 1997–2010, and these were used for comparison with the wind data recorded at this time throughout January during the campaign. The MUMBA site in January 2013 was characterised by slightly less frequent northerly winds and slightly more frequent westerly winds than expected from the long-term average at Bellambi, but otherwise the wind patterns were very similar in the two records. The MUMBA site experienced lower wind speeds than the long-term averages at Bellambi (but this may be due to location differences rather than atypical weather patterns) (Guérette, 2016). Thus we conclude that the measurements made at the MUMBA site during the campaign should be broadly representative of the region as well as of the summer season.

On a larger scale, the dominant circulation pattern during MUMBA was anticyclonic with the main fetch being principally oceanic (as opposed to continental), which is typical of summer (Chambers et al., 2011). This is illustrated in Fig. 4, which shows a gridded back trajectory frequency plot for 96 h precalculated back trajectories made available for Wollongong through the openair package (Carslaw and Ropkins, 2012). The trajectories were calculated using the HYSPLIT trajectory model (Hybrid Single Particle Lagrangian Integrated Trajectory Model; http://ready.arl.noaa.gov/HYSPLIT.php) every 3 h from an initial height of 10 m and propagated backwards in time for 96 h using the global NOAA-NCEP/NCAR reanalysis meteorological fields at 2.5∘ of horizontal resolution. The surface of the plot is coloured by the percentage of the total trajectories which pass through each grid box.

The MUMBA campaign was designed to characterise the atmospheric composition at the ocean–forest–urban interface and thereby provide a dataset that could be used to test the skill of atmospheric models within a coastal environment. In this section, the major urban, marine and biogenic sources that influence the atmospheric composition in the region are described.

The dominant anthropogenic sources in the region are the Port Kembla steelworks, located approximately 10 km south of the main MUMBA site (for PM2.5, PM10, CO, NOx and SO2) and motor vehicles (for NOx, CO and VOCs) (http://www.npi.gov.au/npidata). The ocean lies to the east of the site and large forested areas are to the west. Outflow from the Sydney basin (80 km to the north) may accompany winds from the northeast.

The impact of the different air masses sampled can be illustrated using a bivariate polar plot, which shows how a pollutant varies by wind speed and wind direction as suggested by Carslaw et al. (2006). Figure 5a shows a bivariate polar plot for CO measured from the main MUMBA site throughout the campaign. Several distinct regions are evident, with the most obvious being the very high amounts of CO that are measured when the site experiences southerly winds with speeds between 2 and 6 ms-1. This direction brings air masses over central Wollongong and also over the industrial area centred on the steelworks at Port Kembla. In contrast, easterly to south-southeasterly winds bring very low amounts of CO to the MUMBA site as the air masses come from the Pacific Ocean. There were a number of occasions during the campaign when easterly winds brought predominantly marine air to the measurement site. These periods were identified by using radon values below a threshold of 200 mBq m-3, indicating minimal terrestrial impact in agreement with back trajectories. One episode in particular, on 26 December 2012, lasted several hours and was characterised by greenhouse gas concentrations similar to those measured in December 2012 at the Cape Grim baseline air pollution station on the northwest tip of Tasmania, Australia (40.683∘ S, 144.689∘ E) (http://www.csiro.au/greenhouse-gases/).

CO mole fractions from the northeast (that also come off the ocean) are nearly double those from the south-southeast, indicating that the MUMBA site may be influenced by outflow from the Sydney basin 80 km to the north. Elevated CO is also measured from the northwest in the direction of the nearest suburban shopping centre, multi-lane road and local industrial sites (including a cokeworks and mining operations). In contrast, relatively low concentrations are seen from the southwest where there is a steep escarpment and eucalypt forests beyond.

Figure 5b shows the polar bivariate plot for toluene, which is predominantly emitted from motor vehicles and is not emitted from the steelworks. The plot shows the largest concentrations with low wind speeds, as is indicative of local sources building up in the nocturnal boundary layer; however, there is a directional bias with much cleaner air to the east. This is due to clean marine air coming from the east and is also obvious in the low amounts of toluene coming from all wind speeds from the southeast. In contrast, there are slightly higher mole fractions of toluene that accompany winds from the northeast, again indicating possible outflow from Sydney or more local pollution to the north that is brought in on the sea breeze.

Figure 5c shows the polar bivariate plot for NOx, which shows a mixture of the features seen in the toluene and CO plots; this is indicative of a mixture of traffic and industrial sources as expected.

In Fig. 6, polar bivariate plots are shown for the main criteria pollutants of concern within the airshed (PM2.5 and O3) along with the most significant biogenic volatile organic compounds, isoprene (PTR-MS m/z 69) and monoterpenes (PTR-MS m/z 137). Both isoprene and monoterpenes show very elevated concentrations with strong northwesterlies, which occurred on the two extremely hot days (8 and 18 January 2013). The monoterpenes are also high with still winds because (unlike isoprene) these compounds are also emitted during the night and hence build up in the nocturnal boundary layer. Also, under more stable nighttime conditions, katabatic flow down the escarpment will bring air predominantly influenced by the eucalypt forests to the site.

The highest PM2.5 concentrations are seen with strong to moderate winds from the south, which bring industrial sources from the Port Kembla steelworks. Elevated PM2.5 is also seen with northwesterly winds that bring biogenic influences from the escarpment and densely forested regions beyond. The highest O3 concentrations are also seen with the hot northwesterly winds, with the influence of NOx titrating out the O3 clearly seen with low concentrations observed at low wind speeds and with wind from the south. The high O3 and PM2.5 values that accompany the high levels of isoprene and monoterpenes imply that biogenic influences are important for both O3 formation and secondary organic aerosol formation in the region. This may be due to the VOC-limited environment (the formaldehyde-to-NOx ratio averaged 0.3 over the campaign) coupled with the fact that anthropogenic emissions of VOCs are relatively low in the area, so that biogenic VOCs are extremely important to the overall budget. Despite the importance to air quality, biogenic emissions from Australian eucalypt forests are poorly understood (Emmerson et al., 2016), and further research is needed to better characterise biogenic emissions in this region of Australia.