Long-term halocarbon observations from a coastal and an inland site in Sabah, Malaysian Borneo

Short-lived halocarbons are believed to have important sources in the tropics, where rapid vertical transport could provide a significant source to the stratosphere. In this study, quasi-continuous measurements of short-lived halocarbons are reported for two tropical sites in Sabah (Malaysian Borneo), one coastal and one inland (rainforest). We present the observations for C 2Cl4, CHBr3, CH2Br∗2 (actually∼ 80 % CH2Br2 and∼ 20 % CHBrCl 2) and CH3I from November 2008 to January 2010 made using our μDirac gas chromatographs with electron capture detection (GC-ECD). We focus on the first 15 months of observations, showing over one annual cycle for each compound and therefore adding significantly to the few limited-duration observational studies that have been conducted thus far in southeast Asia. The main feature in the C 2 l4 behaviour at both sites is its annual cycle, with the winter months being influenced by northerly flow with higher concentrations, typical of the Northern Hemisphere, and with the summer months influenced by southerly flow and lower concentrations representative of the Southern Hemisphere. No such clear annual cycle is seen for CHBr 3, CH2Br∗2 or CH3I. The baseline values for CHBr 3 and CH2Br∗2 are similar at the coastal (overall median: CHBr 3 1.7 ppt, CH2Br∗2 1.4 ppt) and inland sites (CHBr 3 1.6 ppt, CH2Br∗2 1.1 ppt), but periods with elevated values are seen at the coast (overall 95th percentile: CHBr 3 4.4 ppt, CH2Br∗2 1.9 ppt), presumably resulting from the stronger influence of coastal emissions. Overall median bromine values from [CHBr 3 × 3] + [CH2Br∗2 × 2] are 8.0 ppt at the coast and 6.8 ppt inland. The median values reported here are largely consistent with other limited tropical data and imply that southeast Asia generally is not, as has been suggested, a hot spot for emissions of these compounds. These baseline values are consistent with the most recent emissions found for southeast Asia using the p-TOMCAT (Toulouse Off-line Model of Chemistry And Transport) model. CH 3I, which is only observed at the coastal site, is the shortest-lived compound measured in this study, and the observed atmospheric variations reflect this, with high variability throughout the study period. Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
Reactions involving halogen compounds play an important role in the chemistry of both the stratosphere (e.g. Molina and Rowland, 1974;Yung et al., 1980) and the troposphere (e.g. Read et al., 2008). The sources of halogens are many and varied with important anthropogenic and natural components for many compounds (Montzka and Reimann, 2011). Total atmospheric bromine concentrations are generally dominated by the halons (compounds of only anthropogenic origin widely used for fire control) and CH 3 Br (used in agriculture, but with many sources). Their atmospheric concentrations are well known and their lifetimes long enough to know with reasonable accuracy how much of these gases reaches the stratosphere. It was thus puzzling to find that they could only account for 70-80 % of the bromine observed in the stratosphere (Wamsley et al., 1998;Dorf et al., 2006), though there remains uncertainty as to whether the discrepancy is that large (Kreycy et al., 2013).
Since there are no other known long-lived sources, the finger of suspicion points to the shorter-lived CHBr 3 and CH 2 Br 2 which are observed in high concentration in the marine boundary layer and are naturally produced in the surface ocean (e.g. Yokouchi et al., 2005;Butler et al., 2007;Zhou et al., 2008;O'Brien et al., 2009). For these compounds to be the source of the missing 20-30 % of the stratospheric bromine budget, they (or their breakdown products) have to be rapidly transported up to the stratosphere. The only credible place for this to occur is at low latitudes where upward transport can be very rapid in the intense tropical convection. This possibility has been investigated in a number of measurement (Schauffler et al., 1999;Laube et al., 2008;Park et al., 2010;Brinckmann et al., 2012) and modelling (Levine et al., , 2008Aschmann et al., 2009;Hosking et al., 2010;Hossaini et al., 2010Hossaini et al., , 2012Hossaini et al., , 2013 studies, and overall it seems likely that the short-lived bromocarbons are responsible for the apparent deficit. One of the limitations on these studies is the scarcity of high-quality measurements in the regions of most importance, i.e. in the highly convective areas. The majority of reported tropical boundary layer measurements have been made using whole-air samplers or short-term deployments of gas chromatograph (GC) systems (Yokouchi et al., 2005;O'Brien et al., 2009;Pyle et al., 2011b), or ship-borne measurements made during research cruises (Butler et al., 2007;Carpenter et al., 2009;Quack et al., 2004Quack et al., , 2007Yokouchi et al., 1997;Mohd Nadzir et al., 2014;Brinckmann et al., 2012). No long-term and continuous measurements in the tropics have been reported to date. Accordingly, we have set up a small network of instruments to make continuous measurements at a number of sites in the western Pacific/southeast Asian region. This region was chosen as it is where convection is strongest, especially during Northern Hemisphere (NH) winter (e.g. Newell and Gould-Stewart, 1981; Gettelman and Forster, Figure 1. Site location map for Sabah. The inland rainforest site on Bukit Atur is shown by the blue circle, the coastal site near Tawau is a red circle and the coastal site at Kunak (green circle) is also shown as featured in Pyle et al. (2011b). The Tawau site is ∼ 85 km to the south of Bukit Atur. The inset (lower left corner of main panel) shows the southeast Asia region with a red square to show coverage of the main map. Fueglistaler et al., 2009). In addition, an early model study suggested high CHBr 3 emissions and atmospheric concentrations here, due to the many islands and therefore large length of coastlines found in this region (Warwick et al., 2006), although this estimate has since been revised downward (Pyle et al., 2011b). The first two instruments were installed in 2008 in the State of Sabah in Malaysian Borneo, one at a coastal site and one inland (Fig. 1). Early measurements were reported in Pyle et al. (2011a, b). Here we report measurements from the subsequent years, focussing on the period from late 2008 to early 2010. In addition to CHBr 3 and CH 2 Br * 2 , we report our observations of C 2 Cl 4 , a manufactured compound which we use to identify air with anthropogenic influence, and CH 3 I, a compound produced naturally in the ocean which has a short atmospheric lifetime (7 days; Montzka and Reimann, 2011).
In Sect. 2 we describe the instruments used in this study, their calibration and the measurement quality reported here. Section 3 contains a discussion of the measurement sites, including their physical characteristics, while in Sect. 4 we describe the modelling tools (NAME, p-TOMCAT) used in the interpretation of the data. The measurements for C 2 Cl 4 , CHBr 3 , CH 2 Br * 2 and CH 3 I are presented in Sect. 5, with particular emphasis on their variability, and with air mass history maps to aid in interpretation. We also consider the implications of the measurements, including an analysis of the total A. D. Robinson et al.: Long-term halocarbon observations from a coastal and an inland site in Sabah 8371 organic bromine in short-lived bromocarbons at each site. Finally, Sect. 6 summarises our findings and ongoing/future work.

Instrument
The instruments used for these observations were the original µDirac gas chromatographs built by us for use on balloons, with electron capture detection (GC-ECD), in a modified configuration appropriate for extended ground-based deployment as described in Pyle et al. (2011b). The instrument is described in detail in Gostlow et al. (2010). Briefly, each sample is pre-concentrated using a dual-bed adsorbent trap (1 mg Carboxen-1016 60/80 mesh followed by 1 mg Carboxen-569 20/45 mesh). The trap is then heated and the gaseous constituents are separated on a 10 m long, 0.18 mm inside-diameter (ID) capillary column (Restek MXT 502.2) prior to measurement in the ECD. As the instruments are designed to run unattended, the control software runs chromatograms according to a pre-determined sequence of samples bracketed by calibration and blank chromatograms. The sequence includes frequent calibrations using the same volume as the samples (20 mL) for correction of instrument sensitivity drift and precision determination. The sequence also runs calibrations at a range of different volumes so that instrument response curves are generated for each target compound to allow for non-linearity. The response curves are fit using a third-order polynomial. Chromatograms and system data are stored on a host computer for later analysis using in-house software to determine peak heights for the target compounds, which are then converted into mixing ratios by comparison with the calibration standards. Mixing ratios are reported as dry-air mole fraction in parts per trillion.
Measurement performance is monitored for each target compound during the analysis of the chromatograms, and the typical performance for the measurement period reported here is summarised in Table 1. For C 2 Cl 4 the typical measurement precision (1 SD) is 2.5 and 4.0 % at Bukit Atur and Tawau respectively. The typical measurement precision of CHBr 3 is 6.0 and 7.5 % at Bukit Atur and Tawau respectively, and those for CH 2 Br * 2 are 5.0 and 6.5 %. Sample concentrations for C 2 Cl 4 , CHBr 3 and CH 2 Br * 2 are always above the limit of detection at both sites. CH 3 I is only measured at Tawau (the µDirac at Bukit Atur did not achieve satisfactory chromatographic separation for reliable quantification) and the typical precision is 11 %. The CH 3 I peak is small in the sample chromatograms, so particular care is taken in assessing its signal-to-noise ratio (SNR). The SNR varies over time, depending on the instrument condition and environment, and is calculated separately for each analysis period (typically 1-2 weeks) for which concentrations are calculated. Between January and June 2009, the estimated detec-tion limit (defined using a SNR of 10) decreases from 0.9 to 0.3 ppt as a result of reduced noise in the chromatograms, and for the rest of 2009 it is below 0.2 ppt.
The chromatographic set-up uses a 10 m column for these measurements, which provides a good separation for C 2 Cl 4 and CHBr 3 . However, the CH 2 Br 2 peak co-elutes with that of CHBrCl 2 , and so we report a combined value called CH 2 Br * 2 which we estimate to be 10-30 % higher than the CH 2 Br 2 value alone (O'Brien et al., 2009). This co-elution problem is further discussed in Sect. 5.3 of this paper and has been previously assessed in Gostlow et al. (2010). In a recently improved version of the instrument, where a 20 m long column is used, the CH 2 BrCl 2 and CH 2 Br 2 peaks are partly separated and so individual values can be derived by using two overlapped Gaussian fits. By analysing calibration chromatograms from three of our newer instruments using the longer column, we find that the ECD is significantly more sensitive to CH 2 Br 2 than it is to CHBrCl 2 , the peak height per parts per trillion for CHBrCl 2 being typically 80 % of that for the CH 2 Br 2 peak, confirming the assumption made in O'Brien et al. (2009). Measurements of CH 2 Br 2 and CHBrCl 2 using instruments with 20 m columns in similar tropical locations show the ratio (10-30 %) to be still valid. For example, in a 2010 cruise from Singapore to New Zealand we found that, for the southeast Asia samples, the average CHBrCl 2 mixing ratio was 0.18 ppt, ∼ 16 % of that for CH 2 Br 2 at 1.12 ppt (unpublished data). In this case, had a 10 m column been used, the CH 2 Br * 2 value from the combined peaks would have been 1.26 ppt (allowing for the 80 % lower sensitivity to CHBrCl 2 compared to CH 2 Br 2 ). This compares favourably to the value of 1.30 ppt for the sum of the separately calculated values for CH 2 BrCl 2 and CH 2 Br 2 . Other studies in the southeast Asia region have also reported similar values for CHBrCl 2 and CH 2 Br 2 : during the OP3 campaign at Bukit Atur in July 2008, gas chromatographymass spectrometry (GC-MS) observations made by the University of East Anglia (UEA; Pyle et al., 2011a) found an average CHBrCl 2 mixing ratio of 0.25 ppt, ∼ 27 % of that for CH 2 Br 2 (0.91 ppt) and during a 2009 cruise in the Strait of Malacca, South China and Sulu-Sulawesi seas, mixing ratios of CHBrCl2 were on average 33 % of those for CH 2 Br 2 (Mohd Nadzir et al., 2014) The instrument based at the coastal site developed a problem with desorption efficiency caused by incomplete thermal desorption of the target molecules from the adsorbent. This affected the quantification of the CH 2 Br * 2 peak by carry-over from one chromatogram to the next. However, we are still able to use calibration chromatograms for CH 2 Br 2 as these are preceded by a blank chromatogram. About one in every 10 sample chromatograms is also usable for the CH 2 Br * 2 peak as these are also preceded by a blank. As a result of this problem CH 2 Br * 2 is only reported for about 1 in every 10 samples at Tawau. The C 2 Cl 4 and CHBr 3 peaks are not affected by this problem at Tawau as these molecules are Table 1. Dirac target compounds and measurement performance at Bukit Atur and Tawau. Atmospheric lifetimes are taken from Tables 1-4 in Montzka and Reimann (2011); the lifetime value for CH 2 Br * 2 is actually that for CH 2 Br 2 ; estimated accuracies are ±1 standard deviation; typical measurement precisions are ±1 standard deviation of six calibration chromatograms made in a single day; limit of detection is based on a signal-to-noise ratio of 10 : 1; and n/a means no data available. always completely retained on the weaker Carboxen-1016 at the front of the dual adsorbent bed.

Calibration
Samples are calibrated by running frequent chromatograms from cylinders of clean natural air, previously filled at Niwot Ridge (Colorado, USA), at an elevation of 3.5 km above sea level. The cylinder air is subsequently enriched with known concentrations of the target compounds. The cylinders (one for each instrument) are supplied by the Earth System Research Laboratory (ESRL) within the National Oceanic and Atmospheric Administration (NOAA) so that the reported mixing ratios are linked directly to the NOAA halocarbon calibration scales. Figure 2 summarises the period of data coverage for this study together with a timeline of the calibration cylinders in use at the two sites. The calibration gas in use at Tawau for the entire period of this study is in a high-quality, humidified stainless steel cylinder (Essex Cryogenics Inc., cylinder SX-3542). The composition of this cylinder for the four compounds of interest is reasonably stable (  (Table 2b). The agreement with the Aculife standard is much better in the later comparison, which was performed in Sabah under the same conditions as the atmospheric measurements. We do not know why the first comparison should have been poor (it was done in Cambridge just prior to shipping), though we suspect it is related to temporary adsorption on the canister walls. Given the closer agreement between the baseline measurements of C 2 Cl 4 , CHBr 3 and CH 2 Br * 2 at Tawau and Bukit Atur (Sect. 5.1) using the values derived in the second comparison (last column of Table 2b), we choose to use this for the final analysis, rather than the values from the October 2008 check (or an average of the two checks). From May 2009 the NOAA Aculife cylinder (AAL-70424) used for the decanting is used directly at Bukit Atur. This cylinder appears to be more stable than the decanted cylinder though not as stable as the SX-3542 cylinder in use at Tawau (Table 2c). For the analysis of our Bukit Atur data from May 2009 to January 2010 we use the values from the check against UEA made in April 2009 using their NOAA SX cylinder, as this is closest in time to the period of the measurements. We also have more confidence in the comparison made at UEA (April 2009) under stable laboratory conditions using their GC-MS than in an incomplete comparison made at Bukit Atur (November 2012) over 3 years later. During June and July 2008 our measurements at Bukit Atur, analysed using the same NOAA (SX-3542) standard (later used at Tawau), compared well with GC-MS measurements made by UEA (Gostlow et al., 2010;Pyle et al., 2011a, b).
In addition, we took part in an inter-comparison of measurements by several laboratories using a round-robin NOAA calibration standard (Jones et al., 2011). This showed our laboratory calibration scale to be within the instrumental error (10-20 %, 2 sigma) of this new NOAA standard. In July 2008 we ran our two instruments side by side at Bukit Atur, measuring C 2 Cl 4 , CHBr 3 and CH 2 Br * 2 in ambient air, using the same calibration gas on both GC-ECDs. For this 4-day period we found that the observations of C 2 Cl 4 , CHBr 3 and  20 n/a n/a CH 2 Br * 2 from the two instruments were within 6, 1 and 15 % of each other respectively.

Field deployment
We have been making halocarbon observations in Sabah (Malaysian Borneo) since April 2008. Initially, the instruments were deployed in the Danum Valley as part of the NERC-funded OP3 project (Oxidant and Particle Photochemical Processes above a southeast Asian tropical rainforest; Hewitt et al., 2010). Two µDirac GCs were used in OP3: one was based at the Global Atmospheric Watch (GAW) station at Bukit Atur (Gostlow et al., 2010), and the other made "mobile" measurements at several rainforest locations in Danum Valley and during a 4-day visit to the coast at Kunak (Pyle et al., 2011b). At the end of the OP3 campaign the µDirac GC at the Bukit Atur GAW station was left running and the second "mobile" instrument was moved to a coastal site near Tawau, where it remained for the period of this study. The instrument at Bukit Atur is visited at ∼ 2-week intervals (and on request) by staff from the Malaysian Meteorological Department (MMD) who manage the site. The Tawau instrument was visited at ∼ 1-month intervals and on request by one of our collaborators from Global Satria Sdn. Bhd. (an aquaculture company). Neither instrument was connected to the Internet, so these visits were essential for data recovery and monitoring of instrument performance. Periodic service visits (every 9-12 months) were made to both sites by Cambridge University staff. As a result of this low-intensity monitoring, the performance of the instruments varies over time, depending on the condition of the instrument and its environment. Each site is now described in the following subsections.

Danum Valley rainforest site: the Bukit Atur GAW station
The Danum Valley conservation area covers 438 km 2 and is one of the largest, most important and best-protected expanses of pristine lowland dipterocarp rainforest remaining in southeast Asia. The GAW station located on Bukit Atur (4.980 • N, 117.844 • E, elevation 426ṁ) has been in operation since 2004 and is reasonably remote, being ∼ 50 linear km away from the nearest town of Lahad Datu (Fig. 1). The station is equipped with a range of monitoring instruments and an automatic weather station located on a 5 m roof-top platform. The 100 m tower, which adjoins the laboratory, has air intakes and platforms for sampling equipment at various levels and is the tallest instrumented structure on Borneo island. The ongoing measurement programme at the Bukit Atur GAW station includes surface ozone, greenhouse gases (flask sampling), carbon dioxide (using a Loflo Mk2 analyser), precipitation chemistry, aerosol (load, chemical and physical properties), reactive gases (filterpack sampling), persistent organic pollutants, solar radiation and meteorological parameters. The site is operated by the MMD, and staff visit weekly from their nearest office at Tawau airport. The station itself is located on a hill top ∼ 200 m above the valley floor, and the inlet for the µDirac instrument is attached 12 m up the tower, several metres above the roof of the site building. The instrument is placed in an air-conditioned room in the main GAW building. The topography in the area consists of a series of hills ∼ 400 m above sea level with valley floors down to ∼ 200 m above sea level. One of the largest mountains in the area is Mount Danum at 1093 m above sea level. The rainforest extends to all areas, including mountain tops. Primary unlogged rainforest in the conservation area is surrounded by areas of secondary rainforest (which have been logged in recent years). In terms of site meteorology, a clear diurnal cycle in rainfall is observed at the nearby Danum Valley Field Centre (DVFC) with an afternoon peak around 15:00 local time resulting from diurnal development of convective cells following the midday peak in temperature (Hewitt at al., 2010). Boundary layer stability also varies diurnally from highly stable at night (often leading to dense fog at night) to highly unstable during the day due to the onset of strong convection coupled with low wind speeds (Pearson et al., 2010).

Tawau coastal site: Kampung Batu Payong
The Tawau coastal site is located in the peaceful village of Kampung Batu Payong (4.223 • N, 117.997 • E, elevation 15 m) and is ∼ 10 km from Tawau centre and ∼ 85 km south of the Bukit Atur rainforest station (Fig. 1). Between July 2008 and October 2011 the µDirac instrument here was located in an air-conditioned room at a fish hatchery site maintained by the Global Satria company. There are no seaweed beds in the immediate vicinity of the site. The activity at the site does not involve the use of any seaweeds (which are known to emit bromocarbons) and we do not find any evidence of emissions from the hatchery activities. The air inlet is located ∼ 5 m above ground level at the site (about rooftop height) 40 m away from the shoreline. There is an unobstructed view to the sea extending to more than 180 • of horizon to the south, and the site is situated at the foot of a small group of hills extending to the north, reaching an eleva-tion of 350 m. A low-cost weather station at the site gives an indication of wind speed and direction, temperature and humidity. This reveals a typical diurnal variation in wind due to land/sea breezes expected for coastal sites, though this does not obviously influence the observations reported here.

Models
Two modelling tools are used in this paper to assist in the interpretation of the observations. They are now described in turn in sub-Sects. 4.1 and 4.2.

NAME
The Numerical Atmospheric dispersion Modelling Environment (NAME) is a Lagrangian particle dispersion model developed by the UK Met Office (e.g. Jones et al., 2007) which has been extensively used for analysis of long-term halocarbon data sets (e.g. Manning et al., 2003;Simmonds et al., 2006;Derwent et al., 2007). Abstract particles are moved through the model atmosphere by a combination of 0.5625 • longitude by 0.375 • latitude mean wind fields calculated by the UK Met Office Unified Model (Davies et al., 2005), and a random walk turbulence scheme. NAME can be run backwards in time, to see where the air measured at a particular site may have originated, and forwards to see where air from a particular emission source might go. Each of these capabilities has previously been used to investigate relationships between sources, transport and measurements of pollutants (e.g. Redington and Derwent, 2002;Witham and Manning, 2007).
Here, NAME is run backwards in time to reveal the origin of air arriving at Bukit Atur and Tawau. For each 3 h period thousands of inert particles (carrying an arbitrary mass) are released and travel backwards in time for 12 days. At each 15 min time step the location of each particle is recorded. Then, at the end of the run this information is used to create an "air history map". Here, we only record near-surface particles (lowest 100 m of the model) which indicate where air may have been exposed to surface emissions. Monthly mean air histories are then calculated. The air history maps made for Tawau are almost identical to those from Bukit Atur and so are not shown here.

p-TOMCAT
The global chemistry transport model (CTM) p-TOMCAT is used to analyse the causes of seasonal bromoform variability at Bukit Atur. The basic formulation of p-TOMCAT is described in Cook et al. (2007) and Hamilton et al. (2008). Tracer transport is based on 6-hourly meteorological fields, including winds and temperatures, derived from European Centre for Medium-Range Weather Forecasts operational analysis for the years 2008, 2009 and 2010. Here we use a high-resolution (0.5 • × 0.5 • ) version with bromoform tracers "coloured" according to the region of emission and simple OH oxidation chemistry, described previously in Pyle et al. (2011b) and Warwick et al. (2006). The degradation of bromoform is determined using 3-D fields of pre-calculated hourly OH values taken from a previous p-TOMCAT integration and photolysis frequencies from an integration of the Cambridge 2-D model using cross-section data summarised by Sander et al. (2003). For further details see Pyle et al. (2011b). Here we compare time series of bromoform from the model output with the measurements.

Results and discussion
In this section we present the measurements of C 2 Cl 4 , CHBr 3 and CH 2 Br * 2 from both sites and CH 3 I from Tawau only. We first show that C 2 Cl 4 can be used as a good tracer of air mass, differentiating clearly between air masses with anthropogenic and unpolluted origins. This provides a valuable background for the analysis of the three, primarily biogenic, short-lived compounds CHBr 3 , CH 2 Br * 2 and CH 3 I. In this section we also assess the total organic Br coming from CHBr 3 and CH 2 Br * 2 combined. We discuss the implications of these observations, in particular for the regional emissions of CHBr 3 and CH 2 Br * 2 .

Tetrachloroethene
Tetrachloroethene (C 2 Cl 4 ) is an excellent anthropogenic tracer with few (if any) known natural sources. Its main anthropogenic sources are the textile industry, dry-cleaning applications and in vapour degreasing of metals (Montzka and Reimann, 2011). It currently has a Northern Hemisphere background concentration of 2-5 ppt with a pronounced seasonal variation, likely due to variation in the concentration of its major sink, the OH radical. In the Southern Hemisphere the background concentration is lower (0.5-1 ppt), though still with a pronounced seasonal variation (see http: //agage.eas.gatech.edu/data.htm; Simmonds et al. (2006). C 2 Cl 4 has an estimated lifetime of ∼ 90 days (see Tables 1-4 in Montzka and Reimann, 2011), making it a good tracer for studying long-range transport. At Bukit Atur and Tawau the C 2 Cl 4 dry-air mole fractions ranged typically from 0.3-0.8 ppt in the period May-August 2009 to 1.0-3.7 ppt from November 2008 to March 2009 (Fig. 3a). "Above-baseline" spikes are observed occasionally at Tawau, but for most of the time the data at Bukit Atur and Tawau track each other on monthly and daily timescales. As the two sites are 85 km apart and as both C 2 Cl 4 time series track each other closely, it seems there is little local contribution from Sabah to the background mixing ratio.
For the 4-month period between December 2008 and March 2009 there is a systematic difference in the C 2 Cl 4 observations at the two sites, with the observations at Bukit Atur on average 48 % higher than those at Tawau. It is surprising to find this large systematic difference given that the observations of C 2 Cl 4 were within 6 % of each other when the GCs ran side by side at the same site and with the same calibration gas at Bukit Atur in July 2008 (Sect. 2 last paragraph). The differences are probably due to a relative drift in the composition of the calibration gases used at each site. Any drift in composition is likely to be larger for the decanted calibration gas which was used at Bukit Atur between November 2008 and May 2009, and therefore we have more confidence in the time series from Tawau (see Sect. 2.2). The C 2 Cl 4 observations reported at Bukit Atur between December and March are likely to be an upper estimate of the actual mixing ratios at that time.
Although there is a greater frequency of high (local pollution) spikes at Tawau, these do not impact significantly on the underlying background, which is similar to that at Bukit Atur. The seasonal trend at both sites is clear from box-andwhisker plots (Fig. 4a, d) and the tabulated monthly mean, median and modal values (Table 3a), which show low levels and low variability in the Northern Hemisphere summer months in contrast with higher levels in winter. To further explore these seasonal differences and site-to-site similarities, for selected months we use probability distributions from the C 2 Cl 4 observations, using a size interval of 0.1 ppt (Fig. 5a, e). In January/February 2009 there is a broad distribution of mixing ratios at both sites ranging from 0.85 to 3.75 ppt, tailing towards higher mixing ratios. The Tawau coastal median value in January/February 2009 (seen in Fig. 5a) is lower (1.30 ppt) than inland at Bukit Atur (1.91 ppt), though recall that the C 2 Cl 4 observations at Bukit Atur are likely an upper estimate. In June/July 2009 the distribution at each site has a much narrower range (0.25 to 1.65 ppt) and the median values are similar, though the coastal median (0.45 ppt) is still lower than inland (0.63 ppt). In January/February 2009 the Bukit Atur instrument ran on the decanted calibration air from the NOAA Aculife cylinder AAL-70424, whereas in June/July 2009 the calibration came directly from the Aculife cylinder.
The seasonal changes in C 2 Cl 4 are linked to changes in the prevailing monsoon flow, and the inter-hemispheric gradient typically observed for gases with predominantly Northern Hemisphere anthropogenic sources. Air history maps generated using NAME are used here to illustrate the seasonal changes in transport. From the months of November through to February the air mass history is influenced strongly by air coming from the north and east (Fig. 6a, January 2009). The large urban centres in and around China in this region are expected to contribute to the background level of C 2 Cl 4 arriving at Sabah. From March to May 2009 the air mass origin tends to be influenced more by oceanic air from the western Pacific, though with still some trajectories passing the vicinity of China (Fig. 6b, March 2009). During June, July, August and September 2009 the air mass history indicates that Sabah is influenced strongly from the region to the southeast, characterised by Southern Hemisphere air originating from the Indonesian islands and northern Australia (Fig. 6c, June 2009). As there are no major sources of C 2 Cl 4 across this region and the air masses are not influenced by emission from for example Jakarta, we find (as expected) background mixing ratios to be low. October sees a transition between background air influenced by the Southern Hemisphere and Northern Hemisphere air influenced by China and other large urban centres. It is thus an example of a transition period between the two monsoon regimes (Fig. 6d, October 2009).

Bromoform
Bromoform (CHBr 3 ) is an excellent tracer of marine air and is an important source of organic bromine to the atmosphere (Carpenter and Liss, 2000). An open-ocean, phytoplankton source has been identified (Tokarczyk and Moore, 1994), and macroalgae found in coastal areas are also well known as major sources of bromoform (Moore and Tokarczyk, 1993). Emission fluxes in coastal regions are thought to be higher than in the open ocean (Quack and Wallace, 2003;Yokouchi et al., 2005;Butler et al., 2007), though the relative contribution from coasts and from the open ocean to the total global emission remains uncertain. Recent estimates of the global emissions of CHBr 3 are in the range 430-1400 Gg Br yr −1 (Liang et al., 2010;Pyle et al., 2011b;Ordonez et al., 2012; and Tables 1-8 in Montzka and Reimann, 2011), though much lower estimates of 120 and 200 Gg Br yr −1 are given by Ziska et al. (2013). Although bromoform emissions are dominated by natural sources, there are minor anthropogenic contributions from by-products of drinking-water chlorination in the presence of bromide ions, salt water swimming pools and power plant cooling water (Worton et al., 2006). In the atmosphere, CHBr 3 has a lifetime of 24 days (from Tables 1-4 in Montzka and Reimann, 2011), which is shorter in the tropics (15-21 days), as discussed in Warwick et al. (2006). The Table 3a. Statistical summary of our observations at Bukit Atur and Tawau: monthly and overall mean, mode and median values for the mixing ratios of (a) C 2 Cl 4 , (b) CHBr 3 , (c) CH 2 Br 2 and (d) CH 3 I. Table 3c also shows the R 2 value for the correlation between CH 2 Br * 2 and CHBr 3 for each month; the four strongest correlations are shown in bold; and all correlations give a positive slope except June 2009 at Tawau (shown in italics). The total sample number varies between months and between sites due to differences in chromatogram cycle times and periods of instrument downtime. Modal values are based on a bin width of 0.1 ppt. For CH 2 Br * 2 at Tawau the sample counts are particularly low as we are only able to use about 1 in 10 sample chromatograms for reliable quantification. The overall C 2 Cl 4 distribution at both sites is distinctly bi-modal; the secondary mode is shown in parentheses. n/a means no data available. Statistical summary for C 2 Cl 4 at Bukit Atur and Tawau.

Month
Sample total (n) Mean ( The background CHBr 3 concentration at both Bukit Atur and Tawau is broadly in the range 0.5 to 3 ppt. In contrast to the time series of C 2 Cl 4 , there is no unambiguous seasonal variation in the background CHBr 3 concentration (Fig. 3b). During the periods when we report CHBr 3 mixing ratios from both sites, the background concentrations are sometimes the same within experimental error (e.g. Fig. 5b for January/February 2009), and in some months the coastal background is higher than inland (e.g. Fig. 5f for June/July 2009). These probability distributions also indicate that we do not observe strong seasonality in the background mixing ratio at Tawau. In contrast, there is a suggestion of a seasonal cycle, with a Northern Hemisphere winter peak, in the data collected inland at Bukit Atur (e.g. Fig. 4b).
To emphasise this point, the median mixing ratios for the two sites were similar in January/February 2009 (1.67 ppt at Tawau and 2.04 ppt at Bukit Atur), whereas for June/July 2009 the median value for Tawau (1.68 ppt) was higher than at Bukit Atur (0.95 ppt). Clearly, measurements over many years are needed to ascertain whether the somewhat different behaviour we have observed at the two sites in Borneo is a persistent feature. See Table 3b for the statistical summaries of all the relevant periods.
A major difference between the two sites is that there are frequent periods (spikes) at Tawau when very high levels (> 10 ppt) of bromoform are observed, whereas at Bukit Atur such episodes are rare, and the highest concentration is just 5.5 ppt. This difference in variability is shown clearly in the box-and-whisker plots for Bukit Atur and Tawau ( Fig. 4b and e), where CHBr 3 observations > 10 ppt at Tawau are seen in 12 of the 15 reported months.
Various factors determine the concentrations we observe at these two sites. The sources of CHBr 3 are certainly heterogeneous in space, and so the mere fact that the boundary layer is influenced by different regions in different seasons (Fig. 6) leads to a varying influence of recent emissions and hence varying observed concentrations. This heterogeneity occurs due to both natural spatial variations in the populations of algae in coastal and open-ocean areas, and due to the presence of commercial aquaculture in the region, whose influence on bromoform emissions needs to be assessed. For example, northeast Sabah and the archipelagos between Sabah and the Philippines are major seaweed farming areas whose output has grown quickly over the last 20 years (Phang, 2006). At present it appears that entirely natural emissions of bromoform in the region are significantly larger than those linked to aquaculture (Leedham et al., 2013), but continued rapid growth in the cultivation of many species with commercial potential (Goh and Lee, 2010; Tan  d) e) f) g) Figure 4. Box-and-whisker plots for the halocarbon observations at Bukit Atur for (a) C 2 Cl 4 , (b) CHBr 3 and (c) CH 2 Br * 2 and at Tawau for (d) C 2 Cl 4 , (e) CHBr 3 , (f) CH 2 Br * 2 and (g) CH 3 I. In each panel, the coloured boxes show the data falling between the 25th and 75th percentiles, with the median value shown as a black horizontal line inside each box. The vertical black lines above and below each box indicate the 95th and 5th percentiles respectively. Outliers (values beyond the 95th and 5th percentiles) are shown as grey open circles. The mixing ratio (y axis) scale is logarithmic in order to show the extreme high values, such as for CHBr 3 at Tawau (e). The filled circles in the smaller panel above each box-and-whisker plot show the number of days sampled each month, and labels indicate the number of samples used to generate each box and whisker. The total sample count is different at each site due mainly to different periods of instrument downtime and differences in the sampling frequency at each site. this situation (Phang et al., 2010). Further, the strength of individual sources is likely to vary over time. In the case of seaweeds, it seems likely that their halocarbon emissions will depend on environmental factors such as temperature, rainfall, wind, sunlight and, if applicable, the phase of the cropping cycle (Seh-Lin Keng et al., 2013;Leedham et al., 2013, and references therein). Given this complexity, without extensive biological field surveys it is difficult to explain much of the variability in detail.
However, we are able to examine how well the CTM p-TOMCAT is able to simulate the long-term statistical properties of observed bromoform mixing ratios. The model configuration is described in Sect. 4.2, and the bromoform emissions are those developed by Pyle et al. (2011b) as an update to Warwick et al. (2006). These emissions provide the best match with our earlier observations collected at Bukit Atur during OP3 (Pyle et al., 2011a, b). The observed time series for CHBr 3 is consistent with the p-TOMCAT model output using the revised emission estimate as shown for Tawau in Fig. 7a. The model is unable to capture the very high values seen in the observations, which is to be expected given the model resolution and the uniform distribution of emissions around the coastlines in the model. for instance the high tail is not wholly visible for CHBr 3 at Tawau (b, f), though all data are included in the box-and-whisker plots (Fig. 4).  Maps for all months were produced but not shown. Each map is compiled from 11 000 particles released on an hourly basis by the NAME dispersion model using UK Met Office analysed wind fields at ∼ 50 km horizontal resolution. They show the particle density in the lowest 100 m averaged over the preceding 12 days. From November to February the air mass origin is mainly from the Northern Hemisphere, often passing close to China (a). Between March and May the air mass origin gradually changes towards western Pacific equatorial latitudes, away from large urban regions (b). From June to September the air mass history is influenced from the region to the southeast, characterised by Southern Hemisphere air, with sparse anthropogenic emissions (c). October sees a transition back to Northern Hemisphere air masses (d).
the observations of 1.73 ppt is just slightly higher than that for the model output (1.65 ppt).

Dibromomethane
It is thought that the marine sources of dibromomethane (CH 2 Br 2 ) are similar to those of CHBr 3 , as shown by the strong correlations observed between their atmospheric concentrations (Yokouchi et al., 2005;Carpenter et al., 2009). Emission fluxes are probably higher in coastal regions than over the open ocean , but the relative effect on coastal atmospheric measurements is less marked than for CHBr 3 due to a longer lifetime for CH 2 Br 2 of ∼ 120 days (Tables 1-4 in Montzka and Reimann, 2011). The CH 2 Br 2 lifetime is expected to be slightly lower than 120 days in the tropics due to the higher abundance of the OH radical.
Open-ocean sources are also more important globally than coastal sources. Recent estimates of the global CH 2 Br 2 emis-sions are 50-70 Gg Br yr −1 (Liang et al., 2010;Ordonez et al., 2012). The mixing ratios of CH 2 Br * 2 measured at Bukit Atur and Tawau are shown in Fig. 3c. The observed baseline values of between 1 and 2 ppt for CH 2 Br * 2 at the two sites are similar, as reflected in the median values in Table 3. Previous work suggests that 10-30 % of the CH 2 Br * 2 signal may in fact be CH 2 BrCl 2 . During June and July 2009, Mohd Nadzir et al. (2014) collected air samples on a ship cruise around the coast of the Sulu-Sulawesi seas close to Tawau, and the average CH 2 Br 2 mixing ratio was 1.04 ppt compared to 0.3 ppt for CHBrCl 2 . The authors also found that the two compounds tracked each other closely through the entire cruise. In another study, Sala et al. (2014) reported average boundary layer mixing ratios of CH 2 Br 2 and CHBrCl 2 of 1.19 and 0.34 ppt respectively from an aircraft campaign in the western Pacific region during November/December 2011. Both CHBrCl 2 and CH 2 Br 2 are known to have a common Atmos. Chem. Phys., 14, 8369-8388 macroalgal source as found by Seh-Lin Keng et al. (2013) and Leedham et al. (2013) in studies of tropical seaweeds from the Malaysian coast. These studies also found that production rates of CH 2 Br 2 were generally (though not always) much higher than for CHBrCl 2 . In the atmosphere CHBrCl 2 has a lifetime of 78 days compared to 123 days for CH 2 Br 2 (Montzka and Reimann, 2011). Thus we would not expect to see any significant difference in the measured CHBrCl 2 : CH 2 Br 2 ratio between Tawau and Bukit Atur due to differential processing, given the short air mass transit time between the sites of typically 1 day. The variability of CH 2 Br * 2 at Tawau is much less than for CHBr 3 . What may be an important factor here in variability differences between the two sites is the extent to which boundary layer air at Bukit Atur has been diluted with free tropospheric air with a lower ratio of CHBr 2 Cl to CH 2 Br 2 , given that the site is 426 m above sea level. Statistically, the observations at Bukit Atur and Tawau are similar to each other as shown by the box-andwhisker plots, although the 2 months at Tawau with distinctly low median values (March/April 2009) are not reflected in the Bukit Atur observations (Fig. 4c and f). There is a definite hint of some seasonality in the CH 2 Br * 2 measurements ( Fig. 5c and g), though multiyear data are required to determine whether there is a seasonal cycle. Median values in January/February 2009 are almost the same at both sites (1.20 and 1.28 ppt at Bukit Atur and Tawau respectively), though in June/July 2009 the median value at Tawau (1.33 ppt) is distinctly higher than at Bukit Atur (1.00 ppt).
We find that in all reported months at Bukit Atur there is always a positive correlation between the observations of CH 2 Br * 2 and CHBr 3 , suggesting a common source (Table 3c). Mohd Nadzir et al. (2014) also found a strong positive correlation between CHBr 3 and CH 2 Br 2 in whole-air flask samples taken during a 2009 cruise in the South China and Sulu-Sulawesi seas (close to the Tawau site) and Strait of Malacca. The strength of the correlation at Bukit Atur and Tawau is highly variable between months. The strongest correlation at Bukit Atur is found in November 2008 with an R 2 value of 0.86. At Tawau the correlation between CH 2 Br * 2 and CHBr 3 is generally much weaker than at Bukit Atur, though still positive, with the exception of June 2009. Only March and September 2009 show strong positive correlations at Tawau. As we are only able to report a CH 2 Br * 2 measurement in about 1 in every 10 samples at Tawau, there are fewer data available for the correlation of CH 2 Br * 2 and CHBr 3 .

Methyl iodide
Methyl iodide (CH 3 I) is ubiquitous in the marine boundary layer, with typically reported concentrations of 0.1-5 ppt (e.g., Lovelock et al., 1973;Rasmussen et al., 1982;Yokouchi et al., 2001Yokouchi et al., , 2008Yokouchi et al., , 2011Yokouchi et al., , 2012. There are several known oceanic sources of CH 3 I, including phytoplankton (Moore and Tokarczyrk, 1993;Manley and dela Cuesta, 1997), macroalgae (Manley and Dastoor, 1988;Schall et al., 1994;Giese et al. 1999), methylation of iodine by bacteria (Amachi et al., 2001), picoplankton (Smythe-Wright et al., 2006;Brownell et al., 2010), photochemical reactions (Moore and Zafiriou, 1994) and as a by-product of the ozonolysis of dissolved organic matter (Martino et al., 2009). Methyl iodide has an atmospheric lifetime of 7 days (Tables 1-4 in Montzka and Reimann, 2011) and has the shortest lifetime of the compounds reported here. The most extensive, continuous measurements of CH 3 I reported to date in the tropics or subtropics were made hourly   (Yokouchi et al., 2012). Summer maxima were seen at all the four sites apart from Alert, where a wintertime maximum (attributed to transport) was seen. The discrepancy between the continuous and semi-monthly measurements at Hateruma Island presumably results from lowfrequency sampling of the highly variable mixing ratio in the late summer. No significant seasonal cycle is seen in the baseline of the higher-resolution CH 3 I time series. Observations of CH 3 I in the tropical boundary layer have otherwise been limited to measurements of whole-air samples made at specific sites or on research cruises (see summaries in Table 1 in Yokouchi et al., 2008, and Table 2 in . Our CH 3 I record for Borneo (Tawau only) is therefore a unique source of information in the tropics. The measurements for 2009 are shown in Figs. 3d, 4g and 5d and h. The CH 3 I peak is small and hard to separate (using our current column configuration) from the larger peaks that elute at a similar time. The µDirac at Bukit Atur did not achieve satis- factory chromatographic separation for reliable CH 3 I quantification, and these data are not reported here. The CH 3 I mixing ratios at Tawau (∼ 8400 in total) vary from below 0.1 up to 11 ppt, with the large majority of points towards the lower end of the range. July sees the lowest monthly mean (0.34 ppt), and the highest monthly mean (2.50 ppt) is in September. The highest values and the greatest variability are observed in September and October. Back trajectories indicate that air containing high amounts of CH 3 I had previously passed over large biomass burning events in southern Borneo (as observed by MODIS). Emissions of CH 3 I in biomass burning have been reported previously (Andreae et al., 1996;Mead et al., 2008). However while biomass burning emissions may be significant locally at some times of year, it is not the case globally (Bell et al., 2002;Butler et al., 2007). Interestingly some of the lowest observed values (< 0.1 ppt) of CH 3 I occur when the air has previously been in the terrestrial boundary layer, indicating that the tropical rainforests were not normally a source of CH 3 I. The seasonal cycle in the baseline (i.e. if the high values seen in the late summer/early autumn are excluded) at Tawau is weak, similar to that seen at Hateruma Island. A longer record is needed to identify whether one exists.

Total bromine
Here we present and discuss our observations of bromine defined by [CHBr 3 × 3] + [CH 2 Br * 2 × 2] as an approximate measure of total organic bromine from the two sites. For a 2week period of the OP3 campaign (5-12 July 2008) we ran our Bukit Atur instrument alongside a GC-MS system from UEA which measured CHBr 3 , CH 2 Br 2 , CHBr 2 Cl, CHBrCl 2 and CH 2 BrCl (the five main naturally occurring bromine source gases). The GC-MS observations showed that CHBr 3 and CH 2 Br 2 accounted for 93.5 % of the total Br coming from these five bromine source gases (Pyle et al., 2011a). Total bromine as [CHBr 3 × 3] + [CH 2 Br * 2 × 2] from our measurements at Tawau over the period November 2008 to January 2010 is shown as a time series in Fig. 8a. The black line through the data is the running average total bromine (40 point, 50 pass) to highlight average variability. Bromine from these two compounds (accounting for > 90 % of the total organic Br) averages 9.7 ppt over the period of observations and ranges from 3.7 (June 2009) to 189.5 ppt (December 2008). The probability distribution function for total observed bromine as [CHBr 3 ] + [CH 2 Br * 2 ] at Bukit Atur is compared with that for Tawau in Fig. 8b. The distributions are similar (median values of 6.79 and 7.97 ppt at Bukit Atur and Tawau respectively) and Bukit Atur has a narrower interquartile range than Tawau (2.06 ppt compared to 2.80 ppt). The extended tail of very high total bromine values (> 20 ppt) is evident at Tawau and completely absent from the Bukit Atur observations.
As discussed earlier, we are unable to report separate observations of CHBrCl 2 and CH 2 Br 2 and are obliged to report combined observations (CH 2 Br * 2 ). Separate observational studies have indicated that CHBrCl 2 mixing ratios are typically 10 to 30 % of those for CH 2 Br 2 (Pyle et al., 2011a;Mohd Nadzir et al., 2014;Sala et al., 2014). The total bromine observations at each site are relatively insensitive to this range of possible variation. If we apply this range to the CH 2 Br * 2 observations, then the impact on the median bromine observations is to reduce total Br at Bukit Atur by between 0.1 and 0.3 ppt and at Tawau by between 0.1 and 0.4 ppt.
The  Figure 8. Estimated total Br (a) from observed CHBr 3 and CH 2 Br * 2 as a time series at Tawau and (b) as probability distribution functions at Bukit Atur and Tawau. Bromine from CH 2 Br * 2 is shown by the light-red dots (a) and from CHBr 3 by the dark-red dots, whilst Br summed from CHBr 3 and CH 2 Br * 2 is shown by the black dots and the heavily smoothed black line respectively. The probability distributions for Bukit Atur (blue circles with blue dotted line) and Tawau (red circles with red dotted line) for the 10-month period from November 2008 to August 2009 are similar (b). The total Br median values shown as blue (Bukit Atur) and red (Tawau) vertical lines agree with each other to within 1.2 ppt. Occasionally very high total Br observed at Tawau is clearly seen (right of the x axis break); events higher than 20 ppt were never observed at Bukit Atur.
that an inland site can be used effectively to make regional observations suited to model comparison. In fact, as the model we used here was unable to capture the large variability in the observations, presumably due to insufficient information on the spatial and temporal inhomogeneity of the emissions seen at the coast and to model resolution, we suggest that, for evaluating models, an inland site is preferable 8384 A. D. Robinson et al.: Long-term halocarbon observations from a coastal and an inland site in Sabah to a coastal site. We also calculate total modelled bromine coming from [CHBr 3 × 3] + [CH 2 Br 2 × 2] in p-TOMCAT at Tawau and find for the period of our observations a total bromine median value of 6.56 ppt, somewhat lower than the median value of 7.97 ppt from the measurements.
Our data provide a longer-term context for other reports from short campaign-type observations. There are month-tomonth variations in total Br in our data that would be missed by shorter deployments. Overall, our mean value of 9.7 ppt from [CHBr 3 ] + [CH 2 Br * 2 ] seems quite consistent with other work in the region. There have been somewhat conflicting reports of the fraction of Br in CHBr 3 and CH 2 Br 2 compared to the minor very short lived substances (VSLS). Montzka and Reimann (2011) suggest a boundary layer median value of 8.4 ppt Br from VSLS, with 83 % from CHBr 3 (1.6 ppt × 3) and CH 2 Br 2 (1.1 ppt × 2). Based on data obtained in July 2008 from the OP3 campaign, Pyle et al. (2011a) report 7 ppt (5-10 ppt) Br from VSLS, with over 90 % of this present as CHBr 3 , CH 2 Br 2 or CHBrCl 2 . Mohd Nadzir et al. (2014) report a mean of 8.9 ppt Br from all VSLS (range 5.2 to 21.4 ppt in individual samples), with ∼ 90 % from CHBr 3 and CH 2 Br 2 during the Prime Expedition Scientific Cruise during June and July 2009.

Summary
Until recently, few measurements of short-lived halogenated compounds had been made in southeast Asia. These compounds have significant atmospheric importance, and previous studies have speculated that southeast Asia represents an unusually strong source of the naturally occurring species (e.g. CHBr 3 ). Here we present more than 1 year of observations (November 2008 to January 2010) of key short-lived compounds from a coastal (Tawau) and an inland (Bukit Atur) site in this region, which together provide valuable data to the atmospheric science community. The data have been placed on the Halocarbons in the Ocean and Atmosphere (HalOcAt) database for community access (https://halocat. geomar.de/). The data allow seasonal cycles, emissions and the role of local and long-range transport to be studied.
By combining the observations with air mass history information generated by NAME we explain the strong seasonal cycle seen in the anthropogenic tracer (C 2 Cl 4 ). From December to February, northerly air masses lead to higher and more variable mixing ratios, ranging from 0.8 to 3.7 ppt. In contrast, from May to August the observed air masses originate from regions to the south (typical mixing ratio 0.5 ppt). The overall median values for C 2 Cl 4 are 1.33 and 0.99 ppt at Bukit Atur and Tawau respectively.
The naturally occurring brominated tracers CHBr 3 and CH 2 Br * 2 do not show unambiguous seasonal variation, though there is a slight hint of a seasonal cycle in CH 2 Br * 2 . Both compounds nevertheless show strong variability over shorter timescales. At the coast, the CHBr 3 observations oc-casionally peak at over 100 ppt compared with the overall median level of 1.7 ppt. At Bukit Atur, the overall inland median level for CHBr 3 is 1.6 ppt, and the occasional peaks are much lower than at the coast, with maximum observations of just over 5 ppt. Coastal CH 2 Br * 2 mixing ratios occasionally rise to over 7 times the median level of 1.4 ppt, whereas inland the maximum level observed is a little over 2 ppt. The strong coastal variability of both these compounds is likely related to the strongly heterogeneous nature of their marine sources, which are believed to be a combination of coastal macroalgae and both coastal and oceanic microalgae. Accordingly, the measurements from coastal Tawau are more variable than at Bukit Atur, being made closer to these sources.
The observations of CHBr 3 and CH 2 Br * 2 also provide a valuable longer-term estimate of the contribution to total Br in the boundary layer from the main naturally occurring short-lived brominated compounds. At the coast we observe an overall median value of Br from [CHBr 3 × 3] + [CH 2 Br * 2 × 2] of 8.0 ppt, and inland the overall median value is 6.8 ppt. These values are reasonably consistent with other recent estimates of this quantity based on much shorter measurement periods. Our data provide evidence that southeast Asia may not in fact be a region of enhanced brominated VSLS emissions into the boundary layer in relation to other tropical locations where similar measurements have been made.
Our coastal observations of CH 3 I show high variability as expected for a short-lived tracer, with values from less than 0.1 to over 11 ppt. The overall median value is 0.6 ppt.
The highest values are observed in September and October 2009 and are attributed to the influence of biomass burning in southern Borneo that year. In other times of the year, air that passed over land showed some of the lowest values observed, indicating that the tropical rainforests in Borneo are not a source of CH 3 I.
To date, modelling work has focussed on understanding our CHBr 3 data. The CHBr 3 observations presented here were recently used by Ashfold et al. (2014), along with NAME trajectories and an inversion method, to estimate regional-scale CHBr 3 emissions. When extrapolated across the global tropics, the magnitude of their estimated emissions was somewhat lower than most other recent estimates, which have tended to assume strong emissions in southeast Asia. This is consistent with the finding here that observed CHBr 3 mixing ratios in Borneo (and indeed CH 2 Br * 2 mixing ratios and total Br estimates) are in fact similar to those measured in other parts of the world. Our measurements also provide a means of testing the ability of global models like p-TOMCAT to reproduce boundary layer time series of short-lived compounds in tropical regions. Although p-TOMCAT was unable to reproduce the same magnitude of variability seen in the Borneo CHBr 3 observations, it was able to capture background concentrations using the bromoform emission distribution described in Pyle et al. (2011b). This emission distribution is an updated version of the distribution described in Warwick et al. (2006), containing substantial reductions in southeast Asian emissions relative to the original scenario.
We continue to make halocarbon measurements in Sabah and have upgraded to the newer version of µDirac with a 20 m long column and better resolution of species including CH 2 Br 2 . In addition, we have recently begun measuring the same group of compounds using upgraded versions of the instruments used here at three other sites in the southeast Asia region: at Bachok on the Malaysian Peninsula, Kelantan state; at Gunn Point near Darwin, Australia; and a coastal site near Taipei, Taiwan. These sites significantly extend the geographic coverage of our measurements.