TGM continuous record at the Chacaltaya Station (5240 m) in Bolivia

Abstract. High-quality data of atmospheric mercury (Hg) is rare for South America, especially for its tropical part. In consequence, mercury dynamics are still highly uncertain in this region, a significant deficiency, as South America appears to play a major role in the global budget of this toxic pollutant. To address this issue, we performed nearly two years (July 2014–February 2016) of continuous high resolution total gaseous mercury (TGM) measurements at the Chacaltaya (CHC) mountain site in the Bolivian Andes, which is subject to a diverse mix of air masses coming predominantly from the Altiplano and the Amazon rainforest. For the first eleven months of measurements, we obtained a mean TGM concentration of 0.89±0.01 ng m−3, in good agreement with the sparse amount of data available from the continent. For the remaining nine months, we obtained a significantly higher TGM concentration of 1.34±0.01 ng m−3, a difference which we tentatively attribute to the strong El Niño event of 2015–2016. Based on HYSPLIT back-trajectories and clustering techniques, we show that lower mean TGM concentrations were linked to either westerly Altiplanic air masses or those originating from the lowlands to the south-east of CHC. Elevated TGM concentrations were related to northerly air masses of Amazonian or southerly air masses of Altiplanic origin, the former possibly linked to artisanal and small scale gold mining (ASGM), while the latter might be explained by volcanic activity. We observed a marked seasonal pattern, with low TGM concentrations in the dry season (austral winter), rising concentrations during biomass burning (BB) season, and highest concentrations at the beginning of the wet season (austral summer). With the help of simultaneously sampled equivalent black carbon (eBC) and carbon monoxide (CO) data, we use the clearly BB influenced signal during BB season (August to October) to derive a mean TGM/CO emission ratio of (2.3±0.6)·10−7 ppbvTGM ppbvCO−1, which could be used to constrain South American BB emissions. Through the link with in-situ measured CO2 and remotely sensed solar-induced fluorescence (SIF) as proxies for vegetation activity, we detect signs of a vegetation sink effect in Amazonian air masses and derive a best guess TGM/CO2 uptake ratio of 0.058±0.017 (ng m−3)TGM ppmCO2−1. Finally, significantly higher Hg concentrations in western Altiplanic air masses during the wet season as compared to the dry season point towards the modulation of atmospheric Hg by the Eastern Pacific Ocean.


activity, we detect signs of a vegetation sink effect in Amazonian air masses and derive a "best guess" TGM/CO2 48 uptake ratio of 0.058 ± 0.017 ( −3 )

Air mass origin at the regional scale
Bolivia and Paraguay, but also the Pantanal wetland at the eastern frontier to Brazil.

281
• Cluster C4: Eastern edge of the Altiplano. Air masses passing to the east of the Titicaca lake.

284
• Cluster C6: Cloud forest at the North-eastern edge of the Cordillera Real.  certain degree of statistical significance, we excluded those grid cell means based on less than 10 independent 325 data points on TGM concentration (n < 10).

326
To account for growing trajectory uncertainty with increasing distance to the receptor site (CHC), avoid the 327 misinterpretation of a pollution map as a satellite image, and allow for the easy visual comparison between 328 pollution maps and air mass clusters, we used the exact same CHC centered log-polar grid as deployed in height of 1000 m a.g.l). This means essentially that only trajectories passing at low altitudes over a grid cell have 333 an influence on the TGM average calculated for the cell.

336
A summary of the monthly averaged TGM concentrations is presented in Fig. 3a. The data shows an overall rising 337 trend during the measurement period. As this trend exhibits a striking similarity to the evolution of the ONI index 338 ( Fig. 3c), we suggest an important ENSO influence on TGM measured at CHC. This will be discussed in detail 339 in an upcoming publication. In the present paper, we labelled the last nine months of our measurement period

343
We obtained a mean TGM concentration of 0.89 ± 0.01 −3 for NC and a significantly higher mean of 344 1.34 ± 0.01 −3 for EC ( < 2.2 · 10 −16 , Mann-Whitney test). For both NC and EC, we can observe a 345 similar seasonal pattern, with low TGM concentrations during the dry season, rising TGM concentrations during

354
Horizontal blue lines show the total median of the respective period.

357
Given that the metropolitan area of La Paz/El Alto is located less than 20 km downhill of the measurement site, and eBC concentrations (Fig. 4d, Fig. 4e, Fig.4f). During nights around that traditional festivity, numerous  The evident seasonal pattern in transportation pathways towards CHC is visualized in Fig. 5. The most important 387 air mass clusters during NC, measured by mean relative influence and appearance as the dominant cluster, were time of the year (mean relative influence < 10%).

499
Finally, BB season pollution signatures generally showed very high CO and highly variable eBC concentrations.

500
Within those signatures, TGM concentrations clearly increased with rising CO concentrations but did not depend 501 strongly on eBC loadings, even though they tended to be lower in the case of very low eBC (example: H1, H3, 502 G3, F3. Exception: H2).

503
Considering that atmospheric lifetime is much shorter for BC than for CO (days -weeks for BC, Cape et al. Paz/El Alto, a hotspot for BC (Wiedensohler et al., 2018), is quite close to the station (< 20 km) and transport 511 time is therefore usually less than a few hours (compare to Fig. C1 in Appendix C), we can exclude urban 512 influences as contributors to these pollution signatures and assign them to BB.

514
As median TGM concentrations were significantly higher in those pollution signatures occurring almost 515 exclusively in the BB season (over 85% of the time, Fig. 7a), compared to median NC concentrations 516 (0.93 −3 vs. 0.85 −3 , = 1.7 · 10 −11 , Mann-Whitney test), we can conclude that there is an 517 important influence of regional and continental BB on atmospheric mercury concentrations in the Bolivian Andes.

518
This occurs only during a few months of the year (August -October) and it is mostly constrained to northern-