Investigating the impact of aeolian deposition to the Southern Ocean using dissolved aluminium concentrations

Atmospheric dust deposition (aeolian deposition) is thought to be an important source of biolimiting trace elements to the Southern Ocean (SO). Aluminium (Al) has been established as a proxy for dust deposition due to its limited role biologically, short residence times in the surface ocean and supply mechanisms to the open ocean that are typically dominated by atmospheric delivery of dust. Prior to this study, Al distributions within Australian sector of the SO had not been investigated thoroughly and atmospheric dust deposition was poorly constrained by field observations. This PhD project presents the first dissolved Al (dAl) observations in the Australian sector of the SO. These were used: to improve the understanding of Al biogeochemical cycling in the SO; as inputs to the model for the Measurement of Aluminium for Dust Calculation in Oceanic Waters (MADCOW) (Measures and Brown, 1996) to estimate aeolian deposition; to compare and assess the performance of model estimates of surface ocean dAl concentrations in the SO by Han et al. (2008) and van Hulten et al. (2012) to in situ observations; and to compare estimates of dust deposition from this study to those from published atmospheric models. To achieve this, a novel method for the determination of sub-nM dAl in seawater was developed (RP-HPLC with fluorescence detection of the Al‚-lumogallion complex (Remenyi et al., 2011, 2012)). Total required sample volume (including rinses) was 1.5 mL. Analysis time was 2.7 min per injection. Limit of detection was 0.13¬±0.05 nM, with precision of 2.7% at 1.06 nM. Agreement with SAFe reference samples was within 4.6¬±4.6% (n = 13). It provides improvements in sample efficiency, operational robustness and analysis times relative to popular shipboard techniques, especially when applied to samples where dAl approaches the limit of detection. More than 440 samples were then analysed from two voyages (GIPY2, SAZ-Sense and GIPY6, SR3 repeat transect). dAl was depleted throughout the region from 0.05 to 7.5 nM, with most values <1 nM. This places it at the low end of the global range (0.1‚-180 nM). Maximum dAl values were within the proximity of the East Australia Current (EAC). Minimum dAl values were in the Subantarctic Zone (SAZ), coinciding with the region with a very deep winter mixed layer (ML). Trends were latitudinal rather than following watermasses, probably due to the distance from landmasses coupled with the strong eastward flow of the Antarctic Circumpolar Current. Aeolian deposition of soluble Al was low and as such lateral advection of waters that are relatively enriched in dAl appear to dominate the biogeochemistry of the region. Evidence of sedimentary entrainment from the Australian continental shelf and the Kerguelen Plateau was observed. Satellite observations indicate bushfire (wild fire) ash/smoke may be an important source of dAl East of Tasmania, but appropriate field observations were not available (eg. black carbon). Modelling by Han et al. (2008) (DEAD-MADCOW) overestimated surface dAl concentrations in the SO. This was probably because it overestimated the solubility of Al delivered in dust and underestimated dilution (MLD) and scavenging rates. Modelling by van Hulten et al. (2012) (NEMO-PISCES) was much more sophisticated and consequently had much better agreement with observations. Agreement between the NEMO-PISCES model and the observed data is excellent south of the Polar Front, where the model assumptions and mechanisms are appropriate (where van Hulten et al. (2012) use biogenic silica concentrations to estimate scavenging). However, differences between observations and model estimates are much greater in the SAZ and surrounding Tasmania. In the EMLDZ of the SAZ, the NEMO-PISCES model overestimated dAl, probably because the dilution factor applied in their model is too low. However around Tasmania, the model substantially underestimates dAl. This could be due to either an underestimate of sources, or an overestimate of sinks. The biogenic silica field used to estimate scavenging in the NEMO-PISCES model is ~5- fold greater than observations in the SAZ and the STZ (Rosenberg, 2007, 2008). The associated scavenging rate will also be too large (interestingly this does not correct the overestimated values observed in the EMLDZ of the SAZ). It is likely that the source mechanisms in the NEMO-PISCES model underestimate entrainment within the EAC, and almost certainly do not account for supply from bushfires. Dust deposition rate estimates using the MADCOW model (Measures and Brown, 1996) without adjusting the original parameters (intended for use in the mid-Atlantic) were reasonable, and in relatively good agreement with both DEAD (Zender et al., 2003) and INCA (Seze et al., 1991; Whitehead et al., 1998; Schulz et al., 2009) model estimates. However, application of the model with mixed layer depths (M), scavenging rates (Sc) and solubility rates (S) that were more appropriate for the SO resulted in extremely high dust deposition rate estimates and these were considered unreasonable. Investigation revealed the lack of variance of dAl observations with depth and latitude along the SR3 transect, coupled with the physical oceanographic conditions in the Australia sector of the SO invalidate many of the assumptions required by the MADCOW model. Therefore, dAl concentrations in the SO cannot be used as inputs to the MADCOW model to quantify dust deposition rates into the SO. Given the lack of quality satellite observation of dust deposition over the SO and the infrequent collection of aerosol concentrations in the SO, there are currently no reliable methods for estimating dust deposition to the SO. The best estimates of trace element availability in the SO is from direct measurement of water samples. The key finding from this thesis is: aeolian deposition of dust to the open ocean in the Australian sector of the SO is very limited, so sources of bio-limiting trace elements to surface waters are likely to be from lateral advection or from upwelling depending on the region.