Table of Contents

• Coverage
• Dataset Description
  • Methods & Sampling
  • Data Processing Description
  • BCO-DMO Processing Description
• Data Files
• Supplemental Files
• Related Publications
• Parameters
• Instruments
• Project Information
• Funding

Sampling approaches:
Alaskan glacial dust samples were collected in a variety of ways from two different locations, during multiple years. Bulk dust was collected using a Thermo Partisol Plus 2025 on Middleton Island, Alaska (59.42144 °N, 146.3493 °W) continuously during two years from 2011-2012, as described in Schroth et al, 2017. The sampling location, at a high point in the southwestern third of the ~50-meter (m) high island, was selected to be close to a source of needed electricity but far from frequently traveled roads and other local sources of aerosol (e.g. diesel generators at the northeastern end). Size-fractionated samples were also collected on Middleton Island (same location) during a large glacial dust event from 10-12 November 2019 using a Tisch Volumetric Flow Controlled (VFC) high volume sampler (Tisch Environmental, TE-5170V- BL) outfitted with a Cascade impactor. The six size fractions collected ranged from <0.49 micrometers (um) to >7.2 um in diameter. This sampler technology is discussed in greater detail in Morton et al, 2013. Samples were filtered with Whatman 41 (W41) cellulose fiber filters. Filters were acid-washed using environmental grade 0.5 M HCl and very thoroughly rinsed with milli-Q water, then dried, in a HEPA filtered laminar flow hood.

Finally, one large-volume sample was generated by collecting dust on July 29, 2011, from the shelves of a shed in the Copper River delta (close to Million Dollar Bridge, at 60.67454 °N, 144.74811 °W) that was in the path of the Copper River-derived dust plume and had a hole in its roof, which served as the sample collection device. This large dust sample probably integrated over a few years prior to collection. The validity of this sample, despite a very unconventional sampling approach, is confirmed by the striking similarity of its rare earth element (REE) signature to the REE signature of the other samples collected from Middleton Island (especially to the >7.2 um diameter sample)).

Digestion methods:
Samples of a few milligrams (mg) of dust (typical for the largest size fractions; sometimes less) were weighed after equilibration in lab air overnight in a laminar flow hood to minimize moisture content fluctuations. Moisture content of the dust has the potential to alter mass estimates substantially, hence this step was essential. Total particle digestions were carried out in Savillex 15-milliliter (mL) Teflon vials on a hotplate using a three-step digestion, at 140-150 degrees Celsius (°C), patterned after Morton et al (2013) using: 1) Optima concentrated HNO3; 2) a 4:1 mixture of Optima concentrated HNO3 and Optima concentrated HF; 3) Optima concentrated HNO3, followed in each case by evaporation to dryness. Samples were then redissolved in 4 M HNO3 at 90°C for two hours. Digestions and evaporation steps were carried out within a polycarbonate enclosure, with HEPA-filtered air intake, on top of clean polyethylene sheet, within an exhausting fume hood.

Analyses:
Concentrations of the REE and 232Th were determined on a Thermofisher iCAP inductively coupled plasma mass spectrometer (ICP-MS) in KED mode, with He as a collision cell gas, adapted from the approach of Trommetter et al (2020). Concentrations were determined from standard curves using a REE ICP-MS standard from High-Purity Standards (that also contained 232Th). Three internal standards (Ge, In, and Bi) were added to both samples and standards, to correct for short-term variability in the instrument response and to evaluate stability of mass response during the ICP-MS run. Concentration estimates for the REE and 232Th were blank-corrected using full-process blanks that included filters deployed during times when there was no known dust deposition. Most of the full-process blank concentrations were 100 times or more smaller than the concentrations of our lowest standard (with the exception of Ce, the concentration of which was ~7 times smaller than our lowest standard. This means that our blank concentrations were very low but also not quantified extremely accurately. Our best estimates are that the full-process blanks, including filters, ranged from 0.02 picograms per square centimeter (pg cm-2) for Eu, Tb, and Ho, to 2 pg cm-2 for Ce. These blank concentrations were in all cases 40 times or more smaller than our lowest REE sample concentration for the <0.49 um size fraction with the smallest amount of dust, and ~3 orders of magnitude smaller than the signal of the largest samples. Hence, the blanks did not impact our REE concentration estimates significantly.

Solid reference materials were analyzed to evaluate the completeness of the particle digestion process and the accuracy of the standard solution calibrations. These included PACS-3 and a large sample of Columbia River Basalt (BCR-UW) collected, ground, and homogenized by Prof. Bruce Nelson of UW Earth and Space Sciences, from the same location as BCR-1 and BCR-2. In addition, Arizona test dust and a large-volume dust sample from the Copper River delta (see sampling methods discussion) were analyzed multiple times as additional constraints on reproducibility. Our REE concentration estimates agree within roughly four percent of the published concentrations for the reference materials PACS-3 and BCR-UW, while 232Th concentrations agree within ~5%. We used BCR-UW as a reference material because the USGS Geochemical Reference Materials lab was not able to provide us with any BCR-2 upon request, and we assume that this sample is of the same composition as BCR-2, as is suggested by the REE concentrations.

Double normalized REE data processing:
REE concentrations were double normalized as follows. They were normalized to the average Post-Archean Australian Shale (PAAS) REE concentration (McLennan, 1989), following recent practice (Grenier et al, 2018; Zhang et al, 2008; Friend et al, 2008). Such normalization generates smooth plots of the REE because it helps to reduce the effect whereby even atomic-numbered REE are more abundant than odd atomic-numbered REE (see Grenier et al, 2018). Concentrations were then normalized again to the mean PAAS-normalized REE concentration of each sample, similar to the approach Serno et al (2014) used normalizing to average Upper Continental Crust. See Serno et al (2014) for more detail on double normalization. This normalization approach differs slightly from the Upper Continental Crust normalization used by Serno et al (2014), although there is a very minor impact on the REE patterns (Garcia-Solsona et al, 2014, Appx. A.). Another reason we used this PAAS normalization approach is that it has been used in recent publications where the europium anomaly, Eu/Eu*, is estimated (e.g. Friend et al, 2008; Zhang et al, 2008; Grenier et al, 2018) and is thus consistent with recent practice in the oceanographic community.

The double normalized data are in the attached Supplemental File, "925359_v1_ree_th232_normalized_copperriverdust.csv".

The raw REE concentrations were double normalized, and REE data were saved in this format, as well, as follows. Concentrations were normalized to the average Post-Archean Australian Shale (PAAS) REE concentration (McLennan, 1989), following recent practice (Grenier et al, 2018; Zhang et al, 2008; Friend et al, 2008). Such normalization generates smooth plots of the REE because it helps to reduce the effect whereby even atomic-numbered REE are more abundant than odd atomic-numbered REE (see Grenier et al, 2018). Concentrations were then normalized again to the mean PAAS-normalized REE concentration of each sample, similar to the approach Serno et al (2014) used normalizing to average Upper Continental Crust. This normalization approach differs slightly from the Upper Continental Crust normalization used by Serno et al (2014), although there is a very minor impact on the REE patterns (Garcia-Solsona et al, 2014, Appx. A.). Another reason we used this PAAS normalization approach is that it has been used in recent publications where the europium anomaly, Eu/Eu*, is estimated (e.g. Friend et al, 2008; Zhang et al, 2008; Grenier et al, 2018) and is thus consistent with recent practice in the oceanographic community.

- Imported original file "REE_Th232_raw_data_CopperRiverDustBCO-DMO.csv" into the BCO-DMO system.
- Made longitude values negative to indicate West direction.
- Renamed fields to comply with BCO-DMO naming conventions.
- Converted date field to YYYY-MM-DD format.
- Saved the final file as "925359_v1_ree_th232_raw_data_copperriverdust.csv".

- Imported original file "REE_Th232_PAAS_MeanNorm_CopperRiverDustBCO-DMO.csv" into the BCO-DMO system.
- Made longitude values negative to indicate West direction.
- Renamed fields to comply with BCO-DMO naming conventions.
- Converted date field to YYYY-MM-DD format.
- Saved the final supplemental file as "925359_v1_ree_th232_normalized_copperriverdust.csv".

NSF Award Abstract:
An important part of understanding chemical cycling in the ocean is understanding the processes that deliver elements and nutrients to or remove them from the oceans. The atmosphere can be an important source of material from rain and from dust. In some parts of the oceans, input of elements such as iron from dust (and possibly rain) can play an important role in stimulating the growth of phytoplankton, the base of the marine food chain. This study will involve collection and analysis of a two-year time-series of filtered air samples from two locations in the subarctic North Pacific, aimed at understanding the inputs of metals to the ocean surface from Alaskan and Asian dust and from fossil fuel combustion. The dust time series observations and analyses proposed will provide insight that will aid the interpretation of observations from the 2018 U.S. GEOTRACES Pacific Meridional Transect expedition. This project will support a graduate student for two years and will be the basis of his or her Ph.D. research.

The investigators will collect atmospheric dust samples from two land-based locations well suited to capture the flux to surface waters of the subarctic North Pacific. One is on the northern side of the Aleutian chain, and the second is an island to the northeast (Middleton Island), at the shelf break, in the northern Gulf of Alaska. Sampling will be carried out year round at Middleton Island using an automated sampling system, while high-volume sampling will be carried out for one month at each of the two locations during targeted known seasonal events. These will help infer and characterize inputs from Asian dust, known to occur primarily in the spring, and from Alaskan dust, known to occur primarily in the autumn. Samples from large events will be made available to the GEOTRACES research community for analyses. The HYSPLIT model will be used to simulate dust transport and deposition, for comparison to observations. Previous work documents substantial interannual variability in dust fluxes from Alaska and from Asia, and volcanic ash input is even more variable. Given the seasonality and interannual and spatial variability of the dust sources, and the short residence time of Fe and other trace elements and isotopes in surface waters, this history of deposition from dust and fossil fuel combustion, and its geochemical characterization, will help to distinguish the various inputs more comprehensively than is possible with a single research cruise.&#8232;A variety of size fractions will be collected and analyzed to infer the particle size range that contains each of the metals. Analyses will be carried out on over twenty elements, including Fe, Al, the REE, the isotopes of Pb, Sr and Nd (the tracers that best distinguish the dust source end members), and metals including V and Ag to infer contributions from fossil fuel combustion. Activities in air of the natural radionuclides lead-210 and beryllium-7, and their depositional fluxes, will be determined to convert elemental concentrations in dust into flux estimates to surface waters. Experiments will evaluate the control of different organic ligands on the solubility of iron and other metals in surface seawater, the relative solubility of different size fractions, and the relative solubility of dust- versus fossil fuel-derived metals.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.