Data were collected from the Northern Gulf of Alaska, along the shelf and slope from the mouth of the Copper River to about 50 kilometers (km) past the shelf break. This was roughly along a N/S transect along ~145º West, between 59º North and 60.2º North and a few samples collected along an E/W transect between ~146º West and 145º West. Sample collection occurred during three research cruises in 2010 (7-9 April (CR2010-01), 5-7 May (CR2010-02), and 27-29 July (CR2010-04).

Surface waters were sampled by underway pumping of seawater through Teflon-lined tubing using a technique adapted from that of Vink et al. (2000), with the intake positioned a few centimeters forward from, and mounted to, a PVC towfish towed ~2 meters (m) below the surface. The towfish was suspended by a polyester line from a boom extending ~10 m from the starboard side of the ship to minimize contamination. Seawater was pumped using a shipboard air-operated Teflon-lined diaphragm pump to a shipboard clean lab, where both filtered and unfiltered samples were collected. Acropore cartridges (0.45 micrometers (μm)) were used to filter the surface water samples inline under N₂ pressure, without requiring any intermediate collection vessel, after flushing 10 liters (L) through each cartridge before first use. Surface water salinity, temperature, and sampling depth were logged once per minute using a YSI Sonde attached to the towfish.

Samples below the surface were collected using 8 L externally closed Niskin bottles whose inner face was Teflon-coated, attached to Spectra (Dyneema) line, and triggered at depth using Teflon-lined messengers. Additionally, profiles of salinity, temperature, fluorescence, and turbidity were measured with a Seabird SBE16 CTD, deployed at the same time as the collection of the subsurface water samples. Data from sub-surface samples are available as a related dataset (BCO-DMO dataset 891918).

Filtered samples were analyzed for dissolved aluminum (Al) and unfiltered samples were analyzed for total dissolvable Al using Flow Injection Analysis (Resing and Measures 1994) using fluorescent detection of lumogallion. Samples exceeding 100 nanomolar (nM) were diluted with low-aluminum, low-manganese seawater to ensure linear response of the standard curve. Samples were assigned to a low, medium, or high standard curve depending on concentration. Low concentration samples were considered to be 0-20 nM, medium: 10-100, and high 100-2000 nM. Standards of 0, 1, 5, and 15 nM (low curve); 10, 50, 100 nM (medium curve); and 750 nM and 2000 nM (high curve) were analyzed in replicate. For the low curve, the detection limit was 0.79 nM, and 1 nM standards had a relative standard deviation (RSD) of 33% (n = 8). The minimum concentration of all samples analyzed was 3.7 nM; there were only four samples with concentrations less than 5 nM. The 5 nM standard had an RSD of 6% (n = 10), and the 15 nM standard had an RSD of 3% (n = 10). For the medium curve, the 10 nM standard had an RSD of 4% (n = 10), and the 50 nM and 100 nM standards had RSDs of 1%. The 750 nM and 2000 nM standards were measured to have a relative standard deviation <1%.

Filtered samples were analyzed for dissolved manganese (Mn) and unfiltered samples were analyzed for total dissolvable Mn according to the method of Resing and Mottl (1992), which uses spectroscopic detection of leuchomalachite green. A similar method to that described above was used to determine the accuracy of the wide spread of the data. Standards were run 10 times in succession to determine the daily precision. The 5 nM standard had a relative standard deviation of 3%, the 20 nM and 50 nM standards had RSDs of 1%. A sample collected during the campaign was used as an internal standard and analyzed at least daily, with a concentration determined to be 25.2 nM ± 0.7 (1 SD, n = 20). The limit of detection was determined by analyzing the lowest standard (0.52 nM) in replicate (SD = 0.86 nM). This value is 12% of the lowest sample analyzed.