Dataset: O2/Ar and triple oxygen isotopes discrete samples
Deployment: Tokyo_3

Samples for TOI and O2/Ar analysis were collected from shipboard seawater intake (10 m depth) on basin-wide transects of the North Pacific between Hong Kong and Long Beach, California onboard the M/V OOCL Tianjin and the M/V OOCL Tokyo (each individual transect has a unique Cruise ID) starting in 2008 Samples were collected into pre-evacuated 500-ml flasks every ~2.5° longitude across the basin following the procedures of Emerson et al. [1999]. Sea surface temperature and salinity at the time of sample collection were determined using a Sea-Bird Electronics SBE45 thermosalinograph installed in the ship’s seawater intake. To prevent biofouling that could cause respiration in the ship’s seawater lines and contaminate O2/Ar measurements [Juranek et al., 2010], intake lines between the anticorrosive sea chest and the sampling port were purged with bleach and freshwater between every cruise.

In the laboratory, dissolved gas samples were equilibrated at room temperature for 24 hours and then drained under vacuum to remove seawater. Oxygen and argon were cryogenically extracted and separated from the remainder of the dissolved gas mixture following the procedures of Juranek and Quay [2005]. Samples were subsequently analyzed on a Finnegan MAT 253 isotope ratio mass spectrometer using 75 paired measurements of masses 32, 33, and 34 alternating with measurements of an internal standard to determine δ17O and δ18O, followed by measurements of masses 32 and 40 on a single collector to determine O2/Ar. Reported values for δ17O and δ18O are corrected for the experimentally-determined dependence on both the sample size, as described by Stanley et al., [2010], and the sample O2/Ar ratio. Measurement uncertainty for each sample batch was determined based on daily air standards, and is reported along with all sample values.

Measurement errors for δ17O and δ18O are correlated (r2 = 0.90, n = 489), resulting from mass-dependent fractionation in the sample extraction and measurement process initially reported by Hendricks et al. [2004]. To enable the correlation in these errors to be accounted for in propagating error in calculations made direction from δ17O and δ18O (i.e. gross oxygen production), we report the slope of the linear relationship between δ17O and δ18O error for each sample batch as well as the slope uncertainty.