Dissolved Oxygen
View Data
To assist in the interpretation of the data, it can be displayed using the Hawaii Ocean Time-series Data Organization & Graphical System (HOT-DOGS©).

Dissolved oxygen samples were collected and analyzed using a computer controlled potentiometric end-point titration procedure as described in Tupas et al. (1997). As in previous years we measured, using a calibrated digital thermistor, the temperature of the seawater sample at the time the iodine flask was filled. This was done to evaluate the magnitude of sample temperature error which affects the calculation of oxygen concentrations in units of µmol/kg. The Figure below (upper panel) shows a plot of the difference between on-deck sample temperature and potential temperature, computed from the in situ temperature measured at the time of bottle trip, versus pressure. The lower panel of the same figure shows a plot of the difference between oxygen concentration using the sample temperature and potential temperatures versus pressure. The depth dependent variability in Δ-oxygen is a result of: 1) bottle warming as the rosette is brought up through the water column 2) warm air entering the niskin bottle as samples are being taken and 3) evaporative cooling that occurs while on-deck as bottles are waiting to be sampled.

Precision of the Winkler titration method is presented in the Table below. The pooled annual mean CV of our oxygen analyses in 2007 was 0.25%. It was calculated by averaging the mean CV of N-triplicate samples on each cruise. Oxygen concentrations measured over the 19 years of the program are plotted at three constant potential density horizons in the deep ocean along with their mean and 95% confidence intervals (Figure 12). These results indicate that analytical consistency has been maintained over the past 19 years of the HOT program.

During HOT-145 (February 2003), a noticable negative anomaly in the deep bottle oxygens was observed. Values were 5-7 µmol/kg lower than historical data. The historical frequency of such suboxic events are unusual and several experiments were conducted to assure that our instruments (DOSIMAT, pH meter and REDOX electrode) or fixing reagents were not the cause of the decrease in O₂ concentrations. An investigation, conducted over a 2 month period in 2003, reveled that the electrode was accurate, but imprecise. Changing the electrode returned oxygen concentrations to pre HOT-145 values.

Based on this situation, we have started using a series of protocols to check and maintain electrode performance. An Orion ORP standard is used to check the accuracy of the electrode on each cruise. We also check the precision of the electrode after each cruise. In addition, the electrode fill solution is changed before each cruise and the electrode is replaced yearly.

After the offset in O₂ concentrations were noticed, previous cruise's data were re-evaluated to determine the temporal extent of the problem. It was determined that O₂ concentration data between HOT-142 and HOT-147 should be considered suspect. An attempt was made to correct the previous bottle oxygens, but no satisfactory method of back correcting the data was determined. In the end, the decision was made to not apply any sort of correction and to mark all samples from HOT cruises 142 to 147 as 'questionable'.

Figure 13 and Figure 14 show contoured time-series data for oxygen in the upper 1000 dbar at Station ALOHA. The oxygen data show a strong oxycline between 400 and 625 dbar (26.25-27.0 σ_θ), and an oxygen minimum centered near 800 dbar (27.2 σ_θ). Recurrent drops in the oxygen concentration can be seen throughout the time-series between 25 and 26.25 σ_θ. These features are accompanied by a decrease in salinity and an increase in the nutrient concentration. The anomalous low oxygen centered at 400 dbar in early 2001 is due to the eddy observed during HOT-122.

The oxygen minimum exhibits some interannual variability, with values less then 30 µmol kg^-1 appearing frequently during the time-series. This variability can be seen in a plot of the mean oxygen in the intermediate waters spanning the oxygen minimum (27-27.8 σ_θ, Figure 31). Superimposed on this variability is a general trend towards lower oxygen values from 1989 through 1996, with an increase between 1997 and 2000, followed by a shap decrease during 2001, and reaching record low values during the second half of 2002, and increasing sharply during 2003 and 2004 to reach record high values in mid-2004, to decrease again to values similar to those in 2002 by the end of 2005.

The surface layer shows a seasonality in oxygen concentrations, with highest values in the winter. This pattern corresponds roughly to the minimum in surface layer temperature (Figure 1).

A contour plot of dissolved oxygen concentration in the upper 200 dbar of the water column from 1988-2007 based on analyses of water samples collected at discrete depths is shown in Figure 15. Dissolved oxygen shows a maximum between 60 and 110m depth that develops during the summer-fall. This maximum, presumably of biological origin, is typically eroded during the winter.