Mr. Tom's Blog

A Lagrangian oxygen story.

As water moves through the ocean its properties will gradually change due to mixing processes or more suddenly when exposed to the atmosphere. In a major study called the Atlantic Climate Change Experiment (ACCE) in the late 1990s we deployed 50 isopycnal floats in the North Atlantic west of the mid-Atlantic Ridge to explore water movements on the 27.5 kgm-3 density surface. All floats were equipped with Clark polarographic oxygen sensors. To minimize drift the sensors were first aged for a period (weeks), then calibrated, after which they were kept wet in a dedicated container and mounted on the floats only just before deployment. While most of them exhibited some measure of gradual drift, some up, others down, the records were extremely informative of ventilation processes at the surface when the isopycnal the float was drifting on outcropped at the surface. During the warmer months they also detected sharp but brief drawdowns in oxygen presumably resulting from detritus ‘raining’ from the surface.

But first a little background: By outcropping we mean that as water cools and densifies, successively denser surfaces are exposed to the surface. Outcropping is sometimes referred to as obduction, the opposite of subduction, but I avoid that term because it also means autopsy. The 27.5 kgm-3 density surface is quite deep (>500 m) in the North Atlantic Current. It shoals across the subpolar gyre from east to west and outcrops in winter in the Irminger Sea. In fact, it is the outcropping (or disappearance) of the lighter density surfaces that constitutes the production of North Atlantic Intermediate water (NAIW) as part of the AMOC. This is exactly what happened to a few of the isopycnal RAFOS floats: They bobbed to the surface when the less dense water overhead disappeared during the wintertime cooling. At that moment the dissolved O2 level jumped up and equilibrated with atmospheric oxygen. At this point the float is just barely floating such that it readily bobs up and down in the mixed layer. After winter has passed and the surface starts to warm, the float will begin to subduct and stay submerged as lighter waters form overhead. But when the surface waters cool in fall the float will again resurface. All this is consistent with the isopycnal properties of the float.

What happened to the oxygen record was far more fascinating – reflecting three different processes: First, supersaturation due to entrainment and dissolution of gas bubbles in heavy seas, second oxygen increase in springtime when primary production gets underway, and third, net community respiration at depth. The float is an excellent platform from which to make these measurements from because as it drifts it effectively nulls out horizontal advection such one can view the data as coming from a 1-dimensional system. One sees very effectively supersaturation due to bubble entrainment during stormy periods (as inferred from NCEP/NCAR reanalyses). One sees an increase in dissolved O2 in spring when primary productivity gets underway. It appears to agree well with estimates from the Vertically Generalized Production Model (Behrenfeld and Falkowski, 1997) using satellite-derived chlorophyll concentrations. The decrease in dissolved O2 during summer can be accounted using a relationship to primary productivity, i.e. drawdown due to particulates depends upon primary productivity overhead. These findings were reported in detail Lazarevich et al. (2004).

An aspect of these Lagrangian oxygen data that has not been discussed are the striking negative spikes in O2 that occur preferentially at certain times and areas. I should mention at this point that considerable effort went into designing how to measure O2. Dr. Chris Langdon proposed the sampling scheme. The Clark sensor is operated in the pulse mode and the recorded O2 is the median value of the last 5 samples of 20 total. This approach seems to have been very effective. The data exhibit small amplitude fluctuations that are largely symmetric. They may be due uncertainties in measurement but may also reflect local variability near the float. But occasionally very large departures, > 0.5 ml/l, show up that are strikingly asymmetric and always negative; sharp drawdowns in dissolved O2. They don’t last over several measurement cycles so they must be quite local in space (and perhaps in time?). These large drawdowns are presumably due to bacteria from the detritus that result from the primary production at the surface. I assume the detritus – or at least the bacteria in them – get concentrated in ‘micro-pycnoclines’ that the float’s O2¬-sensor happens to sample. It is also conspicuous that these spikes are more common in spring and summer than fall and winter. While more common near the surface they were observed at all float depths. There is a hint that the spikes are more common along the front of the North Atlantic Current and in the Prime eddy area of the Iceland Basin. My impression from past work at sea is that profiling O2 with lowered CTDs cannot capture these low-O2 spikes suggesting that they are very thin and perhaps rather evanescent. This illustrates the beauty of the isopycnal approach: the floats drift with the water quietly taking note of the local environment as they do so.

To get a better handle on the vertical distribution and prevalence of these O2 extrema it might be instructive to program an isopycnal float to cycle slowly up and down over a suitable density range to sample and resolve O2 at very high vertical resolution. Doing this at different times and sites might allow us to gain a quantitative understanding on how O2 is drawn down. It would make sense to equip these floats with tiny VHF pingers (i.e scatterometers) to quantify particulates in their vicinity to explore their role in the draw-down of oxygen, both locally and generally.


Lazarevich, P., T. Rossby and C. McNeil, 2004. Oxygen variability in the near-surface waters of the northern North Atlantic: observations and a model. J. Mar. Res.,62,663-683.