Piecing Together the Annual Cycle of Oxygen

The Hood Canal dissolved oxygen annual cycle can be broken down into three principal components: 1) the late-summer deep inflow, 2) the winter/spring downwelling and oxygen renewal, and 3) the spring/summer biological draw-down of oxygen (Figure 1). The deep inflow period is characterized by a warm, salty, oxygen-rich bottom density current flowing southward in Hood Canal from the sill. This dense intrusion displaces the low-oxygen water in the south end of Hood Canal upwards and forces it to flow primarily northward---and potentially out of Hood Canal. This is evident by the very strong correlation of shallow northerly flow and low oxygen levels and is further supported by the distribution of dissolved oxygen in along-channel transects (Fig. 1 bottom). Since the duration and speed of the dense intrusion likely influences the rate at which the low-oxygen water is displaced upwards, it may be important to the impact of the low oxygen water on the biota of lower Hood Canal.

The downwelling period abruptly follows the deep intrusion, commencing roughly in mid-November. During this period the deep (sub-pycnocline) circulation pattern is reversed, with inflow (strongest at shallow depths of <50 m) overlying relatively weak near-bottom outflow. A very simple analysis has indicated that most of the downward displacement of isopycnals in time over this period is simply due to downward vertical velocities advecting a background density gradient. Since oxygen, salinity and temperature contours all closely follow the slope of the density contours over this period, it follows that downwelling also strongly controls the time-evolution of these properties. Thus, the intensity and duration of this downwelling---and, hence, the shallow inflow---is important to oxygen renewal in lower Hood Canal.

It appears that this downwelling period may set the background oxygen concentrations at the start of the biological draw-down in spring. This latter period is apparent from the upward slope of oxygen contours in time in Figure 1, indicating decreasing oxygen levels at most depths. An interesting observation which may be linked to the particle fall-rate of (dead) phytoplankton is that this decrease in oxygen appears to occur earliest at shallow depths and latest at near-bottom depths. It is noteworthy that in 2006 the biological draw-down period started about a month earlier and lasted longer than in 2007. There is some evidence that this may have been due to greater solar insolation (less cloudiness) in 2006. Background oxygen levels were also lower by about 0.5-1 mg/L at the start of the 2006 drawdown than in 2007. Both of these factors may have contributed to the anomalously low oxygen levels in south Hood Canal in the late summer of 2006 and to the resultant fish kill in September 2006.

Observations of an Internal Seiche

The combined ORCA/ADCP Hoodsport timeseries has also contributed to the discovery of persistent, internal seiches in Hood Canal with periods of 4--10 days and velocity magnitudes comparable to the surface tide at ~0.07 m/s (Figure 2). These are internal waves that are resonant, or that become amplified if the wave energy reflected from the boundaries (ends of Hood Canal) adds constructively to the incident wave energy. A classic example of a surface (opposed to internal as observed in Hood Canal) seiche is the vigorous sloshing that can be made in a bathtub simply by pushing the water at just the right frequency (like pushing a playground swing). 

For the latitude of Hood Canal, these waves have frequencies well below the lower theoretical frequency limit for freely-propagating oceanic internal waves. Theory suggests that the narrow channel configuration of Hood Canal suppresses the influence of the Earth's rotation for motion in the along-channel direction, causing the waves to behave as linear internal waves without rotation---which have no lower limit on their frequency. Although a likely reason for the waves' narrow frequency range is the preferential response of basin resonant modes (seiches) with the along-channel wavelengths estimated to be similar to Hood Canal lengthscales (~80 km), there is some evidence that suggests dominant variability in wind forcing at the wave frequency may contribute to the persistence and “narrow-bandedness” of the waves.

Observations at the Hoodsport location show predominantly upward phase propagation in time (Figure 2). Internal wave theory indicates that this corresponds to downward energy fluxes. Horizontal energy flux calculations show that these waves are predominantly traveling to the south. The downward energy flux and a strong correlation of the horizontal energy flux with wind stress suggest the waves are wind-generated, explaining a strong seasonal cycle in wave energy. These waves are of particular interest because there is some indication that they may break near the bottom, potentially elevating mixing and influencing nutrient and oxygen fluxes.

Figure 1:
Depth-time maps of 30 day-averaged, along-channel velocities (color) with contours of dissolved oxygen (mg/l, top panel) and sigma-theta or potential density anomaly (kg/m³, bottom panel). The two lower small panels show transects of dissolved oxygen using PRISM/Citizen Monitoring Program data for specific times in the timeseries plots. Velocity vectors are over-plotted at the Hoodsport and Eldon (where data are available) mooring sites.

Figure 2:
Depth-time panels of 2-30 day bandpassed north-south velocities at the Hoodsport location (color) with over-plotted contours of the density anomaly. Four different, roughly 2 month-long periods over 2006 and 2007 are plotted showing the time-variability of the structure and period of the internal seiches. Note the dominant upward-phase in time.

Figure 3:
Sund Rock temperature profiles.