This week I continued looking at scatter plots of the 21 inlets. Though I learned a little more about the 21 inlet behavior, I still have many questions such as:

  • What are the dominant processes (physical or biogeochemical) that contribute to low DO in Puget Sound terminal inlets?
  • Do the most dominant processes that influence oxygen differ between high DO and low DO inlets?

Now that I have two more months for my thesis, I would like to rigorously answer these questions. Thus, I propose the following:

  • Budget analysis of bottom DO in all 21 inlets
    • Determine and compare dominant processes that influence oxygen in high DO and low DO inlets

This analysis was heavily inspired by Jilian’s budget manuscript, as well as Kuo et al., 1991, who created a diagnostic model of DO in a Langrangian water parcel moving along the bottom of the Rappanhannock River (near Chesapeake Bay).

Prior figures

Figure 1 shows the the bottom DO time series at all 21 terminal inlets (data come from a mooring extraction from each inlet). All inlets experience seasonal variability in DO, but the baseline DO levels and the intensity of summer DO depletion differs between each inlet.


Fig 1. Time series of bottom DO in all 21 terminal inlets throughout Puget Sound. Data come from a mooring extraction within each inlet.


Figures 2 and 3 are a reminder of my key results so far.


Fig 2. Average hypoxic season (Aug/Sep) bottom DO vs. different physical characteristics in 21 inlets. Pearson correlation coefficient is reported on each panel. Colors indicate in which sub-basin the inlet is located: Whidbey Basin is green, Hood Canal is pink, Main Basin is blue, South Sound is purple, and Admiralty Inlet is black. (a) DObot vs. inlet depth. (b) DObot vs. bottom minus surface potential density. (c) DObot vs. annual average RMS tidal currents. (d) DObot vs. annual average flushing time.



Fig 3. Average hypoxic season (Aug/Sep) bottom DO vs. average biogeochemical state variables in 21 inlets. Pearson correlation coefficient is reported on each panel. Colors indicate in which sub-basin the inlet is located: Whidbey Basin is green, Hood Canal is pink, Main Basin is blue, South Sound is purple, and Admiralty Inlet is black. (a) DObot vs. average surface 20 m NO3. (b) DObot vs. average surface 20 m phytoplankton. (c) DObot vs. average surface 20 m zooplankton. (d) DObot vs. average bottom 5 m large detritus.


These figures gave us an idea of what distinguishes high DO inlets from low DO inlets. For instance, high DO inlets tend to be shallow and well-mixed with strong tidal currents and short tidal flushing times. However, we still don’t know what differentiates low DO inlets from hypoxic inlets.

Scribble plots

This week, I took a look at how DO fluctuates with daily variation in stratification and tidal currents. Figure 4 shows scribble plots of these fluctuations.


Fig 4. Scribble plot of daily bottom DO vs. daily stratification and vs. daily average tidal currents during Aug/Sep in 21 terminal inlets.


Should stronger stratification always result in lower bottom DO, then we would expect all of the scribbles in panel (a) to be drawn diagonally in the same direction. Similarly, should weaker tidal currents always results in lower bottom DO, we would expect the scribbles in panel (b) to all be aligned in the same direction.

This is not the case. The scribbles are going in different directions. Thus, other processes must explain the daily fluctuations in bottom DO.

More scatter

I also took a look at the interactions between different physical processes.

Figure 5 shows depth vs. RMS tidal currents, colored by the average Aug/Sep bottom DO in each inlet. My hypothesis was that deeper inlets require stronger tidal currents to fully mix the water column. Thus, I anticipated lower DO at deep depths with weak currents, and higher DO at shallow depths with strong currents. While this relationship seems vaguely true, it is not always true. Furthermoe, the scatter plot is shaped more like a 1/x curve– thus, a correlation coefficient didn’t seem appropriate here. Also, we are missing a huge gap of data: deep inlets with strong currents. Perhaps these places don’t exist because deeper inlets will naturally have weaker currents.


Fig 5. Depth vs. annual average RMS tidal currents in 21 terminal inlets. Color indicates average Aug/Sep bottom DO.


Figure 6 shows average Aug/Sep stratification vs. annual average tidal currents in 21 terminal inlets. Again, they are colored by average Aug/Sep bottom DO. Similar to Figure 5, the scatter more closely resembles a 1/x curve. Inlets with strong stratification always have weak tidal currents. We also observe that stratified inlets with weak currents tend to have lower bottom DO, but not always.


Fig 6. Aug/Sep stratification vs. annual average RMS tidal currents in 21 terminal inlets. Color indicates average Aug/Sep bottom DO.


Figure 7 shows annual average tidal flushing time vs. annual average tidal currents in 21 inlets, colored by Aug/Sep bottom DO. The shape of this scatter also resembles a 1/x curve. Inlets with weaker tidal currents may have longer flushing times.


Fig 7. Annual average tidal flushing time vs. annual average RMS tidal currents in 21 terminal inlets. Color indicates average Aug/Sep bottom DO.


Finally, Figure 8 shows average Aug/Sep stratification vs. annual average tidal flushing time, colored by Aug/Sep bottom DO. Though this relationship has quite a bit of scatter, there is a significant positive correlation. In general, inlets with longer tidal flushing times tend to be more strongly stratified. Inlets with short tidal flushing times are well-mixed and tend to have higher bottom DO.


Fig 8. Average Aug/Sep stratification vs. annual average tidal flushing time in 21 terminal inlets. Color indicates average Aug/Sep bottom DO.


Though the shape of some of these scatter plots were surprising, the general relationships (i.e. one thing decreases as another thing increases) matched my expectations.

Figure 9 is a bonus figure containing repeat results from above. I am experimenting with different data visualization methods, and found this figure to be fun (I think it looks like funfetti cake).


Fig 9. Scatter matrix of all physical characteristics of inlets. Each dot represents one inlet. Dots are colored by average Aug/Sep bottom DO. Diagonals show distribution of each characteristic.


Proposed bottom DO budget analysis

Throughout all of this scatter plot analysis, I have not been able to identify which processes are more important in establishing low DO relative to other processes. I need some other method to compare the effects of these different processes. My proposal is to conduct a budget analysis of bottom DO in all 21 terminal inlets. Because I now have two months for my thesis, I feel confident that we can begin answering more interesting questions if I commit to this budget analysis early on (i.e. this week).

I am inspired by Jilian’s budget manuscript and the diagnostic model created by Kuo et al. (1991). They both looked at terms in a DO budget equation, similar to the bottom DO equation I derived in a previous blog post:

\[\frac{\partial}{\partial t}\int_{CV} DO\ \mathrm{d}V = \mathrm{vertical\ mixing} + \mathrm{advection\ in} - \mathrm{advection\ out} + \\ \mathrm{photosynthesis} - \mathrm{O_2\ consumed\ in\ nitrification} - \mathrm{water\ column\ respiration} - \\ \mathrm{sediment\ oxygen\ demand}\]

I hypothesize that the physical terms (i.e. advection and vertical mixing) are weaker in hypoxic inlets compared to high DO inlets. I further hypothesize that the magnitude of biological sinks, respiration specifically, is greater in hypoxic inlets than the high DO inlets. This budget analysis method will give me a way to directly test these hypotheses.

Furthermore, this budget analysis framework provides a good foundation for future comparisons to natural runs (with no WWTP loading). Rather than simply comparing bottom DO between runs, I could compare the transport of nutrients to each inlet. I can also compare the change in the respiration term between runs– this is important because different inlets will respond differently to nutrient loads. Inlets which experience eutrophication and have nutrient-depleted surface waters will have a much larger bloom response to additional nutrient input compared to inlets with plenty of nutrients (Harrison et al., 1994)

One issue: the bottom DO budget analysis work seems tricky to me because I am interested in a budget of DO below the oxycline. Focusing on just the bottom layer will allow us to understand how stratification and vertical mixing influences bottom DO. However, this presents a challenge because I cannot simply use TEF to calculate advective fluxes in and out of the inlet– I will somehow need to separate the bottom from the surface layer (if there even are two layers).

I am open to other suggestions as well.

References

Harrison, P. J., Mackas, D. L., Frost, B. W., Macdonald, R. W., & Crecelius, E. A. (1994, January). An assessment of nutrients, plankton, and some pollutants in the water column of Juan de Fuca Strait, Strait of Georgia and Puget Sound, and their transboundary transport. In Review of the marine environment and biota of Strait of Georgia, Puget Sound, and Juan de Fuca Strait: proceedings of the BC/Washington symposium on the marine environment (pp. 138-72).

Kuo, A. Y., Park, K., & Moustafa, M. Z. (1991). Spatial and temporal variabilities of hypoxia in the Rappahannock River, Virginia. Estuaries, 14, 113-121.

Xiong, J., MacCready, P., & Leeson., A. (2024). Impact of estuarine exchange flow on multi-tracer budgets in the Salish Sea. Manuscript submitted.