Using a Welch’s t-test, I have statistically tested whether oxygenated and hypoxic inlets lose DO at the same rate. They seemingly do. However, I have not statistically tested whether all inlets lose DO at the same rate. This blog post addresses this missing piece of the puzzle.

Figure 1 shows boxplots of d/dt(DO) between mid-July through mid-August of all deep terminal inlets (mean depth > 10 m). The green triangle is the mean value.


Fig 1. Boxplots of mid-Jul through mid-Aug net DO decrease rate of all 13 deep inlets. Data are daily d/dt(DO) rates. Each box contains the middle 50% of all data. The orange line is the median. The green triangle is the mean.


Based on these boxplots, it seems that the inlets have different variances. Thus, I chose to use a Welch’s ANOVA test, which does not assume that the groups have the same variance.

My null hypothesis is that all groups have the same mean d/dt(DO), and the alternative hypothesis is that they do not all have the same mean d/dt(DO).

The test produces p = 1.3E-14, which is much much smaller than 0.05. Thus, it is likely that at least one inlet has a different mean d/dt(DO) rate compared to the other inlets.

Based on the boxplots in Figure 1, it appears to me that Dabob Bay has a somewhat more gradual mean d/dt(DO) rate compared to the other inlets. When I removed Dabob Bay from the Welch’s ANOVA test, I find p = 0.54, suggesting that the other inlets likely have the same d/dt(DO) rate.

Therefore, we can more conclusively say that all deep inlets have the same d/dt(DO) rate, with the exception of Dabob Bay.