Continuum response - Githubissues

kurtisanstey commented 3 years ago

[x] Estimate dissipation rates, using c/GM amplitude ratios calculated in #40.
- What method? There are many, but in this case I'm assuming it's the last one (internal wave interaction theory). X eps = K<N^2>/y where y=0.2. But can't find K (diffusivity) from continuum. (Osborn, 1980; Klymak et al., 2013; Kunze, 2017) X Thorpe scale method. Overturning, would need CTD? (Thorpe, 1977) X Turbulent dissipation rate. Heat-momentum dissipation equivalence; but still need temperature. (Zhang and Moum, 2010) X D = sum(F_n), where F_n is the rate that energy is predicted to be put into each mode. I don't know enough about this one to say. (Klymak et al., 2013)
Internal wave interaction theory (Gregg, 1989; Polzin et al., 1995; Sun and Kunze, 1999; Althaus et al., 2003)
- Need shear variance from shear spectral level over the continuum vs for GM (is this ratio similar to the ratio of continuum amplitude vs GM that I found?).
- Also need shear/strain ratio from

kurtisanstey commented 3 years ago

Dissipation not GM-like due to difference in components. Seasonal energetics in continuum (cascading energy from lower freq. bands).
[x] Ratio of (V2)^2 / (Vgm2)^2 is equivalent to c/gm ratios (squared).
[x] Find a reasonable value of Rw and eps0 from literature. 2.13 as canonical value from literature (Kunze et al., 2002).
[x] Calculate dissipation rates using formula above. Similar to observations by Kunze et al. (2002) for in and above Monterey Canyon (exceeding 10^8 W/kg near topography).
[x] Calculate turbulent eddy diffusivities using K = (gamma*epsilon) / N^2. Again, similar to Kunze observations (exceeding 10^3 m^2/s near topography).

kurtisanstey commented 3 years ago

[x] Vertical scaled U(group)/N (Jody papers) for enhancement effects. Is it different at each site? It should be (N changes).
[x] Kunze 2006 for energy level fits, rather than shear level (recheck work for epsilon).
[x] Be specific in write-up regarding use of WKB-scaled spectra, or not.

Vertical scale of turbulent flow layer by sloped topography and mode-1 internal wave induced currents (Gemmrich and Klymak, 2015), determined from (pi U)/N for cross-slope U about 100 m AB. Results are on page 21 of thesis draft (135 m at Slope, and 250 m at Axis).

kurtisanstey commented 3 years ago

[x] Check U values for vertical scales. U values are fine for 'strong' forcing near the bottom, but my value of N at Axis depth was a bit low. Results are now ~314 m for Axis.
[x] Revisit c/GM, and then dissipation and diffusivity, with narrower bandwidth to account for Axis55 noise floor (and c/GM plots) and the high-frequency hump in Axis75. Removing the high-frequency 'hump' from the continuum range somewhat eliminates the noisy 'blob' from the c/GM, diss, and diff plots. Investigate more.
[x] Better time resolution.
[x] Use total rather than component for dissipation and diffusivity. Either redo or add 1/2 of each component. Done, updating interpretation in thesis.
[x] Forgot to save the total plots... Updated c/GM:

kurtisanstey commented 3 years ago

[x] Compare non-isotropic continuum with other bands to evaluate cascade of energy.

Seasonality comparisons suggest that at both sites (difficult to say at Slope) it's the semidiurnal band that contributes most to the continuum, with some input from the near-inertial band (mostly at Axis).
Only shown are the bands that appear to line up, as others don't seem to line up at all.
Mostly semidiurnal at Slope, it's a bit tough to tell at this site.
A fair mix of both at Axis (perhaps!), and it's more obvious. Seems like the general seasonality is similar to the semidiurnal, with the near-inertial contributing to intermittent pulses.
As the Slope is fairly unclear, perhaps the cascade of energy fueling the continuum is more complex on the slope than in the canyon, where it seems to be more similar to the semidiurnal/near-inertial seasonality?
Will find a few references tomorrow. This makes sense (to me) for a cascade of energy, as the near-inertial and semidiurnal are the two closest constituents.

Slope - Near-inertial (possibly some intermittent pulses, difficult to say at Slope):

Slope - Semidiurnal (appears more similar than inertial, also difficult to say):

Axis - Near-inertial (some intermittent pulses are spot on):

Axis - Semidiurnal (general seasonality looks similar):

kurtisanstey commented 3 years ago

[x] Does spring-neap affect vertical scale of enhancement layer seasonality?
There is no obvious fortnightly modulation of the vertical scale of the continuum that would be attributed to either tidal spring-neap signal (checked using tidal phase-lag process).
Annual seasonality of diurnal spring-neap cycle does not appear to affect continuum vertical scale. Continuum tends to be highest in months 3-5 and 9-11, while diurnal spring-neap strongest in months 12-1 and 6-7 (with this effect obvious in diurnal observations at Slope).
Annual seasonality of the semidiurnal spring-neap cycle is pretty weak, with amplitudes fairly steady throughout the year. Doesn't appear to affect vertical scale of continuum (or vertical scale of semidiurnal observations, with are fairly consistent in depth, through the year).
[ ] What have others seen?
[x] Depth-mean and correlate bands. Integrate in time-steps, similar to wind.
[x] Summary versions of final plots, try not to use plots that have been shown before (like above).
Similar to wind correlations, the depth-mean of band power (each band) was taken and initially smoothed in 1-month rolling averages, with four seasonal (3-month) correlations being calculated for each year (outside of data gaps). Averaging was then relaxed until correlations began to noticeably degrade. Ideal was found to be two-week averaging, and shows that the semidiurnal band has the strongest and most consistent, correlations (see below).
At Slope, continuum 'forcing' is not obvious. Both the subdiurnal and semidiurnal bands have consistently medium-positive correlations in the winter/spring. Other bands show no consistently positive correlations through the years.
At Axis, the semidiurnal band is strongly positive-correlated with the continuum, perhaps due to the focusing that semidiurnal internal tides experience in canyon topography. Additionally, the inertial band is medium positive-correlated in the fall (probably related to the few wind events that make it into the canyon).
[ ] What have others seen?
[ ] Check raw data (single time step when blob is present) for instrument noise.
Adjusting the continuum range to (7e-5 to 1.2e-4) mostly eliminates the noisy patch from analysis range (below about -680 m), so that I continue. However, there is still noise left is above about -680 m. I checked both raw PSD and band-passed velocities zoomed in around early October when the noise is present in continuum plots. Spikes in the PSD and velocities do not appear to be periodic, and I will investigate more.
[ ] Why non-isotropic? Tendencies of other bands that contribute to the cascade of energy into the continuum? Research!
[ ] Finish updating plots and write-up.

Slope (sub-diurnal and semidiurnal):

Axis (near-inertial and semidiurnal):

kurtisanstey commented 2 years ago

Archiving for reference. New issue for discussion updates.

kurtisanstey / internal_waves_barkley_canyon

Continuum response #47

[ ] What have others seen?

[ ] What have others seen?

[ ] Finish updating plots and write-up.