Analysis of individual sub-regimes

[wghuneke]

Similar analysis to here, but instead of calculating total regime averages only averaged over the individual regions.

[wghuneke]

Thick lines are the total regime average, thin lines represent individual regions.

The name convention* is as follows: For Surface:

• PR_WD: surface-intensified regime between Prydz and Weddell
• AS_RS: ... between Amundsen Sea and Ross Sea
• RS_AD: ... between Ross Sea and Adelie Land
• AD_PR: between Adelie Land and Prydz Bay For Deep:
• PR: Prydz
• WD: Weddell
• RS: Ross
• AD: Adelie

Time-averaged velocity and CSHT profiles.

Climatology of depth-averaged velocity and CSHT:

*My fingers automatically wrote "conveCtion" :)

[wghuneke]

Correlation analysis between velocity and CSHT (for monthly time series).

Overall, the individual regions show a similar pattern as the total regime averages, but with varying magnitudes.

Total regime average:

Surface sub-regions:

Deep sub-regions:

[ongqingyee]

Following on from our figure discussion, I did the same analysis for specific regions. Totten, WAP and Ross, for each of the three regimes. The Ross and WAP regions were chosen from Morrison et al. 2020 and Maurice's ENSO paper. Will include a plot with the regions soon.

Below is the time-mean CSHT and velocity. They are color coded with their respective regimes.

And here are the layerwise correlation analysis. IMO the regimes look very similar to the masked regions, with even higher correlations in WAP. This could have contributions from the slightly positive time-mean u_along. For deep regime, we have stronger correlations in the monthly and annual-averaged timeseries, but slightly less correlation in the monthly climatology. Might there be a Ross shelf specific reason for this?

A version of the scatter plots that Will made might be useful for looking at the regions further.

[adele-morrison]

This is great to see, thanks Ellie! I agree, it looks like generally the correlations are higher everywhere (except in the Deep region in the monthly climatology).

It makes me wonder how much it actually makes sense to average ASC speed and cross slope heat transport across whole regimes first before doing the correlations. There could be opposite changes in both these quantities in different regions within a regime that are cancelling out and reducing variability/correlations when we average across such large areas.

Would it make sense to find correlations for small areas first (say 10deg longitude bins) and then average the correlations across different areas within regimes afterwards?

[ongqingyee]

Adele, I think I agree with what you mean - for the idealized model I averaged over the whole domain too. However, I chose high onshore flow regions here, with canyons on the shelf. This might also strengthen the correlations, as opposed to, for example, fresh shelf regions where a slope would inhibit CDW onshore transport even under a weak ASC.

For completeness these are the regions chosen: Totten = (-247,-234.5) WAP = (-90, -60) and Ross = (-200,-170). Definitely happy to tweak the boundaries. Note the discontinuity as well at the Ross Sea and Weddell Sea in the lon_bin_midpoints where the contour was not reconnected when the masking was done. also seen in Taimoor's plot. I can try to do the masking with the full isobath contour.

[ongqingyee]

Below are the regional correlations using daily data, and I also did the annual-averaged and daily climatology to see if a higher time resolution makes a difference.

For daily, we see the same results as in the plot above, with a lot less standard error. The r2 correlations do drop though. Annual averaged and climatology correlations are very similar to that of monthly data, with errors dropping in daily climatology.