New perturbation run

[adele-morrison]

It sounds like we have sufficient hours this quarter to do another perturbation. Let's discuss here what perturbation we want to do. I'm thinking another 15 years of a step function anomaly. What do we want? An addition of the CMIP6 multi-model mean wind anomaly projection for 2100? The same as what we have, but only the zonal component? Do we need a positive and negative perturbation, or are we happy with just one sign? Discuss away.

[StephenGriffies]

Whatever we do, I suggest we hit it reasonably hard to ensure a clean signal. I like just focusing on the zonal.

[AndyHoggANU]

Given the symmetry in these runs, perhaps we only need one simulation initially -- an increase in the zonal component of wind strength?

If we are considering a second run, I would advocate an increase in the meridional component of the wind.

I am also keen for a large signal, but perhaps the same magnitude as the initial runs is large enough??

[matthew-england-unsw]

Yes I'm definitely drawn first to a 'zonal only' version, and for cleanness to match in every other way the existing runs. Yes one sign first is probably sensible, and then maybe next would be the meridional version. Recognizing that alongshore is not always zonal. :-)

[PaulSpence]

I think we will have confidence in the perturbation signal within 3-5 years. Given the Control Ekman downwelling is very coastal, I'd like to see what happens to shelf temps with a zonal easterly wind increase. These experiments aren't shifting the position of the westerlies, which I hypothesize is key to wind warming. I support Stephen in going a little big in the perturbation (for fun): maybe double down on the zonal wind perturbation component. Happy to be overruled.. Also happy to help do the run (create forcing).

I've learned a great deal from our current perturbation experiments (all easterlies increase), primarily: 1) a better understanding of what the shelf winds look like (easterlies, southerlies, Ekman variability). The JRA forcing (and its implementation in the model) is so much better than CORE. 2) Adele's DSW faucets are blowing my mind; the plumbing of coastal Antarctica just pops into life.

Thank you. Paul

On Mon, Aug 2, 2021 at 8:52 AM Matthew England @.***> wrote:

Yes I'm definitely drawn first to a 'zonal only' version, and for cleanness to match in every other way the existing runs. Yes one sign first is probably sensible, and then maybe next would be the meridional version. Recognizing that alongshore is not always zonal. :-)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/adele157/easterlies-collaborative-project/issues/28#issuecomment-890599588, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABSWJXBFDJ5J2ARSRA6W23DT2XF3DANCNFSM5BLRPAAQ .

-- Paul Spence, PhD ARC Future Fellow School of Geosciences University of Sydney, Australia https://paulspence.github.io/

[matthew-england-unsw]

The existing 20% UP / DOWN experiments are a decent hit as that's the wind speed factor, so more like ~50% changes in wind stress (assuming no sea-ice damping feedbacks). I agree good for signal to noise to go for a significant jolt in the forcing, but the existing design does that I think. Could try double UP / DOWN down the track, but one step at a time always cleaner (i.e. try separately forcing zonal // meridional first, to decompose existing runs). My vote at least :-)

[adele-morrison]

Just to note that the existing perturbations are 10% wind speed (~20% wind stress). I agree that these seem large enough in the current runs.

I think if we amplify the existing easterly pattern, we’ll still see a DSW increase in certain locations where the katabatics are zonal (i.e Ross Sea and perhaps Weddell Sea to a lesser extent). If we think this is a problem, we could add a zonal average easterly perturbation instead of amplifying the existing easterly pattern. But that won’t be very comparable to the existing runs.

On Mon, Aug 2, 2021 at 11:10 AM, Matthew England @.***> wrote:

The existing 20% UP / DOWN experiments are a decent hit as that's the wind speed factor, so more like ~50% changes in wind stress (assuming no sea-ice damping feedbacks). I agree good for signal to noise to go for a significant jolt in the forcing, but the existing design does that I think. Could try double UP / DOWN down the track, but one step at a time always cleaner (i.e. try separately forcing zonal // meridional first, to decompose existing runs). My vote at least :-)

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/adele157/easterlies-collaborative-project/issues/28#issuecomment-890637938, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACA44U2WDPGWB6LGC27PI3TT2XV77ANCNFSM5BLRPAAQ .

[matthew-england-unsw]

Thanks for clarifying Adele. Agreed the anomalies at the moment seem big enough to give a decent signal.

Yes good point about the easterly wind component occasionally being cross-shelf.
I guess the other way to decompose the existing experiments is to run additional experiments where we only increase/decrease the alongshore / cross-shore winds. That would require some kind of wind vector decomposition as a function of the large-scale continental shelf outline. Maybe a bit fiddly for the moment?

Matt

[adele-morrison]

I ran the perturbation we discussed last time for 6 years, and the results are not what we expected. To jog your memory, we decided to repeat the UP perturbation, but with no wind speed anomaly applied in the 4 small regions where DSW is formed. You can see the resultant wind stress averaged over the 6 years below. Somewhat overconfidently, I called the new experiment "UP noDSW" (or just "noDSW"). The wind perturbation applied is identical nearly everwhere, except in the regions circled in the difference plot here (UP minus noDSW):

[adele-morrison]

The DSW did not decrease as expected, and is nearly identical to the UP case, as you can see here in the surface mass transformation at each of the dense water regions:

Here are time series of the SWMT, where I've picked out a density that shows the formation of the densest set of waters for each region (the density where half the maximum transformation occurs).( From what I've seen elsewhere, the formation of densest waters, rather than the formation at the peak density seems to match better with the waters that actually overflow.)

[adele-morrison]

I also checked the SSH time series (following Wilma's code) and change in bottom age, which both show minimal difference to the original UP simulation:

[adele-morrison]

The SWMT figure above shows that there is clearly an increase/decrease in DSW formation in the UP/DOWN runs, as we have seen elsewhere. But this change in DSW formation doesn't seem to be driven by the local katabatics as we expected.

I'm wondering if the DSW increase could instead be driven by the change in the large-scale upwelling that occurs in the UP experiment. We saw in the Ekman pumping, that the easterly increase results in more upwelling nearly everywhere, and that there is only enhanced downwelling along a very narrow coastal strip. Could the enhanced upwelling be bringing up more CDW waters, which then result in more heat loss in polynyas and more DSW formation? So a more remote driver of DSW change rather than a local one?

[StephenGriffies]

Thanks for this analysis @adele157. In addition to possible mechanical forcing from winds, could this result also point to the importance of E-P forcing (also induced by winds) as proposed by @matthew-england-unsw ? If so, then presumably the E-P forcing will look very similar between these two runs...

[StephenGriffies]

Is it worth doing the complement expt to fully rule out the importance of local katabatic forcing?