 Hello everybody can hear me? Yes. Okay. I can't hear you all so well So I'm just gonna talk and I can see Fred in the front. So Fred wave your hands if there's anything that I need to Okay So thank you everybody for letting me participate remotely. I'm sorry not to be there and I I'm certainly missing the the Piatto's at the cafe upstairs But I've been able to chime in for most of this Watching it on the web stream, which is really awesome. So thank you all at the center for making that possible So I'll just jump right in This is I'm gonna present some results from a paper that just came out a few weeks ago in science With the same title the Atlantic multi-decadal oscillation without a role for ocean circulation And this is a collaboration myself and a former student could think of Bellomo Lisa Murphy Mark Kane and then Some colleagues from Germany Dorsen Morton Gabby Riddell and Bjorn Stevens So this is figure one from the paper is it you've seen many versions of this in this In the talks earlier today, and I think yesterday as well on the left panel is the time series of North Atlantic sea surface temperature linearly detrended as Neil Tanden warned us against but it's common in the literature at least at this point and On the right hand side is the regression of sea surface temperature sea level pressure and winds on that AMO index and As many people have discussed There's centers of action in the North Atlantic and in the tropical Atlantic, but the whole basin is warmer And it's associated with a characteristic pattern of atmospheric circulation that you see which is a weak subtropical high It a particular phase of the NAO as well projects on to this and then significant trade wind variability in the tropics as well So what drives the NAO and I think we've heard we've heard many talks about this already my read of the literature and And I think it's consistent with what we've heard is Basically that naturally occurring changes in the strength of the AMO can change the amount of heat It's delivered to the North Atlantic and that that appears is it sst signature And then the eye you've heard talks just recently about how those might impact and turn feedback on the atmospheric circulation So this idea has obviously been tested in any of the couple models from around the world and there is does it be a connection between the strength of the AMO and the temperature of the North Atlantic and so that that Relationship has been now and it's also been shown in paleo climate simulations where the Where the Where the AMO is shut down by the North Atlantic with fresh water and you see a cold North Atlantic at that time And that's consistent with paleo climate observations as well so this this does appear to play out in climate models, but the question is causality and So we take a different approach rather than taking looking at all the fields Together, which as we've seen by the talks for example by Tom Dower That there is a there these fields atmospheric fields are correlated with ocean fields and sst But here instead what we did took a different approach and ask the question How do the features of the climate simulation in this case of the AMO? But it can be applied to other climate phenomena as well Changed by progressively turning off elements of the climate model So Slide shows my view of what a climate hierarchy would look like and Actually, we're working with people at NCAR to develop community versions of all of these different elements of the climate Model hierarchy that will be served to the community and available and supported To the community with the next CES and release So that's just a little plug for the work that I've been doing with NCAR people Brian Manderos in particular And I have a research scientist working there Jim Benedict on the aqua planet Version of the model that will be available to the community and then Lorenzo Pavani and colleagues have been working on If you go further down the dynamical core and possibly single-column versions of CESM basically So that's a bigger picture, but now today. I'm just going to talk about the first three components which are the fully coupled model of the top with an atmosphere coupled to land and ocean model and you've seen many results from the type of model thus far and we simplify that by replacing the ocean general circulation with the slab ocean model and Which still has a land model in it and looking at results comparing to and then further Simple by things by prescribing SST simulations or prescribing the SST and simulations with the coupled to the Atmosphere and land model and those are courtesy of people at NCAR Who've done those runs for other reasons? So more details about what we are at what I'm going to show the We're showing I'll show you results from fully coupled models both from the three archive and seem five Pre-industrial runs here. We are simplifying our lives by taking out this the time-bearing forcing so we can at least not not be not Complicate matters as Neil has pointed out why introduce the time-bearing forcing and then The next set of simulations are from the CMIP three archive, which is atmospheric GCMs coupled to 50 meter slab ocean and There are 13 such models 12 of them were in the CMIP three archive and then there's an additional one with NCAR CCSM three or cam four coupled to 50 meter slab that was done at NCAR separately and And these runs were done just kind of interesting historical note for Not for the purpose of looking at internal variability, but actually for estimating Equilibrium climate sensitivity which is done differently in CMIP five And so we don't have the corresponding set of slab ocean models but to all the people from big modeling centers out there I would encourage you to to Make long runs of the slab model because I think that there's a lot of interesting internal variability in those models that is worth looking at I hope to convince you of that today So in this configuration the atmosphere and ocean are thermally couple or thermodynamically coupled So they're changing heat and moisture, but there is no interactive ocean transport So that winds can blow stronger over the ocean and there'll be no change in the ocean heat transport Convergence that then affects the surface temperature and I think an earlier talk in this session showed equation there is the tendency is equal to the surface fluxes plus a few flux which are Which is meant to include the mean influence of ocean heat transport on the simulated sea surface temperature but doesn't vary with time and Then I'll also show some results from the atmosphere a couple to the uncoupled and the prescribed sea surface temperature in camphor So let me just pause there. Are there any questions? Maybe this is a good test of whether I can hear you any questions from the audience No, right, then I'll move on is the sound okay Can someone just yell if the sounds okay because I can't see anybody anymore Answers there was no question right now What's that? Can you hear me? I can hear you. Okay. The the answer is there was no question at the moment And it's the sound okay. The sound is okay Okay, great All right, so the first question we ask is how the spatial patterns pair with and without interactive ocean heat transport So let's return to that picture that I showed earlier of the of the pattern Sea surface temperature and atmospheric circulation that goes with this north that land SS index Joan low at the bottom left And if we look at this same Pattern, which is the regression on the unfiltered time series. So this is just the annual mean sea surface temperature You see that this couple models. These are fully couple models now simulate something that basically looks like the observations in the sense that there's a weaker subtropical high and warmer warmer north Atlantic and tropical Atlantic the time when the Amo is positive and the pattern agrees very well and there are Signals in the tropical Atlantic and there are signals in the wind fields as well that go with this pattern So our argument is this looks very much like the observations and this is from the fully couple models and I would say the usual Interpretation of this would be that this variability is it At least in part relate to changes in ocean circulation But then we we do the same calculation using just the slab models and we find that the pattern doesn't change So that you have the same weakening subtropical high surface wind omelies in fact in this case actually The the tropical anomalies look a little bit like observations In the slab version then fully coupled in the sense that the truck weight anomalies are somewhat larger in response to the heat Warm or at this when the Atlantic is warm and then in the fully coupled models I'll come back to that point a little bit so one question that came up in Reviews of this paper and discussions with people some are who are in the room is Maybe it's only at the low frequencies that the impact is of the oceans in parent And so if you redo the calculation that I just showed and look at the regression of sea surface temperature on the call it that AMO index the top panel is the unfiltered version and the bottom panel is the Look past filter for the ten year ten year Langso's filter so What you see here, and this is for the coupled models So what you see here is that at the cup at the lower frequencies on longer than ten year time scales That you start to see an amplification of the signal in the sub polar region compared to the tropical region Although even in the filtered version so maybe even at higher frequencies the the Subtropical sick component of the AMO is larger than the tropical component But then if we do the same calculation For all the models that we have and look at the multi-model mean The you get basically the same answer, which is that the pattern looks a lot Looks very similar even on low frequencies so at this point doesn't look particularly It looks like this lab model is reproducing a lot of the main features some any of which are discussed here and in this conference as the coupled models Maybe that's not surprising It's just the the mean state that is responding that the pattern is set up But is it possible that oh before I go on I just point out you do the same calculation We're looking at the unfiltered in the top panel and the low-pass filter in the bottom panel from different SST data sets that I think and maybe a subject for discussion and that the amplification of the high-latitude signal is Not as apparent in the observations as it is in the coupled models in other words the in the low frequency the whole basin appears to be responding or Covarying on on the low on the lower frequency timescales just as it does on the higher frequency timescales Whereas in the coupled models the lower frequency seems to be more in the high-latitude regions I think Tom Delworth may mention that or maybe Rong Zhang earlier in her talk, but I missed that That that maybe this has something to do with deficiencies in the climate model simulations of Clouds and Atlantic, but that's another talk So what about the temporal characteristics? So we have this pattern that emerges in the slab model and I'll I'll talk about the mechanism in a moment But let's just look at the time characteristics of the of the coupled models and the slab slick lights the left panel is the spectrum of the slab models in red, so that's a The red line is the multi-model in spectrum and the red shading is the is the min max of all the models and What I think you can see is it compares that with the same calculation for the coupled models that the two line up very well and there Doesn't it doesn't there doesn't seem to be any significant differences in the multi-model meaning sense between the AMO Simulation and slab model couple models Now I will note that don't have spirit We only have a few long slab model runs one is and cars and cars any of you have long slab versions I would love to see them and cars the only one that I have the abilities to actually make simulations of myself so on the right hand panel is the same C-Mip 3 spectrum in Blue and then we put C-Mip 5 on top again. These are pre-industrial simulations, so What you can see is that there's there's no significant difference between these two in a multi-model mean sense between the two spectra It's interesting that C-Mip 5 simulates a much stronger inter-annual peak in the in the AMO, but my interpretation of this is that comes that that's make the eight Tropic or signature that in that appears in the tropical Atlantic SST Any questions here on all right? So this picture is for Gaffin I don't know if he's still in the audience, but I showed him that last figure a while back and he said well, that's not fair You'd be hiding differences between Individual slab and coupled versions that don't appear in the multi-model mean sense or when kind of shade out all the Individual models and so we for the models that we had where we had more than 70 years of simulation of the slab model we compare the the fully coupled with the slab versions of individual six individual models and overall there are The slab versions of the models are not significantly different In fact, some of them actually simulate more variability in the AMO than the couple of versions I'll point out that you came out office simulation and Except for the GFL C2.0 model and there you can see that there's significantly more variability in the multi-decadal timescales in the Coupled model compared the slab model and if you separate if you break down components of the Contribution from the tropics versus high latitudes most of that comes from subpolar gyre variable or subpolar at the steep variability And I think you heard Tom talk about the mechanisms some mechanisms of subpolar heat content variability We've heard a lot about that in this session already so We have a simulation of the slab model that looks a lot like the the Observation or it looks a lot like the couple model both in time and space characteristics and it looks like Surveillance, so what is the mechanism that produces that and our interpretation of this is that the AMO in the slab models Is the response to stochastic forcing from the mid latitude atmosphere circulation i.e. the NAO With thermal coupling in tropics playing an important role in my thermal coupling I mean what the earlier talk in the session pointed out Exchange of heat moisture between the upper ocean and the atmosphere But not necessarily any role for ocean dynamics in the tropics so to demonstrate this we looking at a The NAO in a version in a simulation with cam4 with scribed climatological SSTs I that is uncoupled and you see a regression of Sea level pressure and winds on the subtropical high index could have done this on the NAO and you get basically similar results but the so you see that pattern of wind variability in Particularly in the subtropics and in the latitudes and on the right hand side is this spectrum of the subtropical high index which is should we'd expect and then We do is compare it's in the The same regression now in this slide comparing the left-hand side the sea level pressure and surface winds With on the right-hand side cam4 coupled to a the slab ocean version and This is low-pass filtered so emphasize some the Thermally-coupled parts you could do they get it basically this picture any time you you filter with more than basically a year of Have a look at the terrain and you see Here that the main that the first of all the pattern with the large signals in the mid-latitude and tropical region Emerging as a response that changes the strength of the subtropical high which in our interpretation of this is is basically Internally generated in the atmosphere But with the main difference between the left-hand panel and the right-hand panel is in the tropics so obviously there's no sea surface temperature change and on the left-hand side because we prescribe the sea surface temperature but on the right-hand side what you see is there's wind signal in the tropics that is absent in a couple of version and That that shows that the trapeze weekend at times when the subtropical high is weaker and That so our interpretation of that is is that is the main driver of the though the sea surface temperature anomalies via a Wind evaporation surface temperature feedback that I think you heard about in a talk earlier in the session that is absent in the prescribed case So you can have an NaO But in order to get a tropics water or an Atlantic wide signal in response to the NaO variability You need to have thermal coupling in the tropics that the signal will not send into the tropics unless there's a feedback between the up-push and heat Tends and the atmosphere atmospheric circulation Just one thing that I think just thing about these results is if we compare The uncoupled run in spectra in blue and the slab model in red again, this is the can't just the cam model That the spectra whites for mid-latitude winds Zonal meridional winds and zonal winds in the tropics if the same if I did that showed this subtropical high in index itself But the main difference between the uncoupled and thermally coupled cases is in the meridional wind in the tropics so that coupling with the upper ocean Produces meridional wind anomalies that give you persistence that is more like a red spectrum and most of the rest of the of the wind variability is uncoupled and I think that's you know, I I think that's it's an open question about how much the upper ocean heat content is feeding back on the Atchric circulation and this Simulations it looks like the only piece where that's happening is in the meridional winds in the tropics and just this is the slide for more student line and Illustrating the mechanisms. This is probably not necessary for this audience, but I'll just walk through it So the interpretation is that the NaO variability And the that extends into some proper but not into the tropics Produces trade wind anomaly. So a weaker subtropical high Will produce weaker trade winds through a West feedback uses former sSTs the sST gradients associated with that then allow signals or have an atmospheric circulation response which weakens the trade winds and produces SST anomaly in that extent so you can end up basically filling out the entire tropical Atlantic with They call it a a Forcing from the subtropical subtropical high So three points that I wanted to end on the fully coupled slav ocean miles produced and then That does have the spatial characteristics of the observations and in this case has read specs In an overall multi-model mean sense That interactive push-and-heat transports In models do not change the space-time spectrum of characteristics of the AMO again in an overall robust sense individual models such as the GF Yale may produce variability that looks That is enhanced in the said pole gyre and I don't I don't Be surprised that that's happening real-world also, but it's not necessary to create this pattern and and Does it necessary change the overall? AMO variability most models and then the internal physical interpretation is that the AMO and models is the Uprocean thermal response to the NAO and then a tropical signal emerges through thermal coupling So you can think of the AMO as a hybrid force thermally coupled modes So, I don't know how I'm doing on time and somebody yell out how much time I have left I hear laughter, but I don't earn any any numbers. I think you've talked for about a half hour So, so we should probably take questions if that's okay with you