 Hi everyone. So I'd like first to thank the organizer for inviting me. I'm really happy to attend this workshop and so the the work I'd like to present today is By addressing the question of decadal predictability associated with the amok what actually I'm gonna present today is really more What are the climate impacts of the Atlantic multi decadal variability? And so assuming that the surface signature of the amok is the AMV making this assumption based on the strong body of work What are these impacts and what can we expect in terms of predictability from the amok? so all the credits for this work really go to a postdoc at GFDL Johanna Ripri-Crobert who's working with myself and Tom Delworth and this is a work in collaboration with folks at NCAR listed here Fred, Gokha and Steve and this work I'd like to thank support from the NOAA climate program office for supporting this work So where should we look for a decadal predictability? Well, I would be really happy to find decadal predictability in many places of the globe in terms of the climate impacts Really we expect more predictability over the ocean than over lands and even for the ocean We expect more predictability over certain regions than others because of the Intrastic processes that govern variability and this is what's illustrated here by this measure of potential Predictability proposed a while ago by George Bauer that is basically a ratio of variance between decadal viability and total variance and this Predictability really highlights region like the North Atlantic the North Pacific and the southern ocean as good candidates to look for Skill on Decadal timescale But the picture is quite different when we go to real skill real prediction So when the the models the same models that I use to estimate this potential predictability are initialized from observational estimate We find quite large predict predictive skill over the Atlantic as illustrated here This is from a paper by Paco based on the initialized semi-five Prediction and it shows time series of surface air temperature global mean AMV and IPO For the average over the years two to five of the forecast and these are two different measures of skill and so while the AMV is quite well predicted over this time scale and it It's better than the uninitialized forecast shown with the dash line here The skill for the IPO and the PDO is really is really small and none There's no skill and here Paco didn't show the southern ocean, but it's also doesn't show any skill So really the Atlantic shows larger Predictability than the Pacific and not just for the AMV But also several studies showed that the these models a few models are able to predict abrupt changes in the North Atlantic like the the rapid warming that occurred in the in the supologizer in the mid 90s and you saw this figure earlier Rowan showed it and so they in particular the NCAR model the Met Office model and the GFDL model were able to reproduce this rapid warming of the North Atlantic and The three studies showed that the initialization of the amok was key in capturing this warming Thanks to the associated heat transport that allowed heat to be advected northward and give warming comparable to observation So these the models that were able to predict this supologizer warming showed climate impacts climate anomaly that are comparable To those associated with the positive phase of the AMV as observed and simulated and that's include the northward shift of the ATCZ or increased rainfall over India over Sahel during summer And so the question we're asking here is can we expect some predictability over the Pacific from the Atlantic? Even if the Pacific is less predictable Could we predict some of the climate impacts arising from Pacific? Thanks to the Atlantic and really to answer this question The first step is really to ask what would be the climate anomaly simulated by our current climate models if we properly simulated the AMV and so this is what this talk is about I will present Global impacts of the AMV using observed AMV anomalies and I will focus more on the Pacific response if I have time I'll talk about the Atlantic and I'll conclude So how are we doing this we take global models global couple models and We restore them in the North Atlantic to observed AMV pattern So here we are really interested in the internal variability of the AMV So we attempt to remove the fourth part of the signal and we do that by using a method proposed by Mingfang Ting and colleagues and that is based on a Maximized signal signal to noise maximize EOF So this is our estimate of the internal variability of the AMV And so we take only one standard deviation for each sign We and then what we do is we restore it in the North Atlantic with a restoring timescale of 100 watt per meters Square per Kelvin, which is equivalent to five day restoring for a 10 meter deep 10 meter mixed layer depth and So and the the coupling between the ocean and the atmosphere is kept free and elsewhere outside the Atlantic So we we do AMV plus experiment AMV minus experiment and we have a control where we restore to them model climatology And so these AMV plus experiments are done by applying this positive pattern added to a climatology Estimated separately from a control model and then the other each of these experiments are run for 20 years But here I'm only going to focus on the 10 years so they would be 10 year average and Here really we wanted to find some robustness So we start with two models and we run two to with these the same experiment exactly the same protocol with two models The GFGL CM 2.1 model and the NCAR CSM one model So these models have quite different resolution in the atmosphere I mean double in the NCAR model one degree in the atmosphere to a degree in the GFGL one and same about the same ocean resolution of one degree Note that we have 100 members quite many members for GFGL model and 30 member and I look at the ensemble mean here So in terms of the global impact, this is what we get for the summertime a per panel show the temperature two meter temperature and the lower panel of precipitation and so We can see that yeah in the Atlantic as expected you find the pattern you impose so the protocol basically works hopefully and then For for summertime we find terms of precipitation Impacts that are quite comparable to what has been documented associated with the observed AMV increased rainfall over Sahel drier over Europe dry Western North America and Yeah, we you few impacts in terms of precipitation over land, but quite comparable signals between the two models here The sea level pressure associated with this summer signal shows mainly low pressure anomaly over most of the northern hemisphere And this wave train coming from the South Atlantic If we move to the winter time the response in terms of temperature is dominated in the Pacific by this negative IPV like anomaly with so cooling in the eastern part of the Pacific warming on the west in terms of precipitation Not much over Europe drying again over North America a northward shift of the ITCZ Southwestern shift of the SPCZ and In terms of the precipitation Associated with the temperature IPV like anomaly we have this weakening of the Aleatian law that correspond to a negative PNA So if I'm focusing on the Pacific here and I'll come back to the Atlantic later so What drives so I'm gonna focus first on the Pacific anomaly So I'll ask you to please only look at the Pacific for now and not for the Atlantic I'll come back to that so we ask the question what drives the Pacific response here And I'm looking at the winter time and for the two models to meter air temperature And so to I so this figure I showed before show the That we have these PNA PDO like response This is for the CM 2.1 model And we wanted to know what part of the AMV plays the largest role in driving these anomalies And so we repeated the experiment by taking only the tropical part of the AMV pattern and another in another experiment where it took the Supologia part north of 45 north and these are that the response we get in terms of temperature and As your potential height at 500 millibar in color and this is a stream function in the upper troposphere in contour And so clearly we see that When when it's driven by the tropical AMV we are able to reproduce most of the PNA Pattern, but that there are still some teleconnections even when there is there's no Tropic there so but clearly the tropical and part of the MV is the main driver of the specific response This is the same for the CSM model and while the response is stronger in general the same story holds Okay, so what's the mechanism how the Atlantic anomaly warming are communicated to the Pacific? Anomalies, so we say that there is basically a bridge between the tropical Atlantic to the tropical Pacific and then to the Extrotropical Pacific and the mechanism is quite similar to what has been recently described in a number of paper recently published basically the tropical warming in the Atlantic leads to enhance Walker circulation and drives West Easterlies in the in the Western Endo Pacific that these that are associated with a warming and through these changes in the wind through West West feedbacks lead to a Cooling in the eastern part of the Atlantic and and hence trade winds in the eastern Pacific that drives La Nina like response. I forgot to say what I was showing here I'm sorry This is basically the time evolution of the response from the first winter to the second winter in the CM 2.1 experiment where we put this full AMV So how basically we get the El Nino like response in the in the Pacific by I'm applying this AMV type anomaly So it's really very similar to what has been proposed by by by these studies here So do we need ocean atmosphere coupling to get this PDA? negative IPV PNA response To to to answer this question We did some experiment in addition to the full AMV that I showed before Basically, I only show the same 2.1 model here. We only did them with this model And so what we did we in addition to nudging the chop the Atlantic to the observed AMV we and before we had the whole rest of the globe fully coupled and here in addition we nudged the rest of the globe to the Models climatology in in in this row and the the lower panel show the nudging In the tropics. So basically here We also nudged the whole tropic to observe climatology So we kill and so we kill all the tropical the coupling between the ocean atmosphere in this region and and and here we do it globally and interestingly we see that when when we kill the viability in the tropical Pacific the There is still a PNA like response PDO like response But the magnitude of this response is weakened by about 35 percent So clearly the the the tropics the the the coupling between the the ocean and the atmosphere is key in the in the tropical Atlantic to get the right magnitude and and the extra tropical Pacific also plays a role in enhancing these anomalies So it's really both tropics and extra tropics to that give the full response there Let's move to the North Atlantic now The North Atlantic response is more noisy more difficult to to find here. This is for the two models in For the winter time. So we expected to find a negative NAO in response to this AMV as Documented by a number of previous study and if we look at sea level pressure the CSM model shows some kind of a shifted NAO here negative NAO the CM 2.1 shows something much less clear and if we look at the geopotential height It's actually dominated by a thermodynamic response But we have something also not not very clear there and so the question we ask is is it possible that given our strong Pacific response? Yes That this North Atlantic direct response might be polluted in a sense or just Changed because of the teleconnection from the Pacific and so if if we look at the At the at the anomaly is only driven at the response Sorry only driven by the superlor gyre part of the AMV the two models show much bigger agreement and here It's clear a dipole that is maybe show some resemble with them a mode like the shifted NAO Doesn't really project on the NAO, but it's very similar in the two models and if we look at the Anomaly in the Atlantic for this geo potential height at 500 millibar driven by the tropical part of the MV They have quite opposite sign at least over Europe and high latitudes between the superlor gyre So the contribution of the superlor gyre and the tropical Atlantic are have Opposite and kind of cancel each other in terms of the Atlantic And so our hypothesis is that it's the PDO signal that is actually Weakening that direct North Atlantic response And if we look in the CM 2.1 model the regression of the geo potential height associated with the PDO index It shows some anomalies that are quite similar to what we find when we only Drive the model with the tropical MV and indeed when we nudge the tropics And so we remove this really strong influence Through the tropical Pacific we have a geo potential height response in the North Atlantic that is much stronger So showing that clearly the teleconnection From the tropical Pacific is influencing our North Atlantic response So if we let that aside what drives this North Atlantic response So the direct response so to to show it better I removed here we removed here the zonal mean To highlight the dynamical response and not just the time thermo dynamical response And this is the 500 millibar geo potential and the upper troposphere response at 200 millibar and the control show the Upper troposphere zonal winds and so the zonal winds show a Rosby wave train that originates from the tropics and moves toward the extra tropics That's clearly this very classical response to tropical Atlantic SST the response is barotropic in this case and they're also a role played by Storm track the transient eddies tend to reinforce this anomaly in winter as shown here by store eddies Storm track activity. So the eddies the climatology is shown in contour and the Shading show the the response so we have a southward shift of the transient eddy activity in the Atlantic that is consistent with This negative like an aolike anomaly. So both tropical Atlantic and supologizer anomaly here tend to contribute to the North Atlantic response If we try to go back to that first plot that I showed from George here This is this estimate of a fraction of decade of variance explained So how much of the variance are we explaining in in this model for temperature and geo potential height? So we tend to explain not much over land Unfortunately, but still the variance explained can go up to 10 percent 20 percent in some areas And we were quite surprised to see the differences between the two models So this seems to be quite different noise and variance among these two models despite the same similarity of their response So I'd like to conclude by saying that the MV drives global impacts in temperature precipitation and Sea level pressure that quite similar among the two models We were quite surprised by that. That was a nice surprise There's many differences in particular at high latitude in terms of the response in the Arctic mainly Over the Pacific the observed a MV drives and negative IPO IPV like response and the tropical Atlantic appears to be the main driver of the specific response and the Mechanism through which the tropical Atlantic influences the Pacific in both changes in the Walker circulation Comparable to what has been documented by previous study and extract tropical coupling in the Pacific also plays a role I and that was shown also in previous study So the Atlantic warming we show that it Increases the frequency of having La Nina like events. I didn't show PDF here, but we find a shift in the PDF And so AMV positive AMV tend to lead to more La Nina events in in these models The North Atlantic response to a positive AMV also gives a Significant response in the Atlantic that is weak projecting something that resembles a negative NaO, but it's shifted We don't really understand this pattern yet This North Atlantic direct response is changed by teleconnection from the Pacific And this is something to keep in mind when we try to identify signals in observation and In in models also in control. So maybe that partly explained why we have such a hard time detecting the North Atlantic response and Also, I would say that tropical and extra tropical anomaly Must coincide to give you a significant modulation of the NaO So as perspective and challenges we were asked to to give you to say a few words about what what what would be the perspective we see So I'm asking really the question are these climatic impacts due to the MOC I pre-on we isolated the internal part of the AMV motivated by the fact that we wanted to identify the Climate impact of the MOC, but really we're not sure that the surface signature really Correspond to the MOC signature in particular We show here a really big role for tropical Atlantic part of the AMV and all the models show a very weak Signature of the MOC in the tropics. We don't know if this is true or if this is a model deficiency So but it's really an important question. We need to answer the tropical Atlantic also that is key As I said is badly represented in most couple model and here also This is problematic if we expect to get a prediction to arise from that region Overall, we find a weak signal-to-nose ratio and few impacts over land Even if there are a few they are significant, but remember that we had quite many members there So we have repeated this experiment using a 50 kilometer atmosphere coupled to an ocean so quite high resolution and the the signal doesn't increase much over land So it's not just a matter of resolution I'd like to highlight quickly some limitation of the protocol we adopted here, but Christophe will discuss this more on Friday We proposed this experiment for the DCPP meep the component C at this fixed AMV pattern is With this current protocol is okay But this cannot be applied that way for pacemaker experiment because when we restore the model to surface surface temperature There are issue that arise and models that drift and so Christophe will talk more about that And this is a very important point. So here really are we only isolated the influence of the Atlantic on the Pacific But clearly the Pacific and other region can influence of the Atlantic as well So we need to reconstruct all these teleconnection together and see what's the ad up the the response at the end And which one is predictable. Thank you