 Okay, thank you so, yeah, we we started this analysis from a Mediterranean angle to a regional perspective, but it became quite sort of immediately clear that the implications of this investigation had a sort of much wider sort of relevance also for the you know the North Atlantic domain, so I'm gonna use this kind of The the Mediterranean of the sort of Trojan horse to get into the Atlantic domain, so to say This is not surprising because we know that the Mediterranean Multidicada variability is sort of enslave it to what happens in the neighbor Atlantic domain And so there is a vast body of literature showing how I mean correlated is the low frequency fluctuations in the Mediterranean area with the Atlantic Multidicada variability particularly showing for the sea surface temperature evaporation fields and also temperature of the air temperature over land So that this evidence somehow points to the predictability of the signal We I mean is this that the question is is this really predictable and how I mean if we look at the When we try to establish how predictable is this sea surface temperature variability By using for example, see me five decade of predictions This is a picture from a work by Virginie Gemma and the recent recent work showing What is the added value of initialized versus non-initialized, you know Simulations and there's our differences in the anomaly correlation and Telling you what is the advantage of in, you know climate predictions with respect to most standard historical sort of uninitialized Simulations and you see there are some differences the yellow patches show where there is this added value But this is relatively small is significant is there you can see it's quite confine it to some sub areas of the med-sea and in particular in the Levantine Mediterranean Eastern Mediterranean Basin But to some extent this is also telling us that most of the Skill the predictive skill is already in the uninitialized simulation So we can get a lot of predictability even without an initialization. So it's telling us that That there's some important role not only from the you know internal variability of the system the uninitialized component, but there's also Some important role potentially for the four things And so this led us to the major question of what is the role of the forcing by forcing I mean I'm meaning here nature and anthropogenic force things and then what is the role of the forcing on the overall predictability We see in the in the med-sea region in the Mediterranean area So to tackle this issue we decided to use a sort of hierarchy of different historical integrations Ranging from more standard historical ones that we know of very familiar with the The ones which include the whole set of four things nature are plus anthropogenic and but we also try to decided to use another typology of historical simulations using the so-called heist Michelin heist misc Sort of integration. This is a type of a subset of semi five historical integrations where Specific forcing had been used like either using individual anthropogenic for things or specific, you know natural forcing such as solar Or volcanic and so forth and I need to give credit for collecting this interesting set of Simulation to the NASA group in particular who made a lot of work to collect this in from this kind of coordinating this sort of Me in the semi five and so we're going to use also this historical Michelin is with either Anthropogenic only just to I mean a disentangle. What is the relative role of anthropogenic versus, you know, natural and anthropogenics together and and finally in order to further narrow down our attention to Which particular anthropogenic might be we also used another subset even narrower subset of models where Exit essentially all of four things are there natural plus anthropogenic except for the anthropogenic aerosol the so-called no a Simulations and just to remind us that we know very well. What do we mean by natural? Do what do we mean by anthropogenic natural here include no canals plus solar variability and anthropogenic is a blend of different Four things going from greenhouse gases anthropogenic aerosols land use changes And Okay, so this is the size of the ensembles that we've been using and so as I say historical simulations Provide quite a large body of simulations in here We decided to stick to those models which had provided Ensembles of a historical simulation not just one individual single realization by the number of it And we decided to stick to those which had provided a minimum set of three members And so this led us to select only 13 models there are many more But of course at the models which had provided the ensembles had them a lot and a smaller set of the whole set and this Amounts to about 69 different Historical realizations the his misc with only anthropogenic four things is a still I mean quite large to some extent It's 53 different ensembles Sorry ensemble members overall Divided by 10 different models and finally this is a much smaller set unfortunately of a historical Michelin use Four things without anthropogenic aerosols, which is an ounce to 14 only three models did that But and that and that is the sort of sites that we had to deal with but as for a historical and Anthropogenic only it was quite a reasonably large set of integrations Now if we look at the other historical simulation, this is This is the sea surface temperature record for the 20th century actually it goes from 1870 up to 2010 and the black curve is the observations and the gray cloud is the whole set of historical integrations This and the red curve is the multimodal ensemble average made across the whole set of you know The grand ensemble of all the force all all the force of simulations and what do you see here? Well, first of all, you can see that the cloud is encompassing the observations so that when you you Did the models are somehow able to capture? I mean the variability of the real variability of several variability is included in the cloud But most interestingly that the red curve is the multimodal ensemble mean and after you average over such a large number of Simulation you expect this signal to be essentially the force of component But averaging out the internal variability of individual models And so the interesting point that you can see here is that After averaging over such a large ensemble of integration You see that you see that it's quite a good particular for the second half of the 20th century there's a reasonably good correlation with the observations and Particularly if you look at this sort of a worm to call transition occur from the 50s to the middle 70s There is this declining trend which is still there after such large ensemble average But you can also notice that there is a sort of mismatch in the face and the models in a rather systematic way tend to anticipate this minimum by a Few years with respect to the seven ones Whereas not such good agreement is in the in the in the first part of the record actually there is some good agreement in the late 19th century, but then it becomes a bit more elusive during the early part of the 20th century So when we try to establish what is this partial structure Associated in particular to this particular cycle we decided to focus on on this specific a semi cycle of the all and you know AMV like Variability and in you see if you compost it, you know across the worm face Sorry, I can't see okay if I compost it across the worm face about 20 years from 1930 to 1950s minus the cold face of this semi cycle and the the multimodal ensemble mean sort of composite is a pattern which shows of quite a familiar structure with this comma-shaped Structure as a polar Basin amplified response and the Atlantic is clearly part of it and it's eastward On the on the European African seaboard we have this sort of comma shape And as a cold lobe over the subtropics if you do the same exercise for the ops You see a sort of a grossly Consistent pattern again. We need to bear in mind that this is not the whole AMO This is not the AMV pattern as we are used to see it because it's only about one specific semi cycle of it So there's a lot of also noise and noise. It's not noise. This is I mean since this is the reality in a way We cannot filter out the internal variability from the first one but this is only the first component and Even after removing so the internal variability the first component bear some good consistency with the observed pattern So okay, so the question is that I mean having said that of course historical Forcings include both natural and anthropogenic Forcings and the question is to what extent is this that give due to the some natural forcings or due to some anthropogenic Forcings and so in order to address this question, of course We moved through through the hierarchy of models from purely historical to sort of idealized Anthropogenic only for things so this is the so I'm invoking here the historical MISC a set of simulations With only anthropogenic for since so everything which is anthropogenic is there basically greenest gas aerosols So it's on land use and so forth and Well now the issue here is the following we we can clearly see each of these two curves And the blue one and the green one are Multimodal ensemble mean themselves and they have been obtained from a fairly 50 50 partition in this Population of models about 27 model not there are 27 models in the blue one and 26 in the green one And what is the difference between these two clusters? So clearly this is a clusterization of the behavior of the models under anthropogenic conditions And and the clusterization is such that the green one clearly reproduces this SSD dip this minimum during the middle 70s and then the temperature after that grow again and Whereas the so-called anthropogenic a is at the blue curve the temperature just monotonically increase Although there's a clear change in the pace of the growth as the trend changes and in the in the latest part of record Accelerates quite consistent with the other ensemble But it is clear that we have an issue here as not all the model Despite the fact that all the models have been forced presumably in the same way in the consistent way This is a protocol Then yet there is a large model to model uncertainty But I can anticipate that this comparison is only partly fair because actually the blue curve is strongly biased by one specific model There are in terms of the model Interdependence this population are quite diverse in that the blue one is quite Affected by one specific model which has been run under very different configurations But the model interdependence in the blue one is strongly biased one specific model I'm not going to say which one is not interesting or if you want I can tell you but I mean this is just to say that the green Curve is somehow representative of more diverse zoology of models Right so and if we overlap the historical forcing some of this one So did this is the red curve that we were looking at before in the early earlier slides? And you can clearly see that the historical and this particular subset of anthropogenic for things Quite overlap quite well. Thanks for the in the second half of the century and you can see that Also another feature that you will notice only once you have a lot of observations as well as that under anthropogenic You still don't reconcile this mismatch in the phasing of the minimum of this particular warm to cool transition So anthropogenic and the historical are quite consistent over there And this is suggesting that is this actually I mean aside from this consideration in the model uncertainty There's a quite a good. I mean agreement between One particular subset of anthropogenic simulations and historical ones that which lead us to believe this is According to this picture. This seems to be really anthropogenic early driven at least this particular Part of the cycle we cannot say we cannot make such a good attribution for the earlier part of the record And we see not obviously the anthropogenic and don't seem to be playing any role as expected in the earlier part of the They say the earlier part of the record But not even the historical ones are capable to do that Except that you can see that there is some influence in the volcanic From you do the volcanoes in the historical which is absent of course by definition in the anthropogenic ones because the novel canals are there by construction So right and then we look at the spatial structure of this and again of this Warm to cool transition we show again the composite for all set of models that we've been looking at so far And you see the the green frame at the Plot is the one referring to the so-called anthropogenic be the one show in the cooling and the Anthropogenic a is the one of the models that just the temperature keep on increasing and you can see is that the green models Yeah, I'm coming the the green models Patterns sure is a it's again is the comma-shaped Pattern comes out again with a subpolar amplified response and the Atlantic part of this Large-scale structure and the negative lobe of this are tropics whereas anthropogenic a is the models Which do not show any such as the cooling and that the temperature just keep on increasing and that of course the pattern is totally different So, okay, if it were anthropogenic what kind of anthropogenic would that be and then I'll pay on a very purely intuitive basis You wouldn't think that the greenhouse gas could determine such Let's say how high at use of the middle 20th century and to make a fair comparison We find we decided to use this and no anthropogenic aerosols those models that we could rely upon and Compare those to the historical ones But only to those models corresponding exactly to exactly to the no a ensemble So that we just stick to those three models and we confront to the historical versus no anthropogenic aerosol And you see that after you remove anthropogenic aerosol. This is the green The green belt of the mop In the in the picture the the temperature increase and that don't show any higher to us anymore So it looks like the effectively the anthropogenic aerosol have played a role in this sort of inversion in the in the trends of the temperature Finally, I will skip conclusion. This is just kind of summary I mean we that the emerging picture is okay This is this nice cycle is the a and v like cycle appears to be somehow as a Fertuse not fortuitous, but it looks like more a patchwork of different. I mean sources of variability and predictability and If we cannot attribute the earlier part of the record to any force of response that clearly internal variability is playing a role There there's also concur from the volcanoes So I would say but the second cycle of it the second part of it seems to be much related to the anthropogenic aerosol and this lead us to the kind of Frankenstein a and v like structure were different You know, we're merging together different things, but overall contribute to Create this is nice sinusoidal like picture and I will stop here. Thank you