 So, we have another exciting talk about teleconnection in general with this book. So, I'm this talk, I'm going to focus actually not on all teleconnections, but specifically on two teleconnections. And one, the first one is actually the probably the first teleconnection that has been studied and found which is the teleconnection between El Nino and the PNA or PNA-like pattern. Basically linking the anomalies, SSTs in the equatorial pacific, the warm anomalies, SSTs in the equatorial pacific during El Nino with the circulation on the North America and the pacific North America. And basically this is in the figure you can see, this is a map in the DJF and there are edi SST between 1880 to 2011 and it's a special composite taking the Nino 3.4 anomaly and compositing with the SSTs all around the globe, the surface temperature and the mean sea level pressure that you can see in the contour. So this is a typical pattern of linking these anomalies on the pacific with what happens in the circulation and in the surface temperature. So one can, the first question can be, well, when we have climate simulation, basically one of the way of assessing the ability of the climate model to reproduce the correctly or satisfactorily enough, the climate is just to assess how well it is able to represent the teleconnection. And actually this is an example from one ensemble member from integration done with the ECE Earth model. It is an historical simulation more or less in the same period as in the other ISSTs and we can see that actually in this case, 130, 140 years, basically the teleconnection is very well reproduced, we can see the centers in North America, the anomalies in the temperature that are quite well simulated and probably there is some mismatch, something quite different but in the signal over the Eurasia. So let's say we have one ensemble member with enough years and one can say that, okay, it's quite well reproduced. However, in this case we have more than one ensemble member and the other ensemble members which are totally equivalent in the sense that they have the same model and the same forcing. And once, for example, you look at these other members from the same model and you can see that there is already some differences. For example, the surface temperature anomalies in the eastern, northwestern of North America is quite different and there are also other features that are not actually the same. And looking at other ensemble members, you can see that there are all them are slightly different and, for example, in this case that I wanted to point out is number five, actually basically we don't have any more the teleconnection. We don't have the PNA teleconnection in the mean sea level pressure, almost disappeared and actually it's an equivalent ensemble member with the same condition. It's a couple integrations with the same model, with the same forcing and it's not even a short integration because it's 140 years integration. So basically, my point is what we should expect in terms of reproducibility. It's what we should expect, what is the natural variability? Is the natural variability in 130 years of observation well sampled or there is something more? Or rather are the models which exhibit too much variability? They are too sensitive to something that we don't know. This is one initial point and it's about this El Nino PNA, but I will warn you that I won't show any result from couple models because I wanted to keep it simple. So basically what I'm going to show you later on is some results from AMIP integrations. So the AMIP integration will reduce the degree of freedom because the systems are always the same and there are only ensemble members with only basically the atmosphere which is different. But the forcing coming from the radiative forcing which is basically CMIF5 and the forcing coming from SSD is actually the same. So that's the first teleconnection that I try to speak about and the second is something different is let's go to the Atlantic and this is Atlantic multi-decadal variability or Atlantic multi-decadal oscillation and which is here you can see from the ADSSTs, these oscillations basically the AMIP index which is the AMV index is obtained taking the early anomalies of the North Atlantic SSDs in the northern part of the North Atlantic and subtracting just to get out of the climate signal of the trend over the SSDs basically the 10 year running mean of the global SSDs. This definition is well known definition by Trambert and Shear in 2006 so this is the oscillation and over there we can see the map how does it appear in the North Atlantic, the anomaly of the SSDs in the North Atlantic. In a recent paper in 2014 on environmental research letter, Pengs and Magnus Dottir, difficult Swedish name to pronounce for Italian actually showed they look at the sensitivity of the Euro-Atlantic weather regimes to the AMO and they took reanalysis, one of these century reanalysis in particular they took the 20th century reanalysis by the American 20th century reanalysis and they looked at the variation of the frequency in the Euro-Atlantic weather regimes which are the blocking, the Atlantic ridge, you can see here in the two phases of NIO plus and NIO mines and basically they looked in winter and also splitting the winter in the early winter and the second half of the winter and they look at how the frequency of the Euro-Atlantic weather regime is modulated by the phases of AMO and actually they found that for example it is in the case of the blocking and the two phases of NIO there is a sensitivity to what which is expected actually because there is a change in the SSDs and the frequency of the weather regime in particular the blocking frequencies and negative NIO frequencies increase with AMOV positive with the warm SSDs and decrease with the NV negative and for the NIO plus is the other way around also they looked at a model simulation that is with the CM5 and to see if the model was able to reproduce this behavior and actually what they found that they didn't find anything for the blocking and probably because as many models the blocking is not well reproduced or simulated in climate models but they found a similar sensitivity for the two phases of NIO plus and minus so basically and also we, my colleague Paolo Davini, Joss von Hunderberg which is in the audience and myself did an experiment with this year in version 3.1 sensitivity experiment with having only changing basically the SSDs over the North Atlantic and with positive and negative SSDs over the, anomalies over the North Atlantic to see in MIP mode so MIP sensitivity experiments to see if there was, they say we found if we could find the same signal as the one of Pegmus and actually this is as far as the blocking is concerned and this is a bidimensional blocking indices for the reanalysis in the contour and instead you can see in shading the difference between the control era interim and the reanalysis showing that actually the model is Earth sorry, reanalysis Earth and showing that the model actually as underestimated the blocking over the Pacific but as a good sort of representation of both of the Greenland blocking and for the Scandinavian blocking so this is the model and then we can see here the sensitivity to the North Atlantic temperature so to the MV and in this case when all the North Atlantic from 0 for 5 North to the 70 North has been forced with a positive anomaly and one can see here that we have more blocking basically in the Greenland and over Scandinavian and UK which is consistent with what was found previously and also in reanalysis what is in this paper interesting is that if instead of forcing all the North Atlantic with a positive anomaly and a negative anomaly similar to AMO we take, we forced only the tropical part of the Atlantic so basically keeping the climatological SSD on the North Atlantic actually we found the same, exactly the same singland so in this paper we conclude that we try to found the mechanism which is different from the one of Marlborough this is one forced only the extra tropical basically there is nothing but this might be a feature of the model of this year so said that in this presentation I would like to try to investigate these questions so first okay I wanted this teleconnection pattern so basically this sensitivity to AMO and this PNA El Nino teleconnection is reproduced by the model and the second what is the degree of reproducibility of this teleconnection patterns in sister simulation so basically in different ensemble members and last if there are some factors that can in a way weaken or strengthen a given teleconnection pattern so basically if there are external factor or maybe noise that can in a way strengthen or weaken the teleconnection patterns and what the experiments that we use is a part of a big experiment which has been conducted under a praise grant and the experiment has been led by Jost and has been conducting collaboration also with Oxford University and in this experiment which was aimed actually to investigate the sensitivity of a climate to the model resolution and the stochastic physics in the model as a lot of the integrations most of them MIP integration and a few couple integrations here you can see well basically a summary of the experiments so we focused on a 30 year in the present day and there are also integration the same integration for future scenarios ACP 8.5 there are five different resolutions for T255 we have ten ensemble member with or without stochastic physics and for the higher resolution of course the number of ensemble member that you were able to reproduce decreases and at the highest resolution actually we perform only one integration with stochastic physics for the experiments and actually it is the data set of this experiment are available are in a threat server and can be used for research and we are happy these people they sort of wanted to study something using this data set here I just focus on this small part of the experiment so basically the ten ensemble member without stochastic physics all the same in T255 MIP so first question how well is here MIP simulation are able to reproduce the observed weather regimes on the left you can see the weather regimes that I showed before and they are basically the first one on the left the positive NEO the blocking the Scandinavian blocking the Atlantic ridge and negative NEO has in the end separate analysis in an intermediate absolutely the same and here is how they are reproduced in each year T255 I it's not to return here but the special correlation ever has a special correlations between these patterns and the patterns in an set is very high is more than 0.95 and so we can be really satisfied for this simulation and here one can see basically from in the years how is in we can see here the frequency how is the interannual variability inter-ensemble variability and basically one can see a lot of noise however now we wanted to look at the sensitivity of the frequency of these regimes to the Atlantic to IMV of course in this period because our integration are only 30 years long we don't have actually an oscillation of AMV but there is just we are on the ascending branch of AMV so it's more or less like a trend but it's not a real trend because basically the MV index per se has been the trend so has been computed in a way that the trend but we are just looking at this so basically wanted to stratify the frequency according to the phases of AMV and this is how AMV is like so basically this is the correlation between the AMV signal with the all the SSDs in the period okay so then now the result for observation when you take the observation and then you stratify when you look at the how bar is the frequency of the regimes according to the MV one can one find essentially what has been found already by Panksen-Maxunottir so basically and here you can see that basically the triangle is AMV plus and the dots are AMV minus so all the regimes excluding the NIO plus increase the frequency during AMV plus and the other way around with NIO plus so this is what we found in the analysis and now you wanted to see what about the model is the model as sensitive as the analysis and of course since we have a 30 year of the reanalysis basically we look separately at each one of the 10 ensemble members so here are the four weather regimes on the X axis you can see the number of the ensemble member for 0 to 9 and here you can read the departure of the average frequency for each ensemble member for AMV positive and negative so basically let's focus on NIO plus and what we expect is actually that having less cases of positive NIO during the AMV positive and all the other way around and one can see there is a lot of variation between the ensemble members basically some ensemble members like ensemble member number zero or number seven are very sensitive so basically here you have a 6% in difference in frequency and other ensemble members are not sensitive at all basically nothing happened and all the ensemble members are the same and there are 30 years and they are forced by exactly the same SSDs so in principle well I don't know what someone else would have expected but I myself I would not have expected such a variation or having some ensemble member totally unsensitive the other ensemble members are sensitive if one look at the other regimes it's even more confused and just to make things slightly more clear I would just put together sort of the blocking regimes so there are NIO negative and Scandinavian blocking in one bunch and here is the Atlantic region basically here one can see more or less the same behavior as in the analysis only that there are some cases which are very sensitive and some cases which are not so sensitive so basically what the now the question is what are the possibly possible candidates that can amplify or inhibit this sensitivity to the ANV and we look to the positive an ovalee or Asian snow depth that if are in phase or out of phase with ANV could possibly amplify or inhibit the sensitivity to the weather regimes another possible candidates is the temperature in the stratosphere basically the temperatures in the stratosphere we know that stratospheric warming events can lead to NIO negative so if we have more stratospheric warming events during a period in which also from SSDs we have sort of a push towards NIO minus then we can amplify the signal or the other way around so the idea is just to look at these two possible causes just please the ensemble in five good ensembles so the most sensitive to EMV and five ensemble members which are the least sensitive to the ANV and just look at the difference between between in between them in terms of snow depth and just to be simple I just took the temperature that 50 hectopascal as a measure as just a rough measure of the temperature of the stratosphere and so this is what we found for the Asian snow depth basically looking at the snow depth in this area of Eurasia which has been shown as very important for especially at the beginning of the winter in triggering NIO negative I just look at these are all the month of the year and they are different between sensitive minus nonsensitive cases and actually in the case of positive EMV one can see that there is a positive anomaly in the winter of the snow depth could have helped to increase the sense the sensitivity to AMV plus but actually it seems not having much effect on the sensitivity to AMV minus if I look at the instead of the stratosphere basically this is the average temperature in the different month of the years for the five good ensemble members minus the five bad ensemble members and one can see that in this case both phases of the EMV actually they are sensitive and basically one can see that in the case of the EMV minus the negative anomaly of the temperature so which is basically related to having colder stratosphere which has more related to have an intensification of the vortex so basically positive NIO which which when this is in phase with a negative phase of EMV can amplify this signal and the same on the other way around for the NIO negative so I have only five minutes now quickly I wanted to look basically at the other teleconnection which I said it's Pacific North American teleconnection and one can have here the observation these are the correlation between the frequency of this cluster which is the Pacific North American cluster number one which is very similar to a positive PNA actually it's more a Pacific trap because we have this trap the Pacific and with respect to the positive PNA all the structure is really shifted basically on eastward and if one look at the frequency during the winter of this regime and that's the correlation with SSDs basically one finds this pattern in the observation and this can be sort of a measure of the teleconnection between the two and so what happened in the model first of all the model able to reproduce exactly the same regimes well the answer quickly is yes you can see that the pattern is shared in this and set over there so are quite similar and now we will look if in this case so basically in the model case we can find this the same pattern as in the observation and actually it depends so basically here we can see the same so these are the patterns in observation which we saw before and these are the same so remember that SSDs are the same in all the in exactly in all the ensemble members in all the integrations and that these are different ensemble members and one can see that in some cases it's really good for example let's compare the observation with ensemble nine they are practically identical there is really really very good correlation but in other cases for example ensemble zero we don't the teleconnection is really really very weak and well we can see different flavors of that and that it's a bit was surprised well I was in this case really surprised because basically this is one of the connection that is really direct we have basically these changes in the rugby way sources and then these waves is that exciting the P&A or the P&A like pattern yes they are Amy Pransel the SSDs are exactly the same in all the runs so basically it's something that well in principle you don't think about respect well I would have respect something really noisy in the Atlantic but not in the Pacific so let's look for some candidates of this difference and again the only thing that I could think is something related to the stratosphere if there is or and there are papers I here I I I put only in the zone in scape 2009 but there is a can you have the manzini in the same year just looking at the influence of the stratosphere warming basically on on during El Nino amplifying or not the the response to El Nino in the mid latitudes and what in their case in their model basically they found that stratospheric warming they help to have a better response to El Nino and a more recent study which is in 2015 Richter et al this group in Anchor a claridizer group I think they look at the Amy Pransel with CM5 and they look at the signal with stratospheric warming without stratospheric warming and they found almost a slightly a slight signal in the model but I would say in my view that the signal in the observation is the other way around so basically having a better answer in in this case the answer not the answer is the response sorry the response in this case without stratospheric warming than in this case with stratospheric warming while they claim that the observation are are are not there are not many cases of course but then let's look in in the case of this year so basically what I did it just as before the five more sensitive minus the five least sensitive and looking again at the the Nino winter Nino winters and basically one can find that this is a signal which is not significant however if one takes the best one minus the worst one actually there is a signal which is this red line pointing towards a better teleconnection with a colder a temperature in the stratosphere which of course should be studied better because one ensemble member is only 30 years and one should look at something well and if one look actually at the teleconnection as they are and these are the maps are similar are the ones that I I showed before and this is the the the best ensemble member number nine and over there there is the worst and the one can see that and the SSDs are obviously the same in in this plot the surface temperature is not the same so the temperature over the the the land is different but just focusing on on the signal of mean sea level pressure basically the signal is much much stronger of course in in in this case and interestingly enough the signal over the Euro-Atlantic sea is totally totally opposite totally different so and I leave it the concluding remarks if someone want to read it and I think I am most of time