 I just wanted to say hello and good morning and thank you very much for inviting me to this workshop and I'm really sorry I couldn't be there in person and I also wanted to convey my solidarity with all of you and everyone affected by the tragic events in Paris. So Johanan asked me to give an overview of Decadal Climate Variability in the historical record. Is my talk up now? Sorry, we're just getting the slides up. Okay, so I'm going to keep my remarks very simple and the hope is that I'll make a few overarching points that will hopefully be useful as you delve into much more detail over the course of the workshop. So I'm just trying to put us on the same page and make a few simple points about the challenges to identifying and understanding Decadal variability in the observed record. So you've already heard some of this this morning. I had a chance to listen in on a few of the talks using the live streaming which is working very well. So first motivation, of course, why do we care about Decadal climate variability? Of course, there are profound societal impacts. We've seen that it adds uncertainty to projections of climate in the coming decades and we've also seen examples of how it confounds the detection and attribution of past climate change. So plenty of motivation. So I'd like to highlight some of the challenges to defining and understanding Decadal climate variability in the observational records. So I'm just talking about the last 150 years or so when we do have instrumental data. So the first point I'll try to make is that these patterns of Decadal climate variability tend to be global in scale. And that can make it difficult to sort out the causal linkages between different regions. So I think that's a real challenge that we confront. Then secondly, of course, the data are sparse and the records are short compared to the time scales of interest. And then finally, because of this, it may be very difficult to distinguish Decadal climate variability from a random process, a white noise or a red noise process. And just to caution everyone, when we look at records, especially short records or even the paleo climate records, and we low pass filter the data, we're always going to see Decadal variability. But this may not be physically meaningful. It may not reflect a physical process that's operating on the Decadal time scale. So caution when just assuming that even though a low pass filter time series shows variability, it doesn't mean that there's any physical underpinnings. And then finally, I will go through what the main phenomena are of Decadal climate variability. And I'll split this up amongst the three main ocean basins. So the first thing is to bring us back to the actual data coverage with which we have to look at this problem. And of course, the oceans are primary in Decadal climate variability. So I'd like to show you a few maps of data coverage from the ICOADS data set, which is primarily a compilation of merchant ship data from merchant ships that we use to look at this issue. So these are some maps of the showing the percent of time or the percentage of months in each grid square that had at least one observation in each of these 20-year periods. So the upper left that has the black box around it, that's for the 20-year period from 1860 to 1879. And you can see these ship tracks very well-defined, mainly in the Atlantic and also in the Indian Ocean, South Indian Ocean, but really very, almost no coverage in the rest of the oceans. So really the Atlantic and part of the Indian Ocean is all we can look at so far back in time. The coverage gets a little bit better as we step through the various 20-year periods, but again the Atlantic remains the best measured of the ocean basins. And even from the 1920s on, we can see filling in of data for the northern hemisphere, but the southern hemisphere and the tropics also remains under-observed or not observed as well as we would like. And then finally the most recent 20-year period, things are pretty well-covered except for the southern ocean. So this is a basic constraint and of course these data go into all of the reanalyses that we use, for example the 20th century reanalysis or the soda ocean, simple ocean data assimilation, but again we're fundamentally constrained with where the data are from these ships. I'd just like to just give you a pointer to a resource, the Climate Data Guide, which can provide information on climate data sets written by the people in the community who have expertise in particular data sets. And we, I think as a community this is a very useful way of compiling our knowledge of these data sets to make use of in our studies. Okay, so defining patterns of Decadal Climate Variability, I'd like to just start very, very simply and show that you don't actually have to do anything too fancy with the data to see variability. So again I'm just trying to put us all on the same page, I know we come from different backgrounds in this workshop. So we can look at a map of sea surface temperature. This happens to be the one from particular day 29th of December in 2011. And when you look at the map you see the usual patterns with changes with latitude and then you see the warm pool in the western Pacific and Indian Ocean, but it's very hard to see anything beyond the mean seasonal cycle. So of course we subtract the long-term mean for this particular day and when we do that then patterns emerge. So these are the anomalies for this particular day relative to the long-term mean. And we can see large scale patterns, there's large scale organization even on a daily map of sea surface temperature. And I've just highlighted here in the box a pattern that has been mentioned of course many times already, the Pacific Decadal Oscillation. And you can even see this signature of these large scale patterns also in the Atlantic even on a single day. And I think this is really important to be able to see these patterns without a lot of statistical analysis. So moving on, looking at the Pacific Basin and Decadal Climate Variability in the Pacific Basin, I'm just going to show you results based on one of the many SST datasets that we have. This is the HAD-ISST starting in 1870 and going through the end of last year. And I'll be showing you just the leading EOF of monthly sea surface temperature anomalies where we've subtracted the global mean SST. And this is the typical practice and I've highlighted some of the studies below where we want to remove some estimate of the forced response and that's why the global mean SST is removed. And then what we're looking at is the variability about that long-term warming trend. So this is a familiar pattern I know to probably everyone in the room, the leading EOF of sea surface temperature anomalies in this box over the North Pacific. And we're showing it as a global regression map of the SST anomalies regressed onto the principal component time series of the data in that box. And this mode explains 25% of the raw monthly SST anomalies. So it's a leading mode, well separated from the second and higher order modes, but again it only explains a quarter of the variance so there's plenty of other patterns that are also contributing to North Pacific SST variability. And it's very nice to see this correspondence between this pattern that's derived from 100 years plus of data with this map for a single day. And again that can give you confidence that the statistical analysis really is rooted in the patterns that we see in the data. So of course it's very important to look at the sensitivity of patterns to the domain of interest to variations in the domain. So if you look at the EOF using the full or pan-Pacific domain, you see a very similar pattern and explains maybe a little bit more of the variance. But you see of course that the region in the Eastern Equatorial Pacific really dominates in terms of amplitude for this pattern. And this is of course the ENSO pattern with its teleconnections to remote regions from the tropical Pacific. And there's a high spatial correspondence of course between this Pacific Decadal Oscillation or ENSO-like pattern with the ENSO pattern itself except for this different relative magnitudes between the tropics and the extra tropics. And this PDO pattern is very similar to the inter-decadal Pacific Oscillation which is defined as the leading EOF when you go past filtered the data. But I'm emphasizing here that we haven't done any time filtering so we haven't built in any particular time scale into these analyses. So here are the time series of the principal components of these two EOF patterns. And the one on the top is for this North Pacific region and then the Pan-Pacific region below. And all records go from 1870 to 2014. And again these are the raw monthly SST anomalies. And you can already see that the North Pacific really has a much lower frequency character compared to this Pan-Pacific which is really dominated by the Eastern tropical Pacific. And if you do a power spectrum of these two time series that's shown here on the left and the years are shown on the top axis there, you can really see that the North Pacific is dominated by low frequency variability that is not well resolved in a record that is only 100 plus years. And then this Pan-Pacific record is dominated by the intra-annual and so phenomenon. So now I'd like to turn to the Atlantic. And again much of this has been mentioned before. So again using the same approach of looking at the SST from monthly SST from which the global mean has been subtracted. And in this case we're not the community you can do an EOF analysis but even more simply you can just take the SST anomalies in this box of the North Atlantic and average that and then use that as an index to regress the data onto. And that's what's shown here. So the top map is showing you the box is showing you the region that's used and then the spatial expression of that globally through regression analysis. And then the time series of SST in that box is shown in the lower panel. And again emphasizing these are the raw monthly SST anomalies. And even without doing any filtering there's a very pronounced multi-decadal component. And you notice a strong outer phase relationship between the North and South Atlantic and then a PDO-like pattern that's occurring in the Pacific basin. And if you compare the time series of the PDO, the leading EOF of the North Pacific and inverting the sign of the Atlantic multi-decadal oscillation you can see that there's a tantalizing perhaps relationship between the two. And I think this is also a major challenge that we need to confront. Hopefully there will be many discussions in this workshop. Just what is the physical meaning of a possible relationship between these two indices? The maximum correlation is something like 0.5 when the AMO leads the PDO by about a decade or so. But again emphasizing these are very short records and this lead-lag relationship is apparent maybe in the second half of the record but it's really not so apparent in the first half. So again it really puts into question how robust are these relationships and what and if they are what are the physical meaning behind them. So finally turning to the Southern Ocean which I'm not sure we've touched on yet in the workshop. Here I'm showing you the data coverage over the Southern Ocean as a function of time. So this is the percent of grid boxes between 50 and 70 south over the ocean that had at least one observation and this is broken up by season. And the important thing here is that really data coverage really only passes this 50% threshold in about 1950 and not surprisingly it's really the austral summer season when the data coverage is highest. So really when we look at Southern Ocean as a space what we're really looking at is the summertime first of all and then secondly we really have to be careful about how far back in time we look at the data even in the analyses because there really isn't much coverage before mid-century. And so this is the final map here showing Southern Ocean SST record from 1950 to 2014 again the raw monthly anomalies and then the top panel is the regression map onto this Southern Ocean SST time series shown by the box. And again once again we're getting these global patterns very strong connection with a PDO like pattern and some connection to an AMO like pattern. And finally you can compare the Southern Ocean time series up at the top from 1950 on with the PDO and then the inverted sign of the AMO and again for the second half of the record the decadal variability in all three is very striking and very similar and whereas in the first half I would say that's not the case. So again caution when we're just looking at phenomena that have such long time scales when we really only have very short records. So I'm wrapping up here in my time allotted. I'd like to highlight a very nice book that came out of a meeting a couple of years ago and this climate change multi-decadal and beyond has 23 chapters on a wide range of topics and for those not familiar with some of the topics I highly recommend this book. So final remarks, issues and challenges for understanding decadal climate variability in the historical record and I just want to bring up the three patterns that I've highlighted. Of course there are many more and you can go into much more detail on the types of objective analysis that you do on the data but I think in the end it probably boils down to these patterns that no matter where you start from the North Pacific, the North Atlantic or the Southern Ocean you're actually seeing very much of the same thing and this then really brings up the issue of the global connectivity amongst these different phenomena and makes it in my view quite difficult to sort out the causal linkages between the basins and really maybe we should really think about the global mode or structure of variability. Highlighting again sparse observational data and short records really calls into question how robust these patterns are and then of course makes it very difficult to probe the mechanisms and then just like to highlight that we really I think need to have a view that these patterns could be just an expression of random red noise phenomena with some contribution from deterministic ocean processes but I think both are probably contributing and it's not an either or situation and I think really to make progress as you do this week in the workshop I think it's very nice to see the synergy of combining the observations, the paleo-climate records and the model so thank you very much. So Clara thank you very much for this very clear presentation of the historical record and what it tells us and would you stay connected to us so if we have some questions you can answer. I wanted to say that we can start now the discussion session with maybe more specific questions to Clara and then questions to what else we heard today in the meeting and points that we can carry with us for tomorrow and for the future work on the climate variability. So are there any specific questions to Clara or general questions about the subject? Subjects we heard today. Well maybe I can ask one brief questions to Clara in following up her talk. How much, because Clara emphasized the fact of these global patterns and the fact that they are not foreign to one another that they have some connection between them and I wanted to ask how much could we see that the anthropogenic forcing that we are imposing on the climate is maybe the cause or partial cause for the connection between the different patterns that we were looking at. Is there any reason to suspect something like that or not essentially? Well I guess what I showed I subtracted the global mean SST so in a way I do that as removing the forced component so I was hoping that that would you know take out that element of why the patterns might be connected. Okay so I missed that so I apologize. Yes Ed. Hi Clara, very good talk. I'm kind of curious the similarity in the in those leading modes is quite striking and yeah this is all based on a like an optimal analysis of the observations using methods that I think Alexis Kaplan first described in the 1990s I think they had at ISST data use the same basic methods. Is there any chance that that the method of analysis to create the full field of SSTs could in some sense be pushing the the appearance of similarity in a way that is in some sense statistically kind of ordained. What's your thought about that? Yeah that's a great question and I actually kick myself for not making these maps based on a data set such as the Hadley SST which does not have any filling in of the missing grid boxes so it would not be subject to these this possible issue that you're that you're talking about but I guess if my memory serves and we have made these maps based on the uninterpolated data sets we do find similar connections amongst the basins and I'm not saying that these are I'm just saying that this is what we observe for this very short snapshot of decadal variability that we can look at in the data and whether this is a robust aspect or not I think we really have to use the paleo records and then the models to to really get at this. Rowan has a question. Hi Clara just a small question I wanted if you'd like to say something about the Indian Ocean you didn't say too much about that and maybe you'd like to. Thank you Rowan yeah I thought about it I guess I found that the Indian Ocean was very at least a mist type of analysis seemed to go along very much with the Pacific. And and you could see that in all of the maps there is fairly good quite good data coverage in the Indian Ocean back in the early decades of the data sets so it's a very important region it also has a very large signal to noise in terms of decadal versus inter-annual variability so it is a key region I think for us to to look at for the decadal variability but I didn't notice that it had any distinct decadal decadal variability that was distinct from what we were seeing in the Pacific. I guess one question might be we tend to think of obviously the Indian Ocean responding to answer variability on intranial timescales and a question might be whether the relationship between the Pacific and the Indian Ocean is it's just the same on decadal timescales as it is on intranial timescales or whether it has some interestingly different features and if so what they might be. Yeah I may be not up on the most recent literature but when I had looked at this back in the paper in 2004 I found actually that the Indian Ocean had a very strong connection to the North Pacific not only in the ocean but in the atmosphere on the decadal timescales and it was actually that linkage was stronger than with the equatorial Pacific SSTs. Matt Newman in his talk later this week on the PDO I think he may bring in some of those some of that aspect as well that's a good question. I have just a question we had a question today already on these for the hiatus and the role of the seasonality for hiatus so I'm wondering if because we ignore more or less by looking at those patterns the seasonal cycle so there are two questions I have on this the first is what can be the role or the key element maybe to to look to take into account the the seasonal cycle and second would be do we maybe take out this connection because we do not properly remove the seasonal cycle because we look at this on animals because it's also an issue people really ignore that we are maybe not properly removing the seasonal cycle so what we are removing is maybe not the right thing so these are two questions and okay I guess my impression is that of course there is some variation in the spatial patterns of SST variability between the seasons and also because of the different mixed layer depths the timescales of these patterns may be a little bit a little bit nuanced or a little bit distinct and also in terms of how the ocean and the atmosphere couple they may have different strengths of coupling in the different seasons but despite all of that I think that these patterns at least that we've shown on the decadal timescale I don't think that they are too sensitive to the seasonal cycle but I'd be interested to hear other people's views on that yep please other speakers from today or anybody else can chime in from their own experience and answer the questions or ask new questions and so on hi Clara I'm just like to move the target a bit from observation to modeling you've shown these very strong patterns in the observations you've also shown from your own work and work of others that if you look at regional trends you can have a wide range of different decadal trends and and variability at the regional levels when you look at models and you know big ensembles now what's your perception of how do the models reproduce these particular patterns that you have shown do they have the right spatial structure do they have the right timescale of variability or do they just reproduce the decadal variability on spatial scales or timescales which are actually different from what we see in the observation that's a great question and I hope many other people in the room are going to answer this but I'll I'll just say a few words of my impressions from the models but there has been a lot of work looking at this I think in general this pacific pattern I don't know if they're seeing my slide here but the it's okay actually the pacific pattern the models are generally deficient in the magnitude of the decadal scale component in the tropics in the tropical pacific so the linkages between the tropical pacific and the extra tropics is I would say underestimated in the models for the most part regarding the connection between the north Atlantic and the pacific I think I'm not quite as familiar about how the models do that but I believe that there's quite a diversity in terms of how strong the linkage is between the tropical pacific and the north Atlantic between models and southern ocean I'm really not very familiar probably Mojib can speak to that that aspect if anyone else here can comment about how the models are doing do the models show a similar behavior yeah but maybe you can make a comment today preempt your own talk we'll excuse you for that from the I was going to show some of those pictures tomorrow related to AMAC variability and relationship with AMV but AMAC variability in the models both in their time scales and spatial structures quite different among the coupled models and similarly AMV structure is quite different again among the models so there's quite a bit of diversity I was just going to add to Gokhan's comment that in the Atlantic I think a real central challenge is that many models simulate interesting low frequency variability associated with a subpolar gyre but the connections with a tropical Atlantic are highly variable and many models have very weak expressions that very bit in the tropical Atlantic and yet many simulation studies have shown that in fact it's that tropical north Atlantic piece which is has really strong impact on the global climate system so I think there's a real challenge and you can ask questions why that is are there missing physical processes is it dust from the Sahara is it cloud feedback is it resolution scales in the Atlantic who knows but I think that is a central challenge it's a challenge because if you even if you can predict the decadal scale variation of the AMAC and its connection with AMV if you can predict that tropical Atlantic piece I think your ability to predict continental scale impacts goes way up so I think that's a central challenge I actually wanted to ask Ed I can Ed Hawkins and in your talk of uncertainty you essentially refer to model uncertainty but Clara showed us that there is a tremendous amount of observation uncertainty too so how do we really incorporate that when we study the care of variability and use both models and observations to understand the phenomena good question I think one of the points is to try and compare like with like I mean it's one of the points I tried to bring out at the end of my talk was to you know if we're looking at observed SSTs we need to compare with simulated SSTs not necessarily you know with the air temperatures which is what we tend to do so I think that's one important point you know I don't quite know how to bring in the observational uncertainty into separating sources uncertainty but you know we do have to use more than one observational dataset whenever we can because they all construct it slightly differently so we certainly need to take take all these factors in consideration does that answer your question to some extent yes yeah Karina is another question or comment okay I would like to comment on your question because I think this is one more or less one of the points within concept heat so it's in particular this question so first I think one answer would be to compare that the range or the size of uncertainty are different for model and for observations so maybe the ranges are bigger so the observations of course they have uncertainties but they are maybe that smaller and this is exactly what is gone you have to do the all those intercom perishing experiments and you have to define scientific questions what you want to resolve and what what you are resolving for which uncertainties you want to have for what so this is completely different approach for different questions so if you look at the decadent climate variability it is important to develop those intercom perishing exercises for observations as well as for models depending on your research questions and there's a lot of going on already so because I noticed that when we plot for example we saw several graphs today where we show the observations in the cloud of model uncertainty but we don't really show very well the the observational uncertainty that is involved and how does that change our ability to compare models and data so that's that's one one place where we have this kind of yeah this is especially something we have pointed out that that there's a need for the observations to do this so there's one way going forward as I shown by Clara with a with a climate data guide so that you accumulate information for the first and second that you also start these intercom perishing experiments which are public also for the observation side and there's a great need so there's also already something ongoing under Jesus Cliver for the historical data set for example for the in situ data but you have to do this with all the other products there's a great need on doing this is something we have identified and we try to push work in this direction so I completely agree but this has already been identified there is down here Brad brad lindsley thank you I just want to change the topic a little bit cleric this is brad lindsley can I ask you a question about your guess your opinion about the the relationship between ENSO and the PDO and the pacific whether I've heard some discussions about the potentially large very large a linear events triggering a phase shifts in the PDO or I just wonder what your thoughts would be about the processes and the problems we're having I'm going to highlight Matt Newman's talk coming up because a bunch of us have participated in a review paper on the PDO and the views put forth there with I think some good evidence is that maybe this PDO phenomenon is actually consisting of several maybe four different processes and the connections between them the linkages between them are um uh with linkages between them um uh but maybe not to think of it as a single phenomenon that has this spatial structure as far as triggering shifts of the PDO I think ENSO could be one way but there are also other ways I think that you can trigger shifts of the PDO I don't know if anyone else has comments on that I guess we have uh we I don't want to you know Matt and other people are going to talk on this in the pacific where I've been at the session so we'll leave this as an open question uh till I think it's Wednesday or something like that I well I could jump in with another question and and uh with us two people here spoke or specifically about the climate system response to external forcing uh not anthropogenic forcing but natural uh forcing like solar and and volcanoes and I wanted to ask maybe this question can be asked in two ways to what extent is what we see in the observation reflecting some of these external forcings and the other question is okay so if those those are completely natural internal modes to the climate system coming from interactions within the climate system components how much do we have to consider these global modes when we look at the response to solar and volcanic volcanoes so I would like to to ask the people who yeah or other people or anybody else yeah it's it's more a remark than the question following on this I think they are so there are more and more issues on the volcanic impact on the scale of viability I think more and more papers but also more and more papers on weaker volcanoes than what we thought up to now tropospheric volcanoes as opposed to stratospheric volcanoes as the one Davide has shown so there are more and more papers in the in the hiatus literature but more generally people tend to think now that these volcanoes also have an impact on the climate viability and also volcanoes coming from the high latitudes not only the tropical ones so it's it's just an I just send this to the world and I relate this to Gavin's talk on updating the c-meat five forcing I wonder if this shouldn't be taken into account also for future scenarios or like future scenarios of the past climate thanks so there is a there is a lot of work being done on updating the volcano parts in the c-mip six scenarios and one of the things that the at least some of the groups will be moving towards is modeling volcanoes using point emissions of the at the appropriate height as opposed to using you know a stratospheric aerosol obstacle depth which traditionally people have done and that is a way to totally unify both the the other kinds of sulphate forcing but also the tropospheric and stratospheric ending ones and so you can put in you know however high it goes and you can just see what impact it has the the tests that we've run indicate that the if you put in an SO2 source and you just allow your normal chemistry to to take it and move with it you get you get stratospheric aerosol patterns as long as you have a reasonable stratospheric circulation that looks that looks as you know quite good compared to the observations and of course you can use that going back much further than the satellite era which right now you know a lot of the volcanoes particularly in the in the last millennium runs there's a lot of inconsistencies in how those are being modeled right now so I think things will be moving forward in and doing things in a more coherent and synthetic way in next time I just want to ask a question about volcanic eruptions and these mode of variability up till now the scientists agreed that there is no physical relationship between the eruptions and the ENSO and they they think that we have to remove the ENSO impact when we are going to study the volcanic impact so my question is like there was a talk that these volcanic eruptions can enhance the NaO like they can in in the stratosphere they can bring the NaO towards positive phase and we think that okay if they can enhance the NaO there could be some relationship between ENSO and the NaO so when the the scientists think that we have to get rid of ENSO so they simply apply the regression and remove the ENSO signal but how much we are sure that when we are applying the this regression we are removing the NaO as well which we think that could be the reason because of the eruptions so how certain we are that while doing this eruption I mean while doing this regression we are removing ENSO but on the other side we could also remove the NaO signal as well so that's mean our results could be I mean like uncertain so do we have any idea how to get rid of ENSO if this is not the the cause it's not caused by eruptions so then how to remove ENSO efficiently without getting rid of NaO that is caused by eruption I'm pretty sure I never said that we should or do anything like removing ENSO to understand the response the problem I would say is rather than it is quite uncertain then still how tropical Pacific responds to strong volcanic eruption so it depends actually also in which kind of model framework you put yourself to understand that because for example the earlier study is based on this dynamic thermostat mechanism suggested that positive ENSO phase after stronger eruptions but seems not to be at work in more complex climate models and on the other hand we have reconstruction suggesting that there is a positive ENSO phase so I think it's a matter of understanding how the tropical pacific response to strong volcanic eruption rather than to say okay we have to get rid of the tropical asphalt that because it is inserting so that's my point on this any other questions and I'd like to remind you that I avoided questions in the morning so if people still remember points that they want to ask about for the morning speakers can I just ask something else about the volcanic response so I mean there's this understanding that this NaO mechanism is perhaps important but it's not always found in all models and there seem to be some simulations that give an ocean response but that's not linked to the NaO maybe just through a direct surface cooling so I wonder if you could comment on the state of knowledge about those issues how robust the NaO response really is and whether other other mechanisms of ocean response are likely to be important all out first of all I would say that this NaO plus kind of response is kind of robust in the observations and I think the mechanism is rather clear and it's basically the downward propagation of these strengthened polar vortex and sorry yeah the models the problem is that the models don't get that as robustly as one would expect but the possible reasons are several one is again the forcing how we implemented forcing in the model and actually in the case of reconstructed events just talking for example about last millennium okay we are using kind of very simplified approaches to implement the forcing so we have this AOD latitudinal bands for example and the response in the stratosphere is kind of washed out this dynamical response uh linked to the thermal radiation on the other hand there are limitations in the model characteristics for example uh troposphere stratosphere coupling maybe not sufficiently well represented in the model just because of the low resolution vertical resolution of the model so we have to understand that and I hope that coordinated modeling activities like this one will help to answer question related to this fact concerning the oceanic response okay the NaO is one point there are contrasting evidences about in the model world about how the other modes respond atmospheric modes like the PDO sorry the PNA response to the to volcanic forcing but we shouldn't neglect the fact that for example already this the anomaly that are induced i'm talking about the tropical eruptions mostly on direct relative forcing effect on the surface this cooling reduces the marital temperature gradient which leads to an effect also in the stratosphere so we can think also about response in terms of the coupled ocean system that ocean atmosphere system that is independent of this direct um forcing from the stratosphere first of all I would like to make a comment on this statement about the vertical resolution in the stratosphere I hear a lot of times people saying oh you know we don't get the linkages between the stratosphere and the troposphere because we don't have enough resolution now in the s is here under we have model we don't have a very good stratosphere troposphere links we haven't actually explored the volcanic aerosol issue in strong depth but we are thinking about a range of phenomena and still we have 137 levels in our model um metophys model as over 80 levels um mine pray I it's sad for me to believe that it's only an issue of vertical resolution there've been experiments done recently increasing the vertical resolution very much in the stratosphere within the specs project and also we have tested the impact of vertical resolution in the stratosphere you definitely get the much better description of the stratospheric variability so if you look at the qbo for example that's much better if you increase the vertical resolution but you don't necessarily get the the connection between the stratosphere and the stropo and the troposphere better just by increasing the vertical resolution so either we really need an order of magnitude or levels we don't know but in the sort of range that we are using at the moment at least in numerical weather prediction it doesn't seem so obvious to me that just increasing the vertical resolution in the stratosphere automatically or it's the main factor in in reproducing these connections um the other question I I wanted to ask actually refers to another talk this morning and one by Karina on on the energy budget and the ocean heat storage and and she reminded us at the beginning of the talk that actually only a few percent of the imbalance energy imbalance at the top of the atmosphere actually goes in the atmosphere itself most of it the vast majority goes in the ocean and therefore if we actually look at the ocean heat storage that's a much better way of relating basically the energy budget to to the imbalance in in at the top of the atmosphere however we do live in the air and we do live on the continents and therefore the question is do we as far as the heat which is actually basically the changes in near surface and air temperature over the continent do we think that that is actually directly related to this three percent of the energy imbalance or is rather related to the amount of heat that the ocean gives back to the atmosphere and that may actually depend in my opinion much more on the patterns that Clara has actually shown rather than the what actually comes from the top of the atmosphere or basically the the the radiative balance so there's no doubt that in the long-term trend there is a relationship between all these variables but I'm wondering on on the decade of time scale when we look at the basically the the the big temperature anomalies over a large part of of the continents are these anomalies directly related to this three percent or are they more related to how this decadal model variability managed to extract this heat that goes into the ocean and bring it back to the atmosphere was answering this that's what we are wondering maybe Karina can say something I think I cannot answer this hundred percent but I think yes I agree I think you're completely right because the inventory which has been done and where it has been announced that the storage of energy from this accumulation of energy in the climate system is predominantly in the ocean it's it's an inventory which has been done for the long-term trend on the other hand you have to differentiate between the internal climate forcing and the external climate forcing and we have tried to to discuss its inventory in this paper on the discussion and our first figure shows the figure which has been shown in the IPCC on the different climate forcing in order to get to get also more the time scales on this climate forcing so I would give the first answer would be yes and it's a question of time scales and and if you look at climate variability you would not look at an inventory so I think these are two different issues we're discussing but it's more a common than not the deepest I mean did you want to add something to the volcano because I saw your hand up well I've just done something in terms of the vertical resolution question I mean so most of the climate models in c-mip 5 don't have 137 layers I think the maximum was maybe 80 but in our tests weird I mean our c-mip 6 runs will have like you know 96 to 100 layers and that's sufficient to self-generate a QBO that makes a big difference to the drop strut exchange in the tropics and we're seeing much better ages of air we're seeing much better strut tropics change of you know brilliant 7 and nitrous oxide which impact the chemistry and a no zone in the southern hemisphere and so it wasn't sufficient just to change the resolution you have to fiddle around with some of the things that impact that for instance the gravity wave drag parameterization that's a key thing that you really have to concentrate on but you can get even within a climate model at you know 90 to 100 levels really very good stratospheric circulations as long as you know and the lid has to be high enough as well I mean you can't do this with a lid at 1 millibar it has to be at you know 0.1 or 0.0025 or something like that but you can get dramatically improved simulations with not that much of a cost in terms of computational time. Gavin actually if you can keep the microphone I wanted to ask you because the question about volcanoes and the tropical pacific reminded me a question about solar and the tropical pacific because I've just seen a recent paper that showed that the response to solar variability and I didn't read it in details to know what it is but the model is that the response was entirely opposite to the response that Jerry Meale got from the Wacom model in other words when there is a maximum solar they tend to get ends up and El Nino and not El Aninia that Jerry showed that he's getting so do you have something to say about your results in the tropical? I mean that's something that I've looked at and it's very ambiguous I mean we don't see any clear impact on on and so variability my my gut feeling is that we still don't have enough of the right processes in the tropical pacific at the right level of detail to be able to see a robust result and I think that that goes over to you know the huge diversity and responses to anthropogenic forcing in and so as well I think we still have you know we can get the amplitude and the and the spectra about right but I think we're we may be over tuning our models so that that is not a sufficient measure to demonstrate that we have skill in predicting how it's going to change under any kind of forcing whether that's a strong volcano whether it's solar or whether it's an anthropogenic changes if there are no other questions we'll first of all thanks Clara for joining us from from boulder we can head on to the next item which is what is it reception sorry Clara cannot join us in the poster session so thanks again everybody who spoke today and everybody asked questions and let's continue the same the same spirit for tomorrow thank you bye bye bye bye bye bye