 Thanks, Johan. So I'm going to talk about pacemaker experiments. I mean we heard a lot about this type of experiments during this workshop and I'm going to so to focus on the protocols and also to give some use some information about the pacemaker experiments we recommend it for cip6 So what what are the motivation for the pacemaker experiments? So we heard a lot about about this, but I just I have just one slide about the motivation for this type of protocol. So the goal is to constrain the internal to decadent modes of viability of a couple model to follow the observed fluctuations and and so the motivation are to investigate both the local response to the constrained temporal evolution of the SSD modes and also the removed ocean atmosphere response in the over-oceanic basins that are not constrained because we are in a in a in a couple system so basically it's an attribution problem and so I just reproduced here the two the two pictures that we've seen already from from Yucca Saka and Xie showing on the showing the trend in the observation of the of the temperature over the this period here and the result from a pacemaker experiments where the SSD is restored here over the eastern part of the Pacific and so what what you can see on this plot here that they match pretty well showing Suggesting that the eastern part of the Pacific is responsible for a lot of the of the patterns that we are showing in in the observation So in principle, this is a very easy protocol Basically, how does it work? You just add a restoring term over the selected region or selected domain where you want you to restore the SSD And you set a buffer zone because we are in a couple mode between the restored and the rest of the the restored area and the rest of the of the ocean but As always we need to be very cautious In this type of experiments because the restoring term may create energy imbalance leading to the spin-up of the of the couple model And also the restoring term may perturb the the basic local ocean dynamics that may Perturb the entire system because as I said before we have to remember that we are not in a forced mode It's not very constrained. We are in a couple mode So basically what we need in this type of protocol We need to find a set of parameters that perturb the least the equilibrium and the physics of the couple model why we control enough the temporal evolution of the of the SSD and the low frequency change of the SSD and I would say as usual the devil is in the detail of this of this type of protocol so So I'm going to to to go into a little bit in Into detail about this restoring parameter because it's extremely important. So So this is equation is just the way You do the the pacemaker experiment basically you take the the non solar heat flux of your model Which is here and you had a restoring term Which is proportional to the difference between what is simulated by the model and the SSD you want to restore to and you have this this parameter here, which is a gamma parameter which cannot which is also called the feedback coefficient and the unit of this This parameter is what from meter square per Kelvin so they've been depending on this parameter here the the restoring is You control basically the strength of the of the restoring with this parameter and So for for the any type of experiment this parameter gamma is just infinite If you have an infinite restoring to the those of the SSD that you prescribe You a strong restoring could be a thousand watt per meter square Which is equivalent to a restoring of a two-day for a 50 meter depth of the ocean me mix layer and the weak rest of me Could be for instance or 40 watt per meter square, which is equivalent to two months restoring for a similar type of For say the same value of the mix layer depth So I'm going to show that this parameter is complete is very crucial for this pacemaker Experiments and I'm going to use one model, which is a synapse M5 model I'm not going to describe it And I'm going to I've done I performed two type of experiments an authentic pacemaker experiment and the e-specific pacemaker experiment Similarly to what you Kosaka and she have done. I'm not going to talk about this one because we heard a lot about This this pacemaker experiments. I would focus on this north Atlantic experiments So you think this model and what I did I As a test I test different values of the of this restoring parameter, which are given here So I have four sets of experiments with So a strong restoring corresponding to about two days and a weak restoring cost bring to about two months and some in between And I have three members for the two extreme of these restoring parameter And what I did I did an experiment where I branched The the space maker experiment on a historical run of C&R M same file and for the sake of I mean, I didn't have a lot of computer power basically and not a lot of time either So I just did some sensitivity experiment over this period here, which is a 30-year period Basically, it's just to to show how the protocol works and what we should do and should not do So I'm going to describe this Using this very simple Analyzes which is based on the earf decomposition of the SST. So this is the observed earf the SST Over the the novel the North Atlantic for winter time. So you recognize a traditional tribal here, which Projects also on a MV And this also padded here. This is the second mode of viability and is you have here the time series so the observed time series is given by the color here and so Blue and and and pink Showing that this mode over this period here as an abrupt shift In 96 95 96. I mean, this is the traditional shift that we we heard about This this workshop So the the observation are given there and then you have a kind of spaghetti here, which corresponds to the Projection of the of the model SST when the SST is restored to this pattern here So the the red curve Let's concentrate on the red and the blue the red curves corresponds to the strong restoring and basically it follows the observation This is exactly what you want and the blue curve corresponds to the weak restoring where you could see is that you have you have three members here so three curves so you have a dispersion between Between these members but basically with this week parameter because it's two months You still capture the internal the decadal shift of the Of the SST If you do a correlation between these time series and the observation the correlation is given here So for each member and then sample mean is this one so it's point 87 which is pretty good And if it's you have a strong restoring this is point 99 of I mean by construction This is very good too for the the the second one here the more more dispersion But basically you can see that you can also capture the Internal change this is for winter. This is for summer and for summer. You can really see that you capture a lot Even whatever the restoring a strong or weak restoring the you capture the the shift really where there is no dispersion And this is because it's in summer the mix layer depth is it's very shallow So basically the the restoring that you apply is in fact a much stronger Because the mix layer depth is about 10 meter 10 20 meter in In summer which correspond to rest through a restoring time scale even for the for the week one To about 15 days. So basically we capture a lot. So it's seasonality actually in this in in this restoring business That is kind of important too okay, so now that we see that The The restoring whatever the the value works pretty well I mean the target is that we I mean works pretty well in the sense that it follows the internal variation that we want to To produce the the the target with this type of experiment is not I mean is to compare to a control experiment and usually usually the control experiment is a historical experiment that are produced with the same model and And the target is Use you shouldn't introduce any change in the mean state Or in the pacemaker experiments compared to the historical experiments because otherwise you're not going to to compare to to fairly compare the the simulation so here I'm showing you the The biotropic stream function. So basically the ocean Horizontal circulation in the ocean and I'm showing you the difference between the historical experiments and the pacemaker experiments With this acronym here. So 40s Q 40s what you want from the square but Kelvin this is weak restoring and this is strong restoring here And what you want is basically you want to have everything in white With no change in the in the mid-state This is the difference between the historical and pacemaker experiments and you could see that For this week restoring everything is white for the strong restoring here The difference in the mean state is it's quite significant We have a strengthening of the subpolygide here and the strengthening of the sub-topical Right here, and it's it's the the magnitude is about 10 Sphere rep which is not significant which is significant 25% of the of the mean basically If you look at the another parameter, which is a mixed layer depth It's a little bit of change for the weak restoring for the water meter square per Kelvin But it's not as strong but for the the strong restoring the mixed layer depth is getting very is getting much deeper here in the eastern part of Central part of the Pacific and also you have along the along Greenland here You have a very strong deepening and and here shallowing in the center of the of the of the paint of the subpolygide So basically what we what you see here. It's a strong restoring even if the so this is is Very much like like the observation with a very strong restoring you have an acceleration of this circulation here and plunging of the mixed layer in the sea and Here you even have some spurious effect like a very strange signal in the model meaning that the perturbation is really strong and really Effect is here the the circulation If you look at another parameter, which is a amok That's so the amok is extremely important because it's reduced to be the heat of the system and we are in in a couple model here So basically what you want you want to have the the amok the shading here I'll present the viability in the in this model of the of the amok over the from the historical Simulation so we have 10 members for for this one and you want to have the pest maker experiment within this This this this envelope here So what you see is for very strong restoring the amok as a strong as a drift, which is quite important The variants of members can have a six. I mean about more than 10 minutes further up more than The historical experiment and so what you see here is in fact the strong wrestling Leads to a model spin up While the the weak restoring it's perfectly with the the range of the Estimation given by by then some Historical simulation so you could you tell me okay who cares because what what the one what the atmosphere sees is just SST so who cares about the amok below the the ssd Actually the model cares a lot Because as I told you the the the marina recirculation is Reduced to be the heat over the system and I'm showing here the the heat content in the north Atlantic So it's average from 060 north and here the heat content in the south Atlantic so same type of plot the The historical the envelope of the historical are here and what you see that the heat content in the In the north Atlantic is drifting in this model But it's also drifting in the south Atlantic where you don't want to have any influence of the of the ssd So there's an adjustment here of the tropical Atlantic heat Oh, this is a tropics I forgot to tell you a tropical Atlantic heat content and and it perturbs the the southern Atlantic basin and So there's only a 30 year integration But I I could start seeing some change also in the ACC It's not significant because it's for years integration and and we should Carry on the the the integration but because of this change in the We did with this distribution of the heat the ACC is starting to be affected in In this type of model. So the question is can we confidently interpret the remote? Influence of a more in the south Atlantic when the response is very much perturbed by the protocol that that that we are using And it basically on the choice of this parameter of restoring another side effect of caveats of this very very strong restoring is the the the relationship is It perturbs a lot the relationship between the the atmosphere and the in the ocean at high frequency So basically what I'm showing here is the the lead lag relationship on So the correlation the lag correlation the correlation here between the the temperature and the precipitation of this area here the TNA That's on topical North Atlantic area and they have the lags This is a lack of relations So in this part of the of the plot the ocean leads the atmosphere and in this one the atmosphere leads the ocean and You also have this is again this these four curves corresponding to the different restoring so your target is to to to match the historical simulation and And what you see is I mean in this in this area for the precipitation So when you have warm ssd, it's preceded by weak by by less precipitation basic Basically, you have killer as clear sky and the ssd was so the ocean is responding to the atmosphere and you have the feedback here which they have more ssd after this ssd is formed a more precipitation after this ssd is formed and What what you see it with strong restoring? This is the red curve basically you completely destroy here the Relationship which is much better captured with the with a week restoring This is the same thing for the nets for surface net net solar flux and the same also for the for the latent heat flux so with a strong restoring you don't allow the ocean to adjust to the high frequency of the Atmosphere, which is the the main source. I mean the main feature at very at very high frequency This was in a in a in a tropical Atlantic. This is in a in a sub logi Same results in a sub logi There is a relation strong relation I mean relationship between precipitation and ssd meaning that the one ssd is warm you have less precipitation before and This is shown here and in in red you could see that you have the the opposite type of research And this is also true for the sea level pressure the sea level pressure. It's a relationship. It's not it's not good I'm going to skip this one okay, so this first conclusion so the Strong restoring it can be can lead to spurious behavior and so that it can Provide I mean you can trigger some some twist because the energy budget is spits basically not not close in this type of simulation and So you destroy also the interesting Ocean-Atmosphere relationship and so the recommendation is do not put too far too far the the button for the rest of the term and And also it's very it's it's crucial to diagnose precisely the impact of the protocol I Have two more slides for because you can't tell me that it's a Smaller dependent. Yes, it is more dependent and I'm going to show you another flavor of Spurious effect that you can have in the GFDL model. So this is thanks to You can and and and rim work, so you've seen already this This picture from from rim describing the The protocol that they used so basically they did some Experiments where they restore the the system in that case to fix SSD pattern the MV plus pattern then MV minus pattern And they do also a control simulation So I'm show you the difference yesterday between the MV plus and MV minus I'm going to focus here on the control experiments So they did some some test on this Restoring coefficient as well With this model and we I'm going to focus on this area here the the surplus I very important So I'm showing you here Three curves with three different Oops with three different restoring term So the black one is kind of strong rest of me. It's 20 days and is there is a drift even in this So apologize and that way it's kind of significant and if you look at the Amok you have here the overtone escalation In the ocean showing the amok the mean amok in in contour and the it's a difference between the historical simulation and And the amok in this in this pacemaker experiments You can really see that have a weakening of the amok In in the modern as a function of lead times is 20 years here. This is five-year average Period so and this corresponds to the less heat transport it to the to the sub prototype That is something very interesting they apply the same the same Forcing but in they introduce a new correction, which is of salinity correction In the sense that they they added an additional term So that the density is neutral. So it's compensating the The change of the density due to the temperature So in that case you don't have any density change in the subrogio And you can see that the drift is lower and it's the same same type of pattern of the Amok in this in this part here and the last one the red red curve is the same with a weak Restoring so the surface restoring the sea surface restoring the sea surface flux for neutral density and A weak restoring and the drift is very minimum and the drift in the in in the amok is also very Minimum so it's not only one model, which is very sensitive to the to this what meter square per per kelvin And What I said it's extremely important because you know in here we are in a couple more days to to evaluate the flux restoring term And what what you want you don't want to add any energy into your system with this protocol experiment And I'm showing you the the three three curves again for this flux restoring term So you remember the equation I showed you before this term that that is that allow you to restore the density You want basically what you want again is zero so you see in this strong restoring without Density neutral correction there is a very strong drift and you see the value here the value is is At the at the very end is 15 watt per meter square, which is huge It's a huge amount of energy that you put into your system We're not talking about one or two watt per meter square like for for the signal that you want we want to investigate We are talking about 10 watt of meter square here And so the target is to have zero and you could see that this week with this week restoring and this salinity Additional restoring you have a perfect balance, which is what what you want and there's something very important It's the pacemaker experiment. It's not flux corrected experiment You should add you should not add this the energy Okay, so as a conclusion as I told you the Strong restoring can lead to spurious behavior So we we should be very careful about about it and the second conclusion is very important the recipe to avoid the drift is model dependent and And the use of salinity term for density neutral might be required is true for the gftm model. It's not true for our model C&M CM5 model, so there is no magic recipe to avoid the drift and each group has to work on it on all model So it's an interesting possibility Alternate not alternative, but I think hypothesis that need to be to be tested It's what I showed you is as the gamma parameter is always fixed and there are some some preliminary work done at IPSL where the gamma here It would be a function actually of the mixed layer depth in that case you don't you can't you really control the time scale of the restoring that that you want because if With a fixed gamma if the mixed area depth is is really high Then you don't control much and you really want to keep the SSD anomaly that you want to investigate the impact So for CM6 because of all of this this caveats we we have decided in this DCPP component C to place to to put the idealize AMV experiment in tier one and to to put the pacemaker experiment for the Atlantic and for the Pacific in tier three because it's extremely important so the drift must be avoid and And and to to work on this drift It's it's much easier to work on this idealize experiment. So we've we thought that is it was in fact a secondary step after this Idealize experiment and the recommendation for the to the group would be to provide some basic diagnostics to document the drift So for instance the time series of the flux by storing term Which should be close to zero the Mock the seasonal cycle of ENSO because I didn't have time to show you But the pacemaker experiment is an Atlantic can change the seasonal cycle of the of ENSO in your model And then you you start to drift in the Pacific and this is what something that you got in the Friday new experiment, I think because you had to do control experiment in the Pacific So I'm not sure I have to go to into the detail of this So you can find in the document of the component C in DCPP you can find all the all the The protocol so basically I told you the idealize experiments are in in tier one with a weak Restoring and we suggest to do so to follow the the Protocol tested by NCAR and by GFDL So 10 10 years and 25 members the more member you you you can put is better with the Idealize experiments for MV plus and MV minus and we also added the idealize experiments for the the the IPO so same type of experiment but for the for the Pacific and And so the pacemaker experiment are done in Now put in a tier 3 priority Something we also want to test. I mean we heard about the volcano in that are very important in decodal Variability so in this DCPP component C We also want to to evaluate the impact of the volcano in the in the prediction So the the idea in tier one is just to repeat the 91 prediction without a pinot tube or to see the impact of the pinot tube in in the prediction and And also the idea for the Component B so component B is the the real-time forecast the real-time forecast should be done with With and without a pinot tube because what we want we want to have the the range of possibility of the forecast and Ineruption can happen anytime So this is in tier one and and the last experiments that that are in T3 we want to investigate a little bit more the The impact of the initialization, especially in the cephalogy We we've heard during this workshop that the cephalogy is extremely important for predictability and the source of viability So we want to repeat the the the shift Of the of the cephalogy so to repeat I mean to to repeat the the prediction Where we don't initialize the the supologize and repeat this date here to see if the If the shift of the of the MV is forecast or not if the supologize not initialized Okay, so I just to finish I I really liked what dirt knock said we should not be a shame of curiosity in in In our field energy we just had it we should not be a shame of slowness as well Because sometimes I have the impression that we we tend to go a bit too fast in this In this business. Okay. Thank you