 So I would like to thank the organizing committee for inviting me to present at this very interesting workshop. As introduced earlier in this session, the satellite observation revealed a substantial decline in September Arctic CI-610. And in the right panel is the lowest record in September Arctic CI-610 in 2012. And this decline trend has been often attributed in large part of the increase in greenhouse gases. Interestingly, during the same period when we have a decline in Arctic CI-610, we also have an increase in trend in the Antarctic CI-610. And it's not well understood what caused the opposite trend in the two-harm sphere. And historical record also shows there's a much decadal variability in the Arctic surface air temperature. So the recent rapid warming we see in the satellite period is mainly due to the decline in Arctic CI-610. However, in the early of the 20th century around the 1930s, there's another rapid warming period. And after 1940, there's a cooling period lasting for several decades. So it's important to understand the mechanism for this much decadal variation in addition to the long-term trend. The Atlantic-Miridana overturning circulation is known as a major source for the multi-cadal variation. So the key question I would like to address is what is the role of low-frequency AMOCK variability in the observed Arctic CI-610 decline since 1979. In this early study back in 2011 using GFDLCM 2100 year control simulation, for the first time we identified that AMOCK is indeed a key player for the winter Arctic CI-610 variability. And showing here the AMO index is highly correlated with the Arctic surface air temperature and anti-correlated with the Arctic CI-610. And this regression map shows in the model with intensified overturning circulation is lead to a positive phase of the AMO and induce a reduction of winter CI-6 concentration over lab-door-C, Greenland-C and Baren-C. And the spatial pattern is very similar to the observed decline trend in winter CI-6 concentration over the satellite period. This indicates a possible role for the overturning circulation in the observed winter CI-6 decline. And this anti-correlation between AMO and winter Arctic CI-610 is also found in the same three models and also the paleo record. So for this anti-correlation to behold we need to have strengthening of the overturning circulation over the same period as of the satellite period we have a decline in winter CI-610. But before 2004, before the establishment of the rapid program of direct observation of AMO, we do not know how the AMO changed in the past. So we can reconstruct the historical variation of the overturning circulation using some fingerprints. So in this early study one fingerprint identified is in the tropical North Atlantic the surface temperature over tropical North Atlantic is anti-correlated with subsurface temperature in the same region. And using JFD-LCM 2.1 water-hosting experiment it shows a weakening of the AMO leads to a southward shift of the ITCZ and cooling in the surface tropical North Atlantic. Meanwhile the summer climate is different and also the weakened western boundary current induce a warming at subsurface. So this anti-correlated change is shown to be a fingerprint of the overturning circulation suggests that the AMO was indeed weakened during the 70s and strengthened since then. And this tropical AMO fingerprint also found in same five models and paleo records. Another fingerprint is from extra tropical North Atlantic. So which defined as the leading model of upper ocean heat content. And so this very recent study explains the simple mechanism why a change of the overturning circulation can lead to this dipole pattern in the fingerprint. If initially we have a positive AMO anomaly at northern high latitude that will propagate southward with the slow advection speed. And the Muridano heat transport will propagate in the same feature. That lead to a convergence of a Muridano heat transport over the Spala region and divergence of the heat transport over the Gulfstream region. And the time integration of this heat convergence divergence lead to this dipole pattern in the load on integrated upper ocean heat content. And which correspond to this dipole spatial pattern. And this fingerprint can be well predicted on the decadal time scale. In this paper we also shows another experiment if we surprising the AMO propagation without the propagation of the overturning circulation. This fingerprint disappeared and there is no predictable signal. And this southward propagation also exist in the isopikinal models and also in the high resolution models. So this schematic plot explained that in the region north of 34 north due to the existence of interior pathway of Northlandic water, this slow propagation of the overturning circulation and the Muridano heat transport is crucial for the evolution and the enhanced decadal predictability of the AMO fingerprint. Which consistent with the recent decadal prediction studies, successfully predict the warm shift in the Northlandic Supolar Jair in the mid 90s by initializing a stronger AMO at northern high latitude as Rowan Sutton showed yesterday. And south of 34 north because there is no interior pathway, the AMO propagate with the fast coastal wave speed. So the direct influence of ocean heat convergence, divergent in this region is much weaker. That's why we have most predictable signal in the Supolar Jair. And if you look at the surface temperature, the surface temperature, the AMO signal also has a tropical branch that has to be through the atmosphere response to the change in the Supolar region, such as the cloud feedback. If we do not simulate while those atmosphere response, we cannot predict while the tropical signal in the surface temperature. So showing here is the time series of the observed AMO fingerprint. There's a strengthening trend since the late 70s. And after 2007, there's a decline trend, which is consistent with the rapid program observation. And this fingerprint also has coherent change with observed AMO index. That indicates the overturning circulation is indeed a key player for the observed AMO. And the magnets for this fingerprint, you cannot find in the model coupled just with the slab ocean. So now the question is whether the strengthening trend of the overturning circulation can also cause the decline in summer Arctic CI-610. So in this recent paper published early this year in Penas, I studied the magnets for the low frequency variability of summer Arctic CI-610. Using a much longer control simulation, 3600-year CM 2.1 control simulation, it shows that the Atlantic heat transport is indeed one of the key players for the summer Arctic CI-610 variability. So the Atlantic inflow enters into the Arctic region mainly through the Iceland-Scotland ridge and further split into two branches. When enters the Barents Sea through the Barents Sea opening, the other flow northward as Western Spiceburg current and through the east front street. Both eventually reach the central Arctic. So I analyzed the Atlantic heat transport across the Arctic circle, this integrated heat transport along this black curve. And found that there's indeed strong anti-correlation with the September Arctic CI-610 over this entire control simulation at low frequency. So with this long control simulation, you can focus on the multi-decade on centennial time scale. So showing here the right line is the inverted Atlantic heat transport which leads the September Arctic CI-610 by several years. And the spatial pattern shows with the enhanced Atlantic heat transport, we have a reduction of September CI-610 concentration in both the Pacific side and also the Atlantic side. Now the mechanism is because the enhanced Atlantic heat transport induces the enhanced basal melting and the reduction of Arctic CI-610 at all different seasons. And this collision map shows that this anti-correlation is strongest in the Atlantic side and decay to the Pacific side. And this change of CSMAS signals in all seasons contribute to the change in September Arctic CI-610. Now this Atlantic heat transport variability is indeed caused by the overturning circulation variability. Showing the lower panel shows that the overturning circulation actually leads the change in the heat transport by about one year. And this change in the heat transport also causes a coherent change downstream of this heat transport through Barency opening and East Farm Street. As showed in the upper panel, there's a coherent change with the heat transport across Arctic Circle lead, the downstream change by several years. And yesterday there's a nice poster showing this advection process from the MPI group. The heat transport across the Barency opening is strongly anti-correlated with winter CS-610 over in the Barency. And so because the Atlantic heat transport can cause a change in both winter Barency CS-610 and also summer Arctic CS-610, so these two times here are also significant correlated. And the lowest panel shows the observed record of the winter Barency CS-610 and observed September Arctic CS normally. They both shows very similar normalized decline trend and has also highly correlated with each other. And this just shows the observed winter Barency CS-610. In the 1979 most of the region in the Barency is covered by CS in winter boil by 2006 we see a substantial retreat of winter CS in Barency. And there's an optional base study shown by Alkan which also in the audience. This next paper shows the observed increase in the heat transport across Barency opening is a prime driver for the observed CS decline in the Barency. And what is the role of the atmosphere heat transport? It turns out that with the stronger overturning circulation that's induced enhanced North Atlantic heat transport into the Arctic then there's an enhanced upward surface heat flux released into the Arctic region and this enhanced heat flux is carried by the atmosphere here transferred back into the lower latitude from the Arctic region. So the Northward atmosphere heat transport actually reduced and this compensation between atmosphere heat transport and ocean heat transport so-called Biryakni's compensation has been found at Decadal timescale in previous studies. Here with a much longer control simulation at multi-decadal and centennial timescale the anti-correlation is much stronger than that at Decadal timescale. And so the Northward atmosphere heat transport actually provide an active feedback to the September Arctic CS variations. For summer Arctic CS viability the Atlantic heat transport is not the only key player. The Pacific heat transport through the Barring Street also contribute to the change in summer Arctic CS. With enhanced Pacific heat transport we see a reduction of the September Arctic CS especially over the Pacific side. And this Pacific heat transport is also highly correlated with PDO in the model. And another predictor is the special atmosphere pattern the so-called Arctic dipole defined as the spring season CO2 in the CO2 pressure and this positive Arctic dipole induced enhanced transpolar ice drift enhanced CS exports through the Barring Street. So that leads to a reduction of the CS at the Pacific side and slightly increase of CS at the Atlantic side. And because this cancellation so the net effect of the Arctic dipole is not very efficient compared to the other two predictor. So there's a three key player for the summer Arctic CS decline the enhanced Atlantic heat transport and the enhanced Pacific heat transport and positive Arctic dipole. So we can reconstruct September Arctic CS using a multiple regression model with all the three key player. And also it's estimated based on the observed trend these internal viability can contribute substantially to the observed Arctic CS decline summer. In particular the positive Arctic dipole and the enhanced Atlantic heat transport contribute about half of the observed decline and the Atlantic heat transport is a major player here. So in summary the AMOG viability and the associated Atlantic heat transport into the Arctic has play a significant role in the low frequency viability of summer Arctic CS extent. In both models and the observations summer Arctic CS extent variation significantly correlated with winter Barring Street CS extent variations indicate an important role of the Atlantic heat transport into the Arctic. The AMOG fingerprint indicated a strengthening of the overturning circulation since the mid 70s consistent with the observed decline in Arctic CS and at the low frequency the atmosphere heat transport into the Arctic are forced by an anti-correlated change in the Atlantic heat transport into the Arctic. So provide an active feedback to the September Arctic CS extent. Enhanced Pacific heat transport into the Arctic and positive Arctic dipole also contribute to the summer Arctic CS decline. And very recent study identified a seven year pause from 2007 to 2013 in September Arctic CS decline and this updated time series shows we now have a nine year pause from 2007 to 2015 and there's no decline trend over this period. And it's quite likely that the recent decline in the overturning circulation contribute to this nine year pause and if the overturning circulation continue to weaken in the near future there might be a longer hiatus in the September Arctic CS decline. Okay I'll stop here.