 So the original topic as people mentioned before this meeting was supposed to start out as climate shifts and When I was asked to participate I said, oh, I don't know anything about climate shifts. So that'll be interesting And then I was assigned this talk. So now I had to learn a little about about climate shifts So this will not be my work This is a review of what I've found about climate shifts, but I think I know now. So how does this? Yeah, so what I did is I me and my Graduate students IU Wang Who's working actually on pacemaker? We set up we better look at this and we did what's called these days research online To to survey the literature for this talk So we did Google scholar searches on things like climate shift which had remarkably few You know exact phrase climate shift practically very few papers came up And we were more successful. We said well, we can't just talk about climate shift It has to be expanded a little so we look for Regime shift and so forth. So this is what We came up with well first I thought it'd be good to Give my kind of personal view of some of the concepts involved Climate to me is the statistics of the instantaneous weather variables Climate variability then is changes in those statistics. So it changes in the mean changes in variants Frequency correlation any statistic you can think of that's climate change or climate variability and Provisionally climate shift then is a change in the statistics That persists for much longer time than the transition between the two states So it's a kind of climate variability On the other hand a lot of the papers or a lot of the few papers we discovered people use climate shift as Equivalent to a change in sign of some index like Atlantic multi-decadeal variability Or a Pacific decadal variability or not top of the atmosphere heat flux certainly a change in Sahil precipitation index would be very important to change in sign and so forth Okay, and so what about climate versus weather? So weather is is the instantaneous values of the weather variables and and So climate is a lagging into indicator like if you were betting on You know stocks a lagging indicator would be something that's already happened and it's not that interesting In fact, you don't know what the current climate is until it's The past climb really So in order to for example Climate prediction versus weather prediction you have to wait ten years to verify a decadal climate prediction Something might happen in the last of the ten years if you're looking at the mean over the ten years or whatever Whereas a weather prediction you verify against instantaneous data So there's some kind of philosophical issues in in climate prediction that maybe Will be touched on later for example An end so prediction a seasonal prediction events of to me is a weather prediction for SST You might not have to know the atmospheric Evolution exactly, but you do want to know the exact evolution of the ocean temperature Whereas a seasonal hurricane forecast is a climate prediction You want to know some number of hurricanes or something like that over the season So then we come to climate regimes so climate climate shifts are related to climate regimes and Effectively these terms are used interchangeably So regimes are distinct climates and they're motivated. I think by the three variable nonlinear Lorenz attractor model I'll show you a schematic in a second and the regimes are our regions Like the circles of those two donuts their regions surrounding those empty spaces in the middle of the empty spaces or some kind of This is a very mathematical model Unstable fixed points and then as your Climate which is a point that moves around in this space as it goes from one donut to the other it switches regimes And there's a transition between between the regimes and if it if it stays in one place for a long time This is another that's a climate shift Here's another view of what a climate shift is So this is what's called a hysteresis diagram on the vertical axis is Let's say the forcing Or the response say of the system So like global temperature or something like that and the horizontal axis is some measure of the forcing So as the forcing increases to the right and you start on the lower curve You reach some point and you jump up to the upper curve When the forcing gets strong enough and that's a rapid Transition that you could interpret as a climate shift on the other hand when you decrease the forcing It'll follow the upper curve and then kind of fall off a cliff to the lower curve So that's a hysteresis that occurs in these kind of nonlinear Lorenz type models when you have a Response to a change in external forcing And then if you have some noise on top of it you can induce these shifts To occur not just at those cliffs but in some region around them so in reality there are climate shifts, but this is a picture of Over 40,000 years of some climate data and you can see in the various curves This is 40,000 years on the left to current on the right various Quantities from from ice cores Say say look at the top one There's a time of strong variability But not much say trend and then all of a sudden there's a trend around 10,000 years ago and say the ice sheet melts and then you get to the upper part where There's no trend again, but the variability is much reduced So I don't know you know what I'm not interpreting these but I'm just saying when we look at these climate times time series you can see various kinds of shifts climate shifts when you look at them and At the bottom you see a change in the trend and then the shift between No trend and then trend and then no trend for example So what about some more relevant things for us here in terms of decadal prediction? Well, I did realize that this is 2015 which is 20 years of Clivar basically and Back in 20 years ago. I remember the things we were talking about were PDO or PDV AMV global mean temperature and So forth and so the things that luckily I'm talking about the same things here today But we have 20 years more data So this is a picture the time series is the time series of the Pacific Decadal Oscillation up to 2000 from 1900 to 2000 the SST pattern on the left here I can You see there's a horseshoe type pattern in the in the Pacific Ocean Where there is one sign along the equator and then opposite sign in the extra tropics in the Pacific and This is the positive phase so warm in the equator I suppose you see in the time series. There's obvious decadal or multi-decadal variability in this time series But and maybe maybe there's evidence of climate shifts But I think it's it's easier to think of this as just Decadal variability Here's a longer time series of the PDV or PDO And at the bottom you see that this squiggly line is the data Reconstructed Pacific Decadal Oscillation on the bottom this gray curve is Various regimes or various shifts denoted by Jumps in the in the gray line at the bottom Along with the the PDV or PDO there was at least recently We've seen that sea level pressure in the North Pacific has different states for different states of the whether it's positive negative so the Mean sea level pressure. This is a change between 1977 to 88 and 56 to 76 I'll come back to this later so Let's look at some of the kinds of climate shifts that have occurred Or if they've occurred changes in the statistics this is Representing changes in the statistics and actually one of the papers that we came across that mentioned the exact phrase climate shifts Was by kicharski and in sick So they are the actual Experts as far as I'm concerned And I think Fred was actually mentioning this work as part of your the lab program here, but this one is showing the AM AMV index observed from 1900 to 2010 in the upper left and you see the strong multi-decadal variability And on the right side You can see the changes that go up that in the ENSO The magnitude of the ENSO variability basically so stronger ENSO's As as Rowan was talking about during different phases of the AMV So what are recent climate shifts that I was able to identify? Well First there is 1976-7 and that was a shift in the mean Next one that we found was 1988-9 Another shift in the mean Then there's the hiatus 1998 to 2015 which is a a shift in the trend and There was also a change in the NAO AMV What appears to be a sharp trend in that in 1990 a sharp change in that in 1995 and 6? So these are all very Very well documented the the 1976 climate shift. I think was first also Fortunously the year that satellite observations began This climate shift was identified and people spent a lot of effort to say it wasn't just Climate shift. I mean, it wasn't just the satellite data It affected fish for example and things like that. So they couldn't they couldn't have been affected by launching satellites, I guess Anyway, trend birth in 1990 finally, I think it was the first one to identify this 1976 climate shift. So that was after 24 years Recent observed inter decadal and inter decadal changes in the northern hemisphere and What he found was basically this change in the in the sea level pressure and the allusions from high to low in 1976 Sometime later Another 10 years or so later Zang Wallace and Batiste Found that there was a tropical connection to this Multidecadal variability you can see there time series in the top panel with the ENSO Variability removed and the structure of the multi-decadal as shown in the bottom panel with the Change in the the lowering of sea level pressure and the allusions Or this is this is an SST. So it's a lowering of SST in the north Pacific and warming in the in the tropics with a somewhat wider Latitudinally wider structure than And so which is in the middle panel then also in 1997 Mantua at all popularized the PDO PDV and Their structure now you can see also graphics dramatically and got better in the 90s And you see the same picture basically that that from the Wallace paper And a time series of the the PDV which shows decadal variability in the upper left-hand side So you then you just once you have this index you can regress it on any field You want and get a pattern and that's what this is shown the SST and SLP regressed on the on the PDO In the top panel here then in 97 also Minnobay found that you know identified coherent changes in other indices like SST Pressure so forth with the PD with the PDO and And Also, there's the point of view that sometimes these shifts can These changes can occur when different indices kind of very By accident become in phase or out of phase And this is something related to that which is called climate networks, and I understand people will be talking about this kind of phenomenon also, so this this network is PDO NaO and so and global mean temperature or something like that and seems it's a bit arbitrary to me, but The dashed lines are where 1920 1940 1980 where they said there's some kind of climate climate shift from this network approach so summary of what happened with the PDV is it started I think as a climate shift and then it went In the 90s, and then it became a decadal oscillation and multi-decadal variability So I don't think people really think of PDV that much as Climate shift anymore. What about 1988 1989? Was there a regime shift then and this is again related to PDV Mentioned in 2000 and fish abundance has changed and so forth But Seems to have been at least from looking at the current time series This index didn't change sign at that time. It sort of started to and then Looks like it decided to go the other way Change in the global mean temperature trend that's been mentioned already reduced trend 1998 to present it's gotten lots of press huge number of explanations Some of my old work was rediscovered by people from the 90s and But the data has been reevaluated new correction applied I think the conclusion might be at least from the from these that most recent paper Never mind That and but if you look at these pictures the top panel shows both the Or the bottom panel say with and without corrections the actual Curves are practically indistinguishable when they apply these new corrections due to buoys And it's just a matter. I think that we haven't really gone long enough yet to see if there's a Been a change in the trend. Maybe maybe not It's a question this paper is a question of statistical significance, which is Statistical significance doesn't mean it did or didn't happen. It just means we can't Be that sure And finally the NAO 1995 and 6 shift I have a you can see the NAO time series here and where the black line is To my eye it looks like there was a shift in the shift in the sign from positive before a long run of positive to a long run of negative and Primarily negative Until the present so that that could be a climate shift Question of whether it's now attribution and mechanism and so forth in a paper in 2003 while it was still thought in 2003 that this was a positive trend in the NAO You can see actually the negative trend that make the shift if there was one happened Seven years earlier. We still thought there was a positive trend, but we showed that At least up till 1999 this this kind of time series is consistent with just noise in the In the atmospheric model so conclusion My conclusion is that if we define climate shifts as rapid changes between distinct climate regimes To distinguish them from quasi oscillatory climate variability on decadal to centennial timescales We probably don't see much recently, although certainly there are very strong climate shifts in the observational record At least the proxy record for for paleo climate And It's difficult to find climate shifts this way in the recent record But there are plenty of examples of what you might call regime change type shifts or changes in sign of Indices and so forth. Thanks