Inconvenient: New paper finds the last interglacial was warmer than today – not simulated by climate models

I have a couple of problems with this. Forced harmonic oscillators have several essential components: A restoring force (approximately linear, at least for small perturbations). A driving/forcing force — usually harmonic, if only because we have to do nasty convolutions if it is not. A damping force to represent dissipation, (usually approximately linear in the velocity). And a mass.
I realize that you are expressing a metaphor, not asserting a model, but I’m not sure that a metaphor justifies extended conclusions as if it were a model. In particular, I can see nothing in a climate system that performs the same role as a mass (or inductance, if you prefer electric circuit oscillators). Yes, there is an internal storage or buffering of energy in e.g. the deep oceans, but that energy has no momentum equivalent and cannot carry the system through even a single undamped, unforced oscillation if it is perturbed.
I should qualify this statement, of course. The weather involves secular motion of actual mass on the Earth’s surface — atmospheric flow, thermohaline circulation — that have actual momentum and which are forced by a complex mix of Coriolis pseudoforces and real variable buoyancy forces as it is differentially heated and cooled, and some of those motions have natural rotational periods and a very few — e.g. the evolution of the diurnal tidal bulge — could have something like an actual restoring force coupled to a periodic forcing. However, the frequency spectrum of the mass/momentum oscillations of this sort is highly compact compared to even the smallest frequencies relevant to climate evolution on geological time scales — they are simply irrelevant, or at least, it is very difficult to see how they could be relevant.
On the longer timescales, I think you have to think of everything changing very, very slowly — slowly enough that the system for the most part merely tracks a local “quasi-equilbrium” dictated by things like the orbital dynamics, which at least have a very definite and computable effect on insolation as eccentricity changes. It is much more difficult — for me, at least — to understand the effects of changes in obliquity and precession, as they involve the projection of varying insolation onto the also slowly varying geographical arrangement of continents and oceans. This is further complicated by a secondary but extremely significant variation in planetary albedo with the distributed fraction of planetary surface covered with snow and ice, which is also effectively projected onto the slowly varying geographical arrangement of continents, sea bottoms, mountain location and height, and coupled to things like thermohaline circulation in nontrivial ways, and the fact that the dynamics themselves are highly non-Markovian with a time kernel or “memory” of previous climate with timescales that can be very long indeed — hundreds of thousands of years in the case of Antarctic and Greenland ice pack and (perhaps) the deep ocean.
The closing of Panama and subsequent rearrangement of thermohaline circulation, if indeed this was the proximate cause of the Pleistocene’s gradually deepening descent into glaciation, is then very “odd”. If this “flipped a (million year long) switch” as of maybe 1.9-2 mya, with a gradually increasing effect up to that point, one would expect to see some sort of disjunction in the global temperature, but in fact it smoothly continues the 41 ky cycle that was already established. Again, the interesting thing isn’t the 41 ky cycle — that is understandable — it is the continuing gradual deepening of the cold. The cycles, on average, keep getting colder and colder.
This is enormously odd. The average temperature of the planet is systematically changing, but by “average”, I mean in a running window maybe 200,000 years long to start to smooth out the cycle! The character of the oscillations is distinct — warm and cold phase temperatures differ by only 2-3 C, often less, and the Earth spends almost as much time a bit warmer than the mean as a bit colder. This behavior actually extends almost linearly back to an inflection point roughly 3.5 mya, but with what looks like chaos for the first 800,000 years — chaos with features tens to hundreds of thousands of years wide — before the 41 ky cycle asserts itself.
At 1.5 mya, the character of the oscillation changes again, but still within the 41 ky pattern — the oscillation remains roughly symmetric and if anything, it levels off for a half million years, but the amplitude of the oscillation roughly doubles to maybe 4 C with rare cycles as large as 6 C. At ~1 mya, it changes again, this time the most profound change seen anywhere in the last 5 million years. The cycle mean temperature substantially drops, by close to 2 C. The period converts to ~100 ky, but with lots of beats and stretches where the oscillation again looks more like chaos than a “simple” oscillator with a handful of relevant periods (it would be very interesting to do a fourier analysis here if one could actually include enough cycles to make it particularly meaningful, but really one can’t, I don’t think). The cycle peak to trough amplitude increases to as much as 8 or 9 C. And finally, the cycles become highly asymmetric, spending easily 90% of the time “cold or changing” compared to the time they spend warm. And finally, even though the warm cycles are comparatively short, even though the cold cycles are deep and long (long enough to build up 2-3 km thick ice pack on the continents!) the warmest part of the warm cycles warms up well past the peak temperatures seen in almost 3 my!
Oh my! How in the world can one understand this? Sure, it looks like orbital resonance and eccentricity became important but why only 1 mya? What was changing to (comparatively suddenly!) “switch” the system between the regimes discussed above? And what of the paradoxes inherent in the eccentricity even now, where perihelion occurs at the coldest, not the warmest time of the year. This is usually “explained” in terms of the differential albedo of land and sea and a north polar ocean vs a south polar continental ice cap, but all this explanation really teaches us — if true — is that the projection of the eccentricity, obliquity, and particular angle of season in the precession onto the exact shape of the continental land masses, smeared and modulated over thousand to ten thousand year periods by latent heat of any icepack and by the deep ocean temperature in places where thermohaline upwelling carries up deep water and by who knows how many other possible things? — is an extremely sensitive causal factor in progress towards or away from glaciation!
Literally, sweeping the precise angle of the equinox as projected onto eccentricity and the climate even a small amount can make a huge difference. It is very difficult to understand these changes in the pattern of glaciation in terms of the closing of Panama, as one would have expected that to have nearly been “equilibrated by hundreds of thermohaline turnovers even by the time it closed, or at least within 100 ky or so after.
This still ignores many factors that could be important. The annual variation of peak normal TOA insolation is roughly 91 W/m^2 (compared to the order unity W/m^2 forcing expected from doubling CO_2). A clear modulation of the climate occurs with what appears to be the period of an important decadal oscillation (the PDO). The state of the sun is fairly well known over the last 200 (instrumental) years, but is increasingly difficult to precisely infer or date from proxy data from earlier times. We are staggeringly ignorant about possible cosmological or solar modulators of climate, but there is substantial evidence that the climate system is amazingly sensitive to some small variations in forcing while perfectly happy to blithely ignore or even go counterphase to others, even when larger (e.g. the 91 W/m^2 variation). We do not even have a good feeling for the sign of the feedbacks for local short time climate dynamics and are cosmically clueless about longer time scales, e.g. the proximate cause of the Little Ice Age or Medieval Warm Period — none of this sort of century time scale variation seems connectable to Milankovich type long time scale dynamics, and yet the LIA “could” have been a trigger back into glaciation — as far as we can tell — or not — ditto. We don’t know why it happened in the first place, so we cannot say how important that cause is in the grand scheme of things over longer times.
Underlying this is the — IMO obvious — fact that the Earth’s climate system is fundamentally multistable, or at least has been for the last 3.5 my. Before that, there was a slight warming and cooling with fourier components that match up somewhat with obliquity and precession, but no discernible effect from eccentricity in spite of the fact that it should have been modulating as much then as it does now — presumably onto a world with the wrong continental shape and thermohaline circulation to support much oscillation from either that or obliquity and precession. (Even then, the climate appeared to be slowly cooling, but the effect is barely resolvable over millions of years.) Now, multistable with hysteresis, not with mass. It doesn’t seem to switch like an oscillator, it switches more (indeed, very much) like optical bistability in an open resonance fluorescence system, or (less) like a ferromagnet can be switched.
Multistable — lets stick with bistable for simplicity with just warm phase interglacial and cold phase glacial — systems have, for a wide range of their driving parameters, the possibility of being in either phase, with an imprecise boundary in parameter space within which they are locally stable. That is precisely what the data seem to suggest (for a value of multi greater than bi, possibly even for some sort of fractal distribution of possible attractors/stable points in the climate system). Once the Earth is glaciated, it is not at all easy to warm it up. Just the right conditions have to persist for a very long time to melt all of the ice, alter the albedo, restabilize in a warm phase interglacial. Once those conditions cease to hold, though, the system may well remain interglacial for a long time, even though if the earth were glaciated, it would remain so. This negative (cooling) feedback from ice is very likely to be why the 41 ky cycle is almost completely suppressed for the last my. Every four precessions (or so) obliquity, precession angle, the continents, and eccentricity all conspire to create conditions that can flip the switch, and once the switch is flipped they persist long enough to make the warm phase stable, but after a comparatively short time, precession makes it only locally stable — if it were cold, it would remain cold. The system then awaits a fluctuation like the LIA sufficiently long to “tip” it into the stable cold phase.
It might not need to be very long. If glaciers grow every year, you gain a tiny bit of albedo every year, as well as gain the ability to absorb a bit more insolation into latent heat (remelting of ice, evaporating of water) instead of reheating the air and ground to maintain the average. The same is likely true of icecap/sea ice modulation. Even things like mere “normal” climate variability can, in its drunkard’s walk, sometimes by pure chance create persistent heating or cooling because the system is so very nonlinear and multivariate.
The system is very likely crazy nonlinear chaotic enough that increasing atmospheric CO_2, which will very definitely tweak one of many climate forcings in a particular way, could very well cause a transition into the cold phase if the range of parametric stability of the warm phase has narrowed enough due to e.g. orbital precession from the warm phase initiator state of 12,000 years ago. If you study limit cycles of nonlinear systems at all you learn very quickly that as poincare trajectories are being followed around stable points in the dynamical evolution, perturbing the parameters away from stability in any direction will, in time, produce extreme-r behavior in the opposite direction as one shifts to a “larger orbit”. If there are multiple attractors, one can easily be pushed over the boundary and into the well of another attractor.
We don’t — I argue — know anywhere nearly enough to predict the effect of increasing CO_2 on the climate. One of many possibilities is to stabilize the warm phase and put an end to the glacial cycle for the time being, possibly even for geological time (as we don’t really know things like how far the continents have to move before the system stabilizes or destabilizes even more, possibly “suddenly” as in new modes appear). One theory for the Younger Dryas is a bolus of freshwater glacial melt suddenly dumped into the ocean interrupting thermohaline circulation long enough for glaciation to recur for a thousand year before orbital dynamics forced the system back into warming, but that’s only one of several possibilities. Assuming that the seas rise, we don’t know the feedback from sea level rise — does a greater sea surface area actually inhibit warming, so that it is self-limiting? Or does it increase warming, so that it tends to rise faster? Or both, plus other nonlinear effects from ice/albedo/latent heat modulation? Can we even say which one is likely to be dominant, let alone “certain” to be dominant? I doubt it.
I also doubt that we understand the even more subtle feedbacks of water vapor/cloud dynamics well enough to answer the question. There is a very good reason to think that the normal state of the climate equilibrium is the fixed point where water vapor feedback turns negative. That is, less (average) water vapor in the atmosphere causes warming to restore it, more (average) water vapor in the atmosphere causes cooling to restore it. That’s what stable equilibrium is — a point where small perturbations cause the system to return to equilibrium, not be forced further away. But if this assertion is true, warming from increased CO_2 should cause strictly negative feedback from water vapor and the water cycle to reduce the warming. If it is not true, one has to appeal to some subtle and complex things to explain why the climate doesn’t run away from positive feedback from water vapor alone.
Finally, “anthropogenic” contributions to climate variation are not limited to increased atmospheric CO_2 from human operated power plants. Every time an old growth tree in the rainforest is felled, we alter the biodynamic carbon cycle. It is further modulated by the rate that various nutrients are released into the ocean. The ocean itself has optical thermal properties that may well depend on things like pollution and silt from human farming and fouling of waterways. I’ve been in an extended discussion with somebody who thought that a lipid near-monolayer resulting from oil and gasoline dropped into the ocean by human boats and human garbage dumped in the ocean and oil leaking into the ocean etc is a major modulator of the ocean’s climate contribution — affecting things like evaporation rates and surface absorption of energy. I doubt he is correct, but I’d have a difficult time proving it from known data or previously done experiments. Farmland has a different albedo and very different climate impact in lots of ways than forest it replaces, and farmland in much of the world is at far greater risk of desertification, which has hysteresis associated with it — once a desert is formed, it takes a long time and just the right conditions to turn it back into non-desert land.
All things being equal, it would be lovely not to rock the Planet Earth boat too much as we struggle to build an equitable global civilization where the world’s 1/3 poorest people might as well be living in the 19th century or even earlier for all of the fruits of civilization that they can afford or enjoy. OTOH, it is unthinkable that we should — given our enormous ignorance of the effect of any given human activity — condemn millions if not billions of humans to perpetual poverty and misery and famine while we ourselves remain energy-wealthy, just because some models that appear to be actively failing predict a possible, but far, far from proven catastrophe. We simply don’t know enough to predict much of anything at all. If we could afford to leave things unchanged it would be great, but we can’t, and we don’t even know how to predict the future outcome of leaving things unchanged — it might be even more negative.
To conclude, I have to say that I doubt that we have an understanding of climate evolution sufficient, even with Milankovich and help from things he ignored, to take the Earth’s climate system in the Miocene and evolve it through the Pleistocene and into and through the Holocene right up to the human-forced present in the last century, at most, of the Holocene. Until we can, I would argue that we don’t really fully understand the climate, not to the extent required to make quantitative predictions or even to be able to resolve something very simple, like “the measured temperature distribution of Earth” vs “what the temperature distribution of Earth would have been if CO_2 remained flat at 290 ppm”. We don’t even know how to compute a statistical expectation for the delta in any way that is at all convincing, unless one is willing to conclude that our current climate is extremely unlikely because it isn’t doing what the climate models say it (probably) should, rather inverting the general course of science relative to model building.
rgb

rgbatduke
September 3, 2014 at 10:23 am
phlogiston
September 3, 2014 at 12:59 pm
The above was a reply to your long comment, sorry I posted it as a new entry rather than reply, need to get used to the new WordPress system.