164 thoughts on “The Connolley-Dekker Ice Bet: RD probably loses.”

1. And assuming he is a US resident, William needs to remember to declare those winnings on his income taxes. Both should consider applicable state laws prohibiting gambling…. maybe they should meet in Nevada.

2. Lucia,

   The axis labels are switched.

3. Shoot on the axis labels. I’m on my wait out to the gym. They won’t get fixed until…. maybe never!

   I have no idea how these gambling things work on the internet. I’ve been discussion proposing real bets with a “backer”, but I need to figure out what I consider my stop limit on several possibilities. I’d written a (disorganized) srcipt to test out a few ideas (bets on volume, area, multi-year, single year etc). Obviously, I wasn’t really looking at my lables!

4. Lucia, sorry if this is off topic.

   Humm…

   Is sea ice Earth’s thermostat? Open water (even cold water) loses heat to space rapidly compared to ice, so more open water means more total heat loss. Is the relatively slow measured warming since ~2000, in spite of continuing rapid increases in GHG forcing, related to loss of sea ice… which makes up for greater GHG forcing? Sure seems plausible to me.

5. SteveF.

   Yup. that’s a interesting thought.

6. Lucia,

   Setting aside the mislabeling of the axes, since when was the NSIDC September average extent for 2010 more than 12 million square kilometers? Am I going crazy or did you grab the wrong data somehow?

7. Mosher,

   To flesh out the thought a bit: From 2000 to present, there has been a gradual increase in GHG forcing (mostly from CO2), which is now ~0.27 watt per square meter higher. Since 2000, there has been a gradual (but noisy!) decline in total global sea ice area of ~1.5 million square Km. During that same period, the ocean heat accumulation rate has decline by ~0.25 watt per square meter or a bit more compared to the pre-2000 rate, while average surface temperatures have increased very slowly (compared to the pre-2000 period).
   So the relevant question seems to be how much additional heat is lost to space due to 1.5 million sq Km of open ocean compared to ice? My guess: A lot, and maybe enough to explain the above observations.

8. I’m also made suspicious by the fact that 2010 is shown as having a lower September extent than 2007.

9. Jon,

   It is the total global area, not the Arctic area.

10. SteveF,

    That may explain the chart but Connolley and Dekker are betting on the northern hemisphere extent not the global extent.

11. Jon,

    You are right. Maybe Lucia is using the sea ice volume graphic, not the area chart.

12. ARghhh!! You guys are right. I read Rob’s question, thought “Oh. I can do that quickly.” Pulled up a “for fiddling script” and forgot it read both ice area and volume. Worse, the key lines are:

    df_nsdic$anal=df_nsdic$dec_vo
    anal=”December Volume”

    But I hadn’t gotten around to make the graph titles draw from “anal” so that I wouldn’t screw up! ARghhh…

    Now I need to change this. (Very embarrassing.)

13. SteveF (Comment #80816)
    August 26th, 2011 at 12:01 pm
    “Open water (even cold water) loses heat to space rapidly compared to ice, so more open water means more total heat loss.”
    —————————————–
    Interesting. Why might this be so? If it is radiant loss, would the state of the matter or even the composition be of any consequence?

14. Owen–
    I don’t happen to know the relative emissivities of water or ice at similar temperatures. But, like reflectivity, absorptivity and transmissivity, emissivity is a property of materials. Colors and finishes can affect radiative properties. That’s the exact same material in black might achieve a different temperature than the same material in white.

    So, it wouldn’t be surprising if ice and water had different emissivities even at OC. But also, ice is generally colder than 0C and water generally warmer. So temperature could also matter. The warmer water would tend to emit more heat which would during sunless arctic winters mean more heat lost to space.

15. Re: Owen (Aug 26 16:07),

    It’s not the emissivity, which is close to 1 for both ice and water, it’s heat transfer. Open water has much higher effective thermal conductivity than ice, very much higher if the ice is thick. That means the surface temperature of the ice is going to drop rapidly and therefore radiate less. In round numbers, say the temperature of open water is 273 K and the ice surface is 253 K. That means water radiates about 314 W/m² while ice only radiates 232 W/m² for a difference of 83 W/m².

16. So the relevant question seems to be how much additional heat is lost to space due to 1.5 million sq Km of open ocean compared to ice? My guess: A lot, and maybe enough to explain the above observations.

    ####
    Funny when Willis talks about the clouds and tropics being a governer, my first thought was… Not the whole story, the excess heat may be escaping from the pole.. especially with more open water.. but looking at the radiation escaping there would be key I would imagine. That has to have been done.

17. Owen #80825,

    According to Wiki- emissivity follows “Kirchhoff’s law of thermal radiation: emissivity equals absorptivity (for an object in thermal equilibrium), so that an object that does not absorb all incident light will also emit less radiation than an ideal black body.”

    Now, since radiative emission from the surface of water is in the infrared, and since water is a near-perfect infrared absorber, I would guess it is a near-perfect blackbody radiator. Ice is also a strong infrared absorber (and so strong infrared emitter), but as Lucia points out correctly, the ice temperature can be much colder than the water. Ocean water freezes at about -2C, while wintertime ice temperatures can approach -40C, so there is a huge difference in blackbody emission… a ratio of (233/271)^4 = 0.55. Water at -2C emits almost twice as much thermal energy as ice at -40C. Heat transport by convection would (of course) also be much greater for water than ice.

18. lucia (Comment #80826)
    OK, I looked it up. From Wikipedia: “In general, the duller and blacker a material is, the closer its emissivity is to 1. The more reflective a material is, the lower its emissivity. Highly polished silver has an emissivity of about 0.02.” A perfect black body radiator has an emissivity of 1, independent of the material. So, ice is more reflective than H2O so thus would have a lower emissivity.

    The ice / water temperature differential however depends upon the season. In the fall, with ice and open water, I would think that the open salt water will be colder (subzero) than the ice (close to zero), and the emissivity depends on temperature.

19. Steve,

    I missed your post when I was responding to Lucia. There is little open water in the winter, and I wonder about the magnitude of the ice/water differential in the fall.

    But, your hypothesis does seem reasonable and could be modeled to estimate the magnitude of the effect (but not by me).

20. Owen,

    I imagine it is a very complex issue once you get past the simple stuff… like warmer water loses more heat than colder ice. The historical debate seems long and contentious… with some people suggesting that ANY complete (or near-complete) summertime loss of ice would flip the Arcic into a stable permanent ice-free state, due to greater ocean heat accumulation in the summer.
    Apparently, more sophisticated models pretty much always show that even summer-time loss of sea ice (August disappearance) would not keep ice from re-forming by November. A look at the behavior of Hudson bay (which is WAY warmer than the Arctic ocean in summer!), where thick ice re-forms each winter, in spite of complete loss of ice in summer, suggests to me that a year-round ice-free Arctic ocean is unlikely, even if there were complete loss of ice during the summer. Some relevant information is at: http://www.pibburns.com/smmia2.htm.

21. Lucia wrote : ARghhh… Now I need to change this.

    First of all, thank you for paying attention to the details of this bet, and testing your model on it.

    Since the mistake (using Dec volume as opposed to Sep extent) affect the entire post and maybe even the conclusions (and thus the name and title of this post), how will you make the changes ? Will you correct the graphs and numbers or re-do the entire post?

22. SteveF : So the relevant question seems to be how much additional heat is lost to space due to 1.5 million sq Km of open ocean compared to ice?

    Here is my take on this :

    Since open Arctic Ocean water does not get more than a few degrees above freezing, there is actually very little IR increase from open the surface after an area melts out in the summer.

    Comes fall, top layer freezes over and we are back to the state we were if the area did not melt out, with the exception that the ocean water is warmer than without the summer melt-out. This excess heat is trapped under newly formed ice, and thus will cause for prolonged bottom-melting (or slow-down of ice-thickness increase) durinig the fall.

    Conclusion : Open water in summer (where there used top be ice) will cause the ice pack to thinner in the proceeding year.

23. Rob–
    I’m going to do an entire rewrite and post as a second blog post while letting this mistake live in infamy. I’d have already rewritten but

    a) it’s one of the days my husband is at home and we are doing a whole bunch of things (errands, talking etc.)
    b) because the bet is closer, it’s worth discussing WC’s chance of winning money from you and
    c) because the bet is closer, it’s worth providing some numbers.

    I’ll still leave a few thing vague (like why my model does assume Gaussian I don’t worry about skewness — which Paul K suggested I should–etc.) If people want to understand the impact of those things I can leave it to a later post.

24. Rob Dekker,
    <Comes fall, top layer freezes over and we are back to the state we were if the area did not melt out, with the exception that the ocean water is warmer than without the summer melt-out. This excess heat is trapped under newly formed ice, and thus will cause for prolonged bottom-melting (or slow-down of ice-thickness increase) during the fall.

    Whoa friend! That is not how it works.

    When ice forms on the surface of water, ALL the water below the ice is very close to the temperature of maximum density. In fresh water, the maximum density comes at 4C, so that is the exact water temperature a bit below the ice in every frozen fresh water lake. In the ocean, the freezing point is -2C, and the maximum density is also -2C, so the water below the ocean ice is very, very close to -2C all the way to the bottom. It is physically impossible for warmth to “accumulate” under the sea ice in the arctic, even if the surface of the arctic were to warm to much higher temperatures during the summer. It does not matter what its temperature may have been (at any depth!) in the period before the ice forms, it is at the freezing point when the ice starts to form at the surface…. and that same temperature all the way to the bottom.

    Which is why the very deep ocean (~4 Km) is most everywhere very close to 0C. The very cold high latitude water sinks to the bottom and spreads over the bottom of the entire ocean…. it is more dense that the warmer water at the surface, and so only mixes with overlying warmer water very slowly (it takes millennia).

25. And perhaps my head will explode less often?

    When you started this sea ice bet, I imagined regressing the mean curve of sea ice on recent data. Of course the time never happened.

26. I would suggest some sort of logistic curve. As you approach zero it may get too slow but it’s an improvement to models that give negative values. Polynomials are almost never physical.

    Hm, I have some ideas I might look into.

27. Jeff,
    You’re head will still explode. Some fits eventually predict negative amounts of sea ice. It’s not even that far into the future.

    In terms of getting some handle on the best estimate and the possible variability, I think it’s necessary to do curve fits– but you can’t just display them the way Romm did.

    Andrew–If you can give the logistics curve a name I can try to discover if there is a pre-existing function in R that I can use. If it does, I can compute AIC coefficients without too much trouble. Otherwise, it can be too much work and it’s better to think of the polynomial curves as the first terms in a Taylor series expansion (which should, of course, be used carefully!)

28. Oh– I should note: The Connolley-Dekker bet ends sufficiently soon that many of the head-exploding features can be ignored by chopping off curves in 2016. But I think to ensure people who are considering betting have information that screams “be cautious”, I include the head exploding extension to 2020. Of course, the graphs become super-hilarious if we go to 2100.

29. SteveF : It does not matter what its temperature may have been (at any depth!) in the period before the ice forms, it is at the freezing point when the ice starts to form at the surface…. and that same temperature all the way to the bottom.

    Where do you get such stuff ? Did you ever look at FluxBuoy data and, for the Arctic specifically at Ice Tethered Profile data ?

30. Lucia:

    Of course, the graphs become super-hilarious if we go to 2100.

    It would be more than hilarious if anybody who reads this blog made it to 2100! (Friggin’ awesome, living well dude!)

31. Steve, apart from the fact that your assertions are disproven by observation of bottom-melt after freeze-over, your statement It is physically impossible for warmth to “accumulate” under the sea ice in the arctic is absurt even in theory, considering the fact that water is not an ideal heat conductor.

    It would be nice if you would study theory and empirical evidence a bit better before claiming “That is not how it works”.

32. Steve, the Canadian government has this really cool manual on ice (in Canada, that would be more useful than a “manual on sun tanning” probably).

    It’s amazing to me how many stages of ice there are in the development of ice from infancy to “multiyear” ice.

    Perhaps you could compare this to your ideas of how “ice freezing” in the ocean works? Waves make a difference in the dynamics of the freezing of ice, so does salt concentration. (No puns about “barely touching the surface” please.)

33. Carrick: “It would be more than hilarious if anybody who reads this blog made it to 2100! ”

    Well, speaking for myself, if I extrapolate the curve of “alive-ness” vs. time, I conclude that I will indeed be alive in 2100. At any rate, the observational evidence is consistent with that hypothesis. [The step-change in that function (many years ago) is attributed to measurement error. The adjusted aliveness anomaly (relative to baseline of 1981..2010) is 0 for all time to date.]

34. “I would suggest some sort of logistic curve.”

    Yup. That’s a good suggestion. But you might want to switch things around.

    Extent is really the sum of the area of all those grid squares that contain ice. So what you might want to actually calculate is the probability that a given grid cell is “open” at the end of the season.
    I imagine doing this on the entire grid structure with a different estimate for each cell. But you’d need the source data that extent is calculated from rather than just working from the final result..
    essentially a regional estimate that is then combined, where every region is a sensor grid.

    Something like that

35. Lucia wrote : I’m going to do an entire rewrite and post as a second blog post while letting this mistake live in infamy.

    Thanks Lucia. Looking forward to your new, corrected post for 2016, and how the odds work out for the Connolley-Dekker bet when using your models.

36. Jeff Id said : When you started this sea ice bet, I imagined regressing the mean curve of sea ice on recent data. Of course the time never happened.

    Jeff, do you think that to regress the mean curve on recent data is a valid model for prediction for future sea ice extent ? Are there no negative feedbacks that would drive the future sea ice extent back to the long term mean ?

37. Rob,

    “Jeff, do you think that to regress the mean curve on recent data is a valid model for prediction for future sea ice extent ? ”

    On a short timescale (months), it will be difficult to improve on. Long scales are for the more sophisticated people. There are plenty of effects which could drive sea ice right back where it was.

    If I had to guess, I would say the sea ice changes are completely unrelated to global warming. It is just a guess but the amount of change is too great to attribute to 2C of surface warming. It is a weather pattern change affecting ocean currents in the region. Some attribute that pattern change to AGW, however the evidence for that is plausible but weak IMO.

    Lucia,

    Too bad for my head. I’ll get the bandages.

38. Rob Dekker,

    I just looked at the Arctic profiler data for the first time (OK, only the first 15 of the “completed missions”, not all of the data). It is pretty clear that virtually all of the Arctic ocean temperatures under floating ice (including summer and winter) lie between 2C and -2C, with the vast majority very close to -1.6 (slightly lower than the -2C I suggested, due, I suspect, to slightly lower salinity at the surface than elsewhere in the ocean). The winter data is in fact virtually uniform at -1.6C. I had not ever seen that data (I was not aware the program existed), but I do not find the data at all surprising, nor do I believe they suggest any accumulation of summer heat under winter ice.
    As to “where you get this stuff”, it is pretty simple to understand the basics of convective instability in water bodies which are undergoing surface cooling. Short of strong salinity inversions (more rapid density increase due to rising salt concentration than density fall due to warmer temperatures) a significant increase in temperature with depth under the ice is essentially impossible. The question I ask in reply is where you get the idea that convective instability as the ocean surface cools does not preclude the kind of sub-surface heat accumulation you suggest. Certainly the temperature profile data you point to does not in any way support your suggestion (in fact, it pretty much proves what I said), nor does the most basic analysis of the physical processes involved… which is what I based my original comment on.

39. Thanks Lucia. Looking forward to your new, corrected post for 2016, and how the odds work out for the Connolley-Dekker bet when using your models.

    It might be Monday, because I need to add the high-side curves. But you will have a pretty good chance of winning.

    Mind you: this functional forms for this analysis don’t consider the “favorite” candidate model of those who tend to think the ice extent is mostly not driven by AGW. That is: They omit a low frequency oscillatory component. I’d add it…. but I don’t know how. Fist, I don’t know how to do it at all in R (that could be solved) Second, the frequency is unspecified, but thought to be long. So, I don’t think I could pick it out of the data anyway.

    So, to some extent, this model should be thought of as the sort of things a “cooler” or “lukewarmer” should look at before taking the risk of betting. It tells you odds assuming these oscillatory components don’t exist.

40. Jeff-
    I don’t attribute the weather pattern change to AGW. I do think it may turn out that the amount of ice is simply more variable even without AGW than previously thought. But I suspect at least part of the ice loss is due to AGW.

    It is warmer up north, we expect some melting. How much? I don’t know. But it’s plausible that smaller amounts of ice result in a pack that is more movable (not sure why it would– but not impossible). It’s then possible ice being movable tends to slightly more losses each summer, and also, lowered radiative losses in winter result in slightly less ice formation. The movability factor becomes important to figuring out what any “quasi-steady state” might be if we stablilize GHG.

    That said: the movability factor is also what can make natural variability large.

    So, I think the ice melt is “part AGW & part natural”, but I have no idea the ratio. None.

41. Lucia,

    The warmth up north would definitely affect ice and I do have a certainty that AGW is a non-zero effect. I think the melting and warmth that we see is from ocean current changes which may or may not be related to warmer ocean surface temps. These changes could even switch in the opposite direction from AGW in the future. The Antarctic is a good example of that although despite its general non-signficiant or barely significant uptrend, its average ice cover has also decreased in the recent few years.

    So I’m forced to agree that the ice level is certainly affected by AGW but like you, I don’t know how much. My own best guess is that it has little to do with the ice reduction we have seen.

42. Lucia AGW is responsible for 50% of the losses.

    more or less

43. Jeff–
    I don’t disagree. Nevertheless for evaluating bets with shortish-term end dates, I think it’s prudent to consider the trend persisting. That is: Before betting against Rob’s side of the bet, a calculation like the one I’ve done is prudent.

    By the same token, if I were considering betting against WC, I would have to ponder the possibility of that there is some oscillation with a frequency sufficiently long to not be “obvious” in the data wile sufficiently short to cause the ice to recover. But I don’t know how to include that.

    In the end, a person betting has to then gauge some of the outcome based on things other than curve fitting.

    To an extent I see a division between science and prudent betting. This post is more about prudent betting. But in this case, I’m just judging Rob’s likelihood of winning based on extrapolation using time alone. This is partly becasue I have no idea how to include things like the possibility that the 80s were on the ‘high” side of what you’d expect for the temperatures at that time and we might have moved into the ‘low end”, and if so, we might expect to go back to the “high”.

44. Rob Dekker,

    To second SteveF above, think about what happens when the ice melts and you get open water. Yes it absorbs sunlight and warms. But that warmth means the density goes down. The warm water will form a surface layer and a thermocline below it. The depth of the surface layer will depend on the eddy diffusion rate. But it won’t warm the deeper water significantly. When the sun goes down for the winter, that surface layer will cool and eventually freeze. An earlier thaw should mean a later freeze, but it will still freeze. The only way for the Arctic Ocean to warm enough to not freeze over in the winter would be increased heat transfer by ocean currents. Without a truly massive increase in air temperature, air circulation alone won’t do it. The problem with ocean current heat transfer is the geography. The Bering Strait is narrow so there isn’t going to be much total circulation. That’s why the ice exists in the first place. Put a land mass there instead of open water there would be even more ice.

    There does appear to be a cyclic component in Atlantic ocean circulation. Arctic ice extent on the Atlantic side was significantly lower and the sub-surface temperature near Svalbard significantly warmer in the early 20th century than in the middle of the 20th century and the late 19th century. It will be interesting to see what happens when the Atlantic Meridional Oscillation index finally goes negative again.

45. steven mosher (Comment #80857)
    “AGW is responsible for 50% of the losses.
    more or less”
    ——————————————————–
    What forcing agents/events associated with “natural variability” might be responsible for the other 50%?

46. Owen, I’m pretty certain Steven is being facetious.

47. Andrew_FL (Comment #80861)
    Owen, I’m pretty certain Steven is being facetious.

    Andrew, this link does not portray that estimate as facetious

    Kenneth Fritsch (Comment #80810)
    http://www.sciencedaily.com/re…..113956.htm
    Arctic ice could increase or decrease over periods of a decade and models indicate the faster than predicted decreases over the recent years can be attributed one half to natural causes and one half to AGW. Evidently it was the natural causes that the models erred on.

48. Kenneth Fritsch (Comment #80862)-The link is broken. At any rate, I merely meant that Mr. Mosher does not, I think, believe one can make statements like that (“fifty percent from this fifty from that”), certainly not without thorough explanation of how it is actually determined. Hence why he adds “more or less” meaning he is facetiously alllowing for virtually all possibilities.

    He is making fun of someone. Not sure who exactly.

49. I guess it’s safe to say, with JAXA to finish AUG at just under 5.00E6 km^2 and trend bottoming, RD needs an early SEP drop to below 4.80E6, or something in SEP lows at 4.60E6 or lower to offset highs earlier in SEP. With these trends, RD may still squeak into the neutral bet zone on avg, but I sure wouldn’t bet on it now (even if I did in a sense earlier, with my 7-day avg bet).

    Overall, I’m glad to see the wager; too many climate commentors aren’t willing to back their claims with real money — RD was at least willing to do that! With as much controversy as there is in the AGW-land, more should take such a stand.

50. Sorry folks; missed the 2016… never mind.

51. mccall– I suspect that’s what I did the first time I read the wager. 🙂

52. DeWitt Payne (Comment #80827)
    August 26th, 2011 at 4:31 pm
    Re: Owen (Aug 26 16:07),
    It’s not the emissivity, which is close to 1 for both ice and water, it’s heat transfer. Open water has much higher effective thermal conductivity than ice, very much higher if the ice is thick. That means the surface temperature of the ice is going to drop rapidly and therefore radiate less. In round numbers, say the temperature of open water is 273 K and the ice surface is 253 K. That means water radiates about 314 W/m² while ice only radiates 232 W/m² for a difference of 83 W/m².
    ———————————————————-
    So where does the thermal conductivity enter in? I understand the temperature effect on the blackbody radiation, but I don’t see how the thermal conductivity makes a difference/

53. Owen–
    If the water is warmer than the ice (say 2C), the bottom of the ice is say, 0C, and the top of the ice is radiating heat to 0K (approx), then the thermal conductivity of the ice will affect the temperature of the top of the ice.

    The amount radiated into space depends on the temperature of the top surface of the ice.

54. Owen:

    What forcing agents/events associated with “natural variability” might be responsible for the other 50%?

    Before industrialization the Earth warmed and cooled, so there’s nothing that requires that all of the warming since say 1870 be anthropogenic in origin. Indeed the models say that you need not invoke anthropogenic CO2 emissions to explain the observed warming until circa 1970 (the anthropogenic CO2 forcings d was masked by anthopogenic sulfate emissions, or so spake the models.).
    See this.

    I’ll also point out a substantial portion of the glacial ice loss occurred before 1970 and some of the warming that has occurred since then is due to a feedback resulting from the reduced arctic albedo.

    I think the long-term secular trend can’t be explained in any other way than global warming: the more recent phenomena is certainly weather rather than climate related…whether there is a link between that and climate change remains to be established.

    In my opinion, no more than 50% of the long-term ice loss in the Arctic sea can be attributed to anthropogenic forcing. The real number is probably somewhere around 25%.

55. Owen:

    I saw 50/50 for the following reasons:

    1. It’s an inside joke on a comment I made at Judith Curry’s, apportioning half of the rise in temps to AGW and half to natural variation.
    2. I’m also referencing the paper that Kenneth tried to link to but cant recall all the details. I just recall it was 50/50.
    3. When there are two causes and you have no idea which is larger..50/50 is the least wrong bet you can make.
    4. I’m being somewhat facetious, since the apportioning of causes is a funny business.

    make sense.. both serious and funny.

56. Owen,

    Ice is not a good conductor of heat. Neither is water, but the effective conductivity of water also can include heat transported by convection…. when heat is being lost from above or added from below. (The poor thermal conductivity of water is evident if you try to cool water from below or heat from above. Try warming ice-cold water from above with a propane torch if you want to see how good an insulator it is.) It is the convective turnover of the water column when it cools from above which yields rapid loss of heat from the surface of open water. The poor thermal conductivity of ice means the ice surface rapidly cools by radiative cooling to space… and thus reduces total heat loss. The ice is an effective insulator for the warmer water it floats on.

57. Carrick,

    I’m perfectly content to accept the statement that natural variability can and does effect changes in global temperatures and in the melting of ice. Nonetheless, natural variability is not exempt from causality. If we look at the long term changes as deduced from the Vostok ice core, for example, we try to ascribe causes (orbital and precessional forcings, for example, with CO2 feedbacks).

    We have seen a substantial retreat of summer minimum sea ice in the past 30 years – a rather large effect within a short time period. I am wondering what specific natural forcings have produced part of this effect.

58. OK, with Lucia’s and Steve’s help I get DeWitt’s point – since the thermal conductivity of ice is low, radiant heat loss from the surface can not be replenished by conductivity through the ice from the warmer water underneath, making the ice surface even colder. The reduced temperature then reduces blackbody radiation from the ice surface.

59. For sea ice, being influenced as it is by the ocean currents and atmospheric circulation, a regional relative effect on the ocean primarily influencing the ice is all that is necessary to influence it’s trend. Unless you can name me an anthropogenic forcing that preferentially warms the North Atlantic ocean (as compared with the rest of the oceans) it seems to me that the enhanced warming there in the last thirty years, whatever the “cause” will contribute disproportionately to the sea ice trend, and should be a natural influence. In the previous thirty years, during which we have no satellite data, the North Atlantic preferentially cooled relative to the rest of the world’s oceans. How much a role this regional effect has had in ice losses cannot be determined, however, since we have only data for one instance of the warming tendency of this phenomenon. We do have temperature data from locations in the Arctic that suggest recent temperatures have only recently matched and perhaps surpassed the temperatures there in the thirties. One would think that ice should have been fairly sparse then, too, but sadly there were no satellites then. No one knows, incidentally, the cause of the earlier Arctic warmings and coolings. So when you ask “what natural factors could possibly be involved in the recent trend” don’t be surprised if there are no unequivocal answers. For the simple reason that we don’t fully understand what gives rise to large multidecadal variability in regional climate. That we don’t know what natural factors could be at work doesn’t entitle anyone to assume there are none. In factor observational records tell us that factors we don’t understand do exist. Are then involved in recent trends? They might be. That is the crux of the matter in fact: no one really knows.

60. Owen:

    We have seen a substantial retreat of summer minimum sea ice in the past 30 years – a rather large effect within a short time period. I am wondering what specific natural forcings have produced part of this effect.

    I guess my point is even if the long-term trend is associated with warming, not all of that warming is necessarily man-made, even in the last 30 years.

    IMO natural variability lead to a larger than expected, excuse me “worse than expected” (/heh), warming in the 1990s, and a “less than expected” warming during the last decade. (I think ocean-atmospheric oscillations are the culprit.)

    Andrew_FL, I think “arctic amplification”, which is a natural feedback, can explain why the arctic experiences more warming (and cooling) than lower latitudes, when the global mean temperature increases (or decreases). It’s my understanding it’s a combination of the proximity of the land masses & relative isolation of the Arctic Ocean that leads to this amplification.

    In any case, you can see this effect in the climate data. See this.

61. Carrick (Comment #80875)-I am actually thinking of something a little different from Arctic “amplification”. Many people hear me mention an elevated level of warming in the North Atlantic seem to assume I mean “Arctic Ocean” which is weird, since the two may be connected but represent different geographic regions. The elevated level is for the Northern Hemisphere Atlantic Ocean, indeed there is actually even extra relative warming there in just tropical bands, relative to the global tropics, IIRC. Also, there is no obvious reason why “Amplification” should manifest itself in the Arctic at levels that vary with time. I think you’ll find the ratio of Arctic to global change was much higher during the first warming of the twentieth century and much MUCH higher for the mid century cooling. For no good reason, unless on multidecadal timescales the Arctic is influenced by something that influences the globe differently.

62. SteveF,

    I understand that you think you understand it all, but why don’t you slow down for a moment.

    Your assertion was :
    “Whoa friend! That is not how it works” and “It is physically impossible for warmth to ‘accumulate’ under the sea ice in the arctic”.

    Now that you looked at the ITP data, and have that there is enough heat there to melt the Arctic sea ice above some 10x over, don’t you think you stated a bit of an idealistic view of the Arctic ?

63. Andrew_FL:

    think you’ll find the ratio of Arctic to global change was much higher during the first warming of the twentieth century and much MUCH higher for the mid century cooling.

    It’s partly in the nature when the surface land ice was lost. I looked at the period 1960-2010, which is what I generated this data from. I don’t think the data prior to 1960 are reliable for the Arctic, so I personally don’t know what that will teach us.

    I’ll look at the Northern Atlantic data, and see what I see, but when you start narrowing it down to too small of a geographical region, you are more strongly weighting regional-scale fluctuations associated with long-peirod oscillations. You just may be looking at AMO.

64. Rob Dekker #80878,
    .
    OK, it seems you want to consider the details of the Arctic ocean profiles. Fair enough. Here is a graphical representation of the data from one of the “completed missions”: http://www.whoi.edu/cms/images/itp5dat3f_81104.jpg This is a relatively messy data set that may be worth examining.
    .
    ITP5 was deployed on an ice floe in an area of “relatively open water”, and collected data for about 1 year.
    .
    Here is what I see:
    In the top temperature graphic, there is a 25-30 meter surface layer which is at about -1.5C for most of the data set. Below that there is a band about 70 meters thick which is very slightly warmer (about -1C to about 0C). Below that there is another colder band about 120 meters thick (at about -1.5C), which varies very little during the whole of the data set. Finally, from ~250 meters to 700 meters, there is water between ~0C and ~ 0.8C, which also does not change much during the data set.
    .
    Near the end of the data set (summer solar heating?), the 25-30 meter layer warms from about -1.5C to about -0.5C on average.
    .
    Based on the above profile of temperature, we would normally expect there to be rapid (relatively) convective mixing from great depths to the surface. The bottom graphics of salinity profile show why that does not happen. The surface layer is ~20% depleted in salt relative to the deepest layers, so there can be a very slight increase in temperature with depth, which remains stable to convection.
    .
    Based on the above, I would guess that changes in temperature profile for the top 100-120 meters represent all seasonal effects, and water below 100-120 meters is not much involved in seasonal variations of ice cover. Certainly the deep water (that below ~200 meters) is isolated from the ice layer by a continuous overlying layer of -1.5C water, some 100 meters thick, so it is hard to see (absent clear evidence to the contrary) that the water below ~200 meters is involved in melting the surface ice.

    You said “there is enough heat there to melt the Arctic sea ice above some 10x over”. Well, sure, if you could somehow force all the water (especially that 0C to 0.8C water below 200 meters) to mix with the surface water directly under the ice. But that is not at all what appears to be happening, because there is obviously stable stratification. If you consider the top ~100-120 meters as most likely involved in seasonal ice melting (which makes sense considering how deeply solar energy penetrates in the ocean), and assume the average temperature of that water is about -0.5C, then all the available heat in that 120 meter layer (from -0.5C to -1.5C, or about 1 calorie per gram), could potentially melt ~1.5 meters of ice, not 10 times the surface ice thickness. Even this estimate of available heat for melting is physically unrealistic, because it assumes the top 120 meters could be made uniform in temperature (and so salinity!), which is not likely to happen.
    .
    Additional solar warming due to ice loss (more open ocean area in the summer months) will almost certainly cause more heat to be accumulated in the top 100-120 meters during the summer months. This heat will delay the onset of ice formation in the autumn, since that additional heat will need to be lost from the water surface before ice can start to form. And of course, this will also cause thinner 1-season ice, since there is less time for ice to thicken with a shorter period of freezing.
    .
    My objection to your original comment was the suggestion of multi-seasonal carry-over of heat, which melts the ice from below. This still makes little sense to me on purely physical grounds, and I have seen no data (so far) which supports that suggestion.
    .
    As to whether or not my view of the Arctic ocean is ‘idealistic’: Sure, I am not a specialist in Arctic ocean temperature profiles. But I know enough to see when a proposed mechanism appears clearly inconsistent with basic physical processes like thermal convection.

65. Andrew_FL @ Comment #80863, this link appears to work. The excerpt below is from that article. I would think it fits well with what Steve Mosher indicated.

    It would be difficult for me to believe that Mosher would be facetious about anything related to AGW and besides I am sure he intently followed my links to Arctic ice models as he is a very good student.

    http://www.sciencedaily.com/releases/2011/08/110811113956.htm

    “Kay and her colleagues also ran computer simulations to answer a fundamental question: why did Arctic sea ice melt far more rapidly in the late 20th century than projected by computer models? By analyzing multiple realizations of the 20th century from a single climate model, they attribute approximately half the observed decline to human emissions of greenhouse gases, and the other half to climate variability.”

66. “It would be difficult for me to believe that Mosher would be facetious about anything related to AGW”

    LOL

    Andrew

67. Carrick (Comment #80879)-“I don’t think the data prior to 1960 are reliable for the Arctic”

    Could you explain why that would be?

    “I’ll look at the Northern Atlantic data, and see what I see, but when you start narrowing it down to too small of a geographical region, you are more strongly weighting regional-scale fluctuations associated with long-peirod oscillations. You just may be looking at AMO.”

    Actually, that’s my entire point. I don’t expect the Pacific Ocean to influence sea ice much, mostly the Atlantic Ocean. And since the Atlantic experienced anomalous extra warming due apparently to natural variability, it seems to me that the decline in ice is exaggerated by that natural, regional influence.

    As Mosher will no doubt point out, incidentally, there is no evidence that the Atlantic truly “oscillates”. It has variability, that variability sometimes looks “oscillatory” or even “periodic” but really we don’t know what the true nature of the phenomenon is. Which is why I simply refer to it as such. A “phenomenon”, sticking to the observational facts.

68. I apologize for not seeing Mosher’s post on what he thought were his thoughts on the Andrew_FL facetious contention. I am going with my thoughts on Mosher’s thoughts.

69. As Mosher will no doubt point out, incidentally, there is no evidence that the Atlantic truly “oscillates”. It has variability, that variability sometimes looks “oscillatory” or even “periodic” but really we don’t know what the true nature of the phenomenon is. Which is why I simply refer to it as such. A “phenomenon”, sticking to the observational facts.

    ####

    crap am I that predictable?

70. steven mosher (Comment #80886)-Nope, I’m just the Andrew you used to get into arguments with over at CA. I’ve since learned to appreciate your wisdom. Those were the days of my impudent youth.

71. Andrew_FL:

    Could you explain why that would be?

    If you think the geographical coverage of the arctic is poor now, by that standard it would be nonexistent before say 1950. In other words, too poor of a coverage for the average to be meaningful.

    As Mosher will no doubt point out, incidentally, there is no evidence that the Atlantic truly “oscillates”.

    That would be news to the people who study the “North Atlantic Oscillation.” Note its period is somewhere between 50-60 years in duration and it can be seen in long-interval Greenland ice core proxy data.

    Speaking of Steve Mosher, one of his strictures is that we apply consistent rules to how we analyze data. If we are worried for example about recent coverage of temperatures in the arctic, we shouldn’t simultaneously use much poorer quality data to argue one way or the other.

    Steve and I would agree that first you establish how reliable the data is, then you analyze it, then you try and draw conclusions. You don’t (and I’m not saying you do, but others do) look at which data gives conclusions we like, then argue which data should be included based on that.

72. Also, Andrew_FL, the term “oscillatory” is used to describe periodic and quasiperiodic phenomenon, and that’s certainly the case for the NAO. It doesn’t require a harmonic oscillator equation before we can use that term.

73. Carrick (Comment #80889)-That the coverage is poor is a fair point, it may be a problem, I try to assess the data at face value. As I have said in the past, lack of coverage in the Arctic concerns me less than the quality of data in much of the Third World. I am consistent I believe in the way I look at the data. If it is different than how others look at it, well I am not those other people 🙂

    BTW If we are going to get into a debate over “oscillations” I will be content to watch from the sidelines. Others far more knowledgable than me could contribute much more.

74. Also, with regard to explanations of enhanced high latitude effects, please consider the fact that the strongest trends are associated with dry cold winter air masses, especially the Siberian anticyclone. This seems to me disconnected from the usual explanations of amplification but is an interesting effect and I would love to hear anyone’s thoughts as to the reason for it. See this map:
    http://i23.photobucket.com/albums/b370/gatemaster99/GISSNHCOLD.png

    Note the difference from this one:
    http://i23.photobucket.com/albums/b370/gatemaster99/GISSNHWARM.png

75. If I had to guess, the biggest effects of the loss of ice will be seen as a shortened period of covered ice (so the effect should be seen most clearly on the intervals MAM and SON). It’d be interesting to test the data by quarter. I’d still probably integrate over latitudinal angle were I to do it, the lat-laong is a bit too… messy for me.

    It may be that the largest effects isn’t the direct albedo but the influence of the change in albedo (e.g., ice cover) on weather patterns. I think that was seen last fall when the Hudson Bay froze late.

76. SteveF, I’m glad that you are looking into ITP data to get a better understanding of the temperature profiles in the Arctic ocean.
    In fact, I don’t think we disagree much, maybe only in the details.

    Your original question was “how much additional heat is lost to space due to 1.5 million sq Km of open ocean compared to ice”.

    Since this question is essential for the outcome of this (Connolley-Dekker) bet, and I realize that I’ve not been very complete in my answer to this question, here is a better overview :

    Because open ocean has a lower albedo than ice, it absorbs more solar heat than ice covered ocean. The question is, what happens with that excess heat. Here are a number of effects :

    1) During the summer, excess heat accumulated by open ocean causes neighboring ice to melt, creating more open ocean, and resulting in reduced ice volume and extent.

    2) End of the summer, warm open ocean water continues to bottom-melt neighboring ice, resulting in a prolonged melting season, and reduced ice volume and extent.

    3) Since it takes more time to cool down ocean water, freeze-over will happen later, and thus will reduce ice growth, resulting in thinner first-year ice the next year and thus reduced ice volume in the next year. (This is the effect you explain above in good detail).

    4) When freeze-over occurs, some accumulated heat will still be trapped at greater depth (albeit above the thermocline) which will cause prolonged bottom-melt of neighboring ice, and slow down in first-year ice growth during fall and winter, and thus reduced ice volume in the next year. (This is the effect that you dispute). We can argue about this for a long time, and I can show you ITP profiles and FluxBouy data and mass-balance data that confirms the effect, but since it essentially is not much different than the effect described in (3) (only postponing the heat transfer to later bottom-melt) I wonder if it makes sense to continue the arguments. As I mentioned before, I don’t think we differ much in opinion, only in the details.

    5) Thinner ice will make ice more vulnerable to melt and break-up in proceeding years, leading to yet more open ocean and thus amplification of the 4 effects above.

    6) Open ocean causes more turbulence in surface water, which (if open ocean is large enough) may disrupt salinity levels to depths down to the thermocline, and thus may cause deep-ocean heat to be released. In other words, the Arctic may become less stratified. There is some evidence for this occurring already, although not enough research has been done yet on quantifying the effect now, and modeling effects expected in the future if open ocean during the Arctic becomes more wide spread.

    7) Brine rejection by freezing first-year ice is more profound than brine rejection of multi-year ice, which suggests less stratification even during the winter.

    Note that all these effects seem to have a “positive feedback” effects. Open ocean causes more open ocean or thinning of ice, which will lead to more vulnerable ice next year.

    Since the Arctic is the coldest place around, the only real way to loose heat is by increased radiation to space. That (Stefan Bolzmann black-body radiation) effect is a strong negative feedback, since it will resist temperatures to be higher.
    However, again because the Arctic is the coldest place around, it is also a dumping ground for all of the Northern Hemisphere’s excess heat. Whatever excess heat (due to AGW or otherwise) did not get radiated away at lower latitudes, will end up heating the Arctic.

    As a result of this, we see the Arctic heat up.

    Black-body radiation to space (as dampened by GHGs that are also present above the Arctic, just like anywhere) will make sure the Arctic winter is in balance, which appears to result in a 1 C increase/decade for Arctic winter IIRC. However, the Arctic summer CANNOT heat up unless ice melts : Any attempt to get the summer temps above 0 C will result in ice melting. This ice melting will continue until the Arctic summer will heat up about as much as the Arctic winter is now heating up. Now that ice volume has very little ‘buffer’ left over to absorb excess summer heat, this means that substantial areas of the Arctic would have to become ice free in summer for the radiative balance to return.

    And this effect is not linear, but it is amplified by the 7 points (positive feedback factors) described above.

    Another indication that this is happening is that ice extent/area graphs as well as ice volume graphs over time are accelerating.

    And that is what I think is happening, and that is why I think there is a very good chance that ice cover will be less than 3 million km^2 in 2016. And that’s why I made this bet.

77. Carrick, Andrew FL, Steven Mosher,
    Regarding AGW or “natural varibility”, Kay et al concludes that some 50% of the sea ice extent decline over the decades can be explained by natural variability, while 50% can’t.

    Note that this does not mean that 50% IS caused by natural variability. In fact, it may very well be that 100% of the decline is caused by AGW, and natural variability is just doing it’s normal stochastic variability work.

    Owen wrote : Nonetheless, natural variability is not exempt from causality

    I think this is a point very well taken.

    If natural variability is at work here (for 50% or less) then which natural variability force is at work here, and do we have any evidence for that (quite amazing power) that causes 50% of the decline of the Arctic sea ice over the past decades ?

78. Carrick (Comment #80894)-You’ll lose valuable information by eliminating the longtitudinal variations, I think. Try plotting up the gridcell trend against it’s seasonal surface pressure climatology, for dry (low humidity) land gridcells, there will be a significant correlation between higher pressure and larger trend for the cold half of the year.

79. “it may very well be that 100% of the decline is caused by AGW”

    It also may be that 100% of the decline is not caused by AGW.

    Andrew

80. Rob Dekker,
    .
    Thank you for your more detailed comments. As you said, we probably disagree only about a few details, not the basics of what is going on. We should probably agree to disagree about the post-freeze bottom melt question.
    .
    With regard to future loss of salinity stratification: for certain more open water will generate larger waves (assuming there is much wind), and so some increase in surface turbulence. However, the known depth of the well mixed layer globally is usually very modest (~60-70 meters), even where there is enormous surface wave activity. If I remember correctly, wave-driven motion does not penetrate much deeper than ~1.5 – 2 times the distance between wave crests, which restricts wave driven mixing to <40 meters under most circumstances.
    .
    I have only one other technical comment: It is for certain correct that Arctic ocean areas which continue to have floating ice in summer will not warm significantly above the melting point, and the ice floe tethered profilers shows this. However, I think it is important to consider that large, fully ice free areas will in fact warm considerably (at least in the top ~10 meters), which will increase radiative and evaporative heat loss rates, and perhaps more importantly, may produce more cloud cover and substantially changed heat balance in the later part of the melt season.
    .
    I do not know if you will win your bet or not, although I suspect that natural variability may determine the outcome, either way. I certainly would not place such a bet.

81. SteveF–
    Other than the most outlandish bets, natural variability comes greatly into play. In this particular bet it appears Connolley is counting on the ice melt reverting to the linear decay. If so, he will have 1/2 a chance of winning Rob’s money and very close to 1/2 a chance for a draw. Which happens would be purely due to “weather noise”. So, if Connolley is actually right, things are in his favor but weather will determine if he wins or just fails to lose..

    The down side for him in terms of my computed odds is the data just don’t seem to be saying “linear”. That’s not to say it can’t revert to linear and more over, I didn’t even try to look at anything like “linear with 50 to 60 year oscillation” .

    The outcome on the odds computation is heavily dependent on the functional forms one thinks makes sense. In this case, I fit would I could: polynomials and the Gomprets. I haven’t found other fits I can easily use in R.

82. SteveF (Comment #80881)
    August 28th, 2011 at 9:44 am
    —————————–
    Steve, When I read this post (and your others on this topic), all of which make good sense, I come away with the feeling that we really should not be seeing this 30+ year period of gradual but sustained loss of sea ice. You make a strong case that presence of open water does not store thermal energy in the surface layer, and that ice freezes anew each year almost independently of its previous melt history.

83. Lucia,

    I understand your selection of statistical model fits.
    .
    Still, I would only make a bet like this based on a model fit that comes from sound mechanistic rational. AFAICT, nobody is in a position to generate such a model.

84. Owen #80907,
    I find the global sea ice trend more convincing than the Arctic trend by itself. The global loss is the 1.5 million Km^2 value that I originally commented on.
    There is lots of evidence of a long term cyclical see-saw of temperatures between the Antarctic and the Arctic, and it does appear (if only by chance) to be roughly in synch with the ~60-70 year cycle evident in the global average temperature trend. I suspect (but can’t prove) that the relatively rapid decline in Arctic ice cover over the last 20-30 years is in part related to this see-saw cycle, especially since there has been no loss of southern hemisphere sea ice, and in fact a slight increase over the same period. We ought not expect GHG driven warming to be overwhelmingly strong in the Arctic and barely visible in the Antarctic (OK, outside of the peninsula region). I would urge people looking at the Arctic sea ice trends not to be tempted to assign all (and perhaps not even a majority) of that trend to GHG driven warming. Extrapolating an existing trend over an extended period when the underlying mechanism controlling that trend is not clearly understood is fraught with the chance of (comically large) error.

85. SteveF–
    I would bound things to avoid wishful thinking or over confidence my still-un-proven mechanistic theory is right. So for example, if I had a theory ice extent must revert to the linear trend, I’d look at my version of the odds estimate using sea ice extent (rather than volume) see that that linear fit is currently not following the data and that others look better and I would be very cautious about believing I would win a bet based on the assumption the NH ice trajectory is going to revert to linear by 2016. It may well be true that the sea ice will start to increase at some point, but I don’t think anyone knows the lower bounds it might hit or the latest date when it might turn around if it does.

    On the other hand, if Rob was begging to bet me the ice would fall below 1 million km in 2016, I might be tempted to take that bet. The trajectories don’t look like they are going there and as far as I am aware, if anything, the physics says it’s possible for the ice to increase.

86. Lucia #80911,
    If I had a theory of sea ice extent, I would also be very (very!) cautious about placing bets based on that theory unless it was extremely robust with hind-casts. I would also look at the recent trend and see if there is noise/short term variation which might make me lose the bet, even if my model were correct. Most likely, I would never bet on something so noisy as Arctic sea ice, except in the very long term… but then I would likely be dead. 😉
    .
    What I thing we have in the bet is a pair of SWAGS from two friends, both of whom can afford to lose the money. I could be wrong of course… maybe they loath each other, but I doubt it.
    .
    My SWAG: The trend will slow in the next few years and probably reverse a bit. Neither is likely to win any money.

87. Re: lucia (Aug 29 11:47),

    The September average is closely tied to the annual minimum. The plot of annual minimum area shows a break point in 2007. The trend starting in 2007 does not appear to be the same as before. With this few points and the large amount of noise, however, one can’t rule out a null hypothesis of no trend. The 1979-2006 trend is within the confidence limits as well so one can’t say the trend is less than before either.

    Of course that’s area. Extent will be greater. In spite of what looks like very low concentration during recent Septembers, the trend for the monthly average is not significantly different from zero and there are no significant outliers…yet. I haven’t looked at August yet, but the bet is for September, not August.

88. SteveF–
    I think they must like to bet; $10,000 may not break someone, but bets for fun tend to be something like $50.

89. The trend starting in 2007 does not appear to be the same as before. With this few points and the large amount of noise, however, one can’t rule out a null hypothesis of no trend. The 1979-2006 trend is within the confidence limits as well so one can’t say the trend is less than before either.

    Can we put AICc criteria on models with break points? I’d like to add that sort of thing. It would be fun to look at.

90. Lucia,

    I think they must like to bet; $10,000 may not break someone, but bets for fun tend to be something like $50.

    Who knows, it could all be a publicity stunt and no money will ever change hands.

91. Lucia,

    Now if it were Romm v. someone else (anyone!), we would know that he for certain loaths the other bettor, and is at the same time looking for publicity.
    .
    Loathing anyone who disagrees with you seems a common trait among those on the left. Actually, come to think of it, those on the left seem to loath most everyone. Seems a stark contrast with someone like Ronald Reagan. 😉

92. SteveF–
    Romm ultimately did bet Connolley, Annan and Brian Schmidt. I don’t think he loathes them. But I also think he didn’t originally intend to bet them.

93. Re: SteveF (Aug 29 11:23),

    As an indicator of overall warming, I prefer a global index as well. It does look like the rate of decrease of global area is accelerating. The quadratic fit is not meant to be a real model. It’s also smoothed data, which is evil too.

94. DeWitt,

    Smoothing to see a trend does tend to raise doubts about the true size of that trend…. not sure about the ‘evil’ part. ( 🙂 )

95. Lucia,

    But I also think he didn’t originally intend to bet them.

    No, he was probably wishing for Anthony, with bet terms that made it impossible for him (Joe) to lose the bet. Too bad you didn’t have enough site traffic to quality under Joe’s odd rules for who may bet against him.

96. SteveF (Comment #80920)
    August 29th, 2011 at 6:35 pm
    “Smoothing to see a trend does tend to raise doubts about the true size of that trend”
    ————————————————-
    If the trend is a low frequency signal (or is not an oscillating signal), and the noise is of higher frequency and is reasonably random, should not the smoothing process be expected to accurately represent the trend?

97. Lucia:

    Can we put AICc criteria on models with break points? I’d like to add that sort of thing. It would be fun to look at.

    You certainly can!

    Of course, in this case, we have other information. For one, it’s more than four data points (more like 4 years x 12 months), for a second, the anomalization process demonstrates that a “phase shift” occurred circa 2007….

98. Carrick-
    What I want to do is use AICc to weight proposed fits. So, I want to put AICc’s on Dewitts fit to the 30 or so 1979-now sept nh extent data. I don’t think the phase shift gets involved. It might give someone who is aware of it more confidence, but it’s not going to come out of the sept nh extent data vs. year.

99. Owen (Comment #80922)-I begin to see why you and many others expect “explanations” for all trends. You assume that climate is signal + white noise. Especially for regional variations that is much too naive a noise model. A more reasonable model would be signal plus red noise. If “weather noise” were really “random” in the since I take you to mean, it would be the case that one could just average it out. If however, as makes more sense, the weather is auto-correlated, and hence the weather “noise” is not random in the sense of each bit of weather being independent and identically distributed about the “signal”, then no, no amount of smoothing will only include the “signal” it will include some aspect of “noise”, too.

100. Andrew_FL (Comment #80925),
     “If however, as makes more sense, the weather is auto-correlated, and hence the weather “noise” is not random in the sense of each bit of weather being independent and identically distributed about the “signal”, then no, no amount of smoothing will only include the “signal” it will include some aspect of “noise”, too.”
     ————————————————-
     If the smoothing window is big enough to remove oscillating noise of certain frequencies and above, much of the noise can be removed. I would agree that markedly non-random forms of noise will find their way into the smoothed signal. But all forms of noise find their way into unsmoothed signal.

     Not quite sure how this relates for my looking for explanations for trends such as natural variability. If we believe that non-cyclic natural variation is driving a climate change, we ought to be able to postulate mechanisms that could be used to model such processes. We ask for no less of the current climate scientists who claim anthropogenic causes.

101. Re: Carrick (Aug 29 19:45),

     For one, it’s more than four data points (more like 4 years x 12 months)

     I don’t see any evidence of a shift in the maximum area. 2008,9,10 were right on the 1979-2006 trend line. It would be about 5 months/year rather than 12 according to this plot. Using the whole year would reduce the size of the effect. The linear trend would be ~0.15 Mm2 over 4 years.

102. Owen–
     Out of curiosity, do you consider ENSO “cyclic” or “non-cyclic”? I can’t understand exactly what you mean by “non-cyclic natural variation is driving a climate change” unless I know which things you consider cyclic. Also, not knowing what you consider “cyclic”, I don’t know whether people are necessarily postulating “non-cyclic” natural variation.

     Climate scientists do postulates mechanisms for some (probably all) named oscillations. But even if they don’t know the mechanism, some, like ENSO clearly exist.

103. Lucia,

     I consider ENSO to be cyclic, but not entirely regular, I consider the AMO to be cyclic. I consider anthropogenic CO2 forcing and volcanic forcing to be non-cyclic. If we smooth the average global temperature data since 1900 (or even if we do not smooth) we see what appears to be a (generally) monotonically increasing non-cyclic signal (unless the frequency is really low) (http://www.woodfortrees.org/plot/gistemp/from:1900/mean:120).

     If this temperature increase is due to natural variability, it cannot be due to cyclic processes (unless they have a very long period). If not cyclic, then the energy balance on earth has been changed and some mechanism must have affected that balance. If it is claimed to be natural variability, it is incumbent on those making the claim to postulate some cause. Otherwise the claim is attributed to magic.

104. Lucia,

     I should add that I see cyclic processes as involving intra-planetary and reversible transfers of energy from one part of the climate system to another (e.g., ENSO). They have no net effect on the energy balance.

105. DeWitt, I’m referring to this pattern, which certainly involves more than four points (2007-current):

     see this

     You can perform a spectral analysis comparing this period to the previous (1972-2006):

     see this.

     Clearly there has been a shift in the phase of the oscillation, so that when you subtract the averaged behavior from previous years, it no longer is canceling the annual component + its harmonics.

     While I agree you see the effect of the shift mostly in summer months (it’s hardly surprising that winter doesn’t change), try and perform a spectral analysis without the winter months! I think the pattern includes them too.

106. Owen:

     If this temperature increase is due to natural variability, it cannot be due to cyclic processes (unless they have a very long period).

     Not necessarily, nonlinearities inherent to parametric forcing allow for shifts in DC offsets too. If you had an increase in the amplitude of parametric forcing, this would give rise to an apparent increase in the “secular” (non-cyclical/aperiodic) component of the total forcing.

107. Not all trends must have a “cause”. One needs to know the probility of trends arising by chance (have the appropriate noise model) before one can claim a trend requires anything to explain part of it. Even if one then determines that the change is large enough that a purely random trend is unlikely, this does not in fact mean the entire trend requires an explanation. Some of the trend could still be by chance. One has to show that this ISN’T the case, one does not have to show that such IS the case. The mere fact that it could be the case nevertheless gets complete ignored because of a belief that 100% of every trend always needs an explanation. For my part I have explained that I think there is a regional influence which could have just as easily acted to hide the trend but exaggerated it instead. I have even described why it is an exaggerating influence. It is neither due to some cycle nor is it “forced” in my opinion. It is regional scale noise.

108. Re: Andrew_FL (Aug 29 21:56),

     Just because you can’t predict it doesn’t mean it’s random. Chaotic behavior may be completely deterministic.

109. Wow. You guys sure have time to talk. Don’t you have a family and a job ?

110. SteveF,
     About ITP buoy data, you show ITP5 data :
     http://www.whoi.edu/cms/images/itp5dat3f_81104.jpg

     This buoy was placed under thick ice which did not melt out in the first summer. In the second summer (2007) it’s ice was melted out by highly salty warm water (meaning it ended up in wide open ocean) just before it’s transmissions ended.

     It does not show what happens after freeze starts again, but ITP23 does :

     http://www.whoi.edu/page.do?pid=25636

     The stream starts more or less where ITP5 ended : Salty warm water mixed through the upper 50 meters, drifted under more fresh water from late summer melt. This heat sits there until Dec, when salt-rejection from the ice above allows the heat to reach the bottom of the ice, causing the postponed bottom-melt that I was referring to in my point (4) in my post of 7 positive feedback mechanisms.
     See the same thing in this profile :
     http://www.whoi.edu/page.do?pid=41795&tid=441&cid=106814&ct=61&article=72312
     Notice the cold/fresh water at the surface at the right side of the temp graph, locking the warm saltier ‘open ocean’ water below it.

     Another indication that heat flux below the ice does not stop after freeze-over, is shown in ice freeze profiles from IMB bouys (measuring ice thickness). Here is a great example :
     http://imb.crrel.usace.army.mil/2006Csum.htm

     In this one, notice that bottom-melt continues until Novenber, long after after snow covers the ice and air temps are falling to -10 C. And notice that bottom-freeze stops and turns to bottom-melt as early as May. With air temps back at -10 C, where would the heat come from that melts 3.5 meter ice from below if not from below ?

     Needless to say, the statement that “It is physically impossible for warmth to “accumulate” under the sea ice in the arctic” is simply not correct. It can, and it does.

     Now as I noted before, I think we only disagree on the details, since this is only ONE point from the 7 positive feedback factors that I stated in my post.

     And note that feedback factor will ‘accellerate’ any linear trend, which is exactly what we are seeing in Arctic sea ice area, extent and most profoundly volume data.

     So unless any negative feedback factor starts to show itself rapidly, I am going to win my bet.

111. Let me add one more comment.

     Are you guys surprised that when it comes to AGW effects, that scientists after the fact are claiming “it was worse than we anticipated” ?

     Well, Arctic summer sea ice extent is one of these effects that seem to be “worse than we anticipated”.
     http://www.realclimate.org/images/seaice10.jpg

     The median IPCC projection for 2011 minimum sea ice extent is 7 million km^2.

     Reality shows an extent playing tag with the 2007 record, heading for 4.5 million km^2, open ocean to 85 N in some places, both NW and NE passage wide open, and a pack in the Central Arctic less than 1 meter thick, allowing the Healy to steam ahead at standard cruising speed all the way to the North Pole, and all this while there were NO 2007 conditions this year : no excessive Fram strait export, nor excessive solar irradiance due to reduced cloud cover. This year, ice was so thin that simple, average weather conditions melted the pack in place, and still tracks 2007 record.

     Rather that responding to red-herrings about global ice cover and the stochastics of chaos, is this kind of basic reality recognized here at Lucia’s ?

112. Rob:

     “Are you guys surprised that when it comes to AGW effects, that scientists after the fact are claiming “it was worse than we anticipated” ?

     no. Given the complexity of the models and there utter lack of regional skill I think its a certainty that for some areas it will be worse than expected, for some areas it will be better. On average the models are high, so its not been as worse as they expected.

     What to make of that?

     Well you can point to a region where they got it really wrong and conclude…err.. conclude..errr.. conclude… that they got it really wrong in that area. ya.

113. “it was worse than we anticipated”

     Translation: We got that wrong as well.

114. That IPPC sea ice chart is made from non-factual data. The minimum sea ice extent was not 8.0M km2 in the early 1970s. It increased throughout the 1970s reaching a peak in 1980 at 7M km2. It was stable until at least 1996, which was another peak year similar to 1980.

     Many suspect that the extent was lower in the 1940s, rose in the cold-for-the-Arctic 1950s and 1960s but there is no real data that suggests the minimum was so much higher in the 1950s and 1960s. I have data going back to 1947 when the air defence bases were set-up in the Arctic and there is no real change in sea ice or sea ice thickness since then.

     It is another rewriting history chart that the IPCC and this science is prone to do.
     http://www.arctic.gov/publications/AMSA/history.pdf

115. Rob–
     Some things are worse, some things aren’t. Many things (like hurricanes) are in “can’t even tell” territory.

     I think global surface temperatures are much more important than sea ice. Moreover, given recent papers we now aren’t even sure the ice melt is entirely due to AGW. It may just be much more variable than previously thought.

     So, given the two things we actually know: temperatures and sea ice, I would assess things as not worse. The not very important thing has progressed faster, the important thing slower.

     I have a question on this
     
     If we made a similar graph for temperature– showing model means instead of runs–we’d be told we can’t compare it that way. We need to compare to the full ensemble of runs. So…. why should that be meaningful for ice?

116. Rather that responding to red-herrings about global ice cover and the stochastics of chaos, is this kind of basic reality recognized here at Lucia’s ?

     Sorry, I missed the direct question.

     I think AGW is real. I think ice has melted faster than projected by the multi-model mean and is outside the spread of the multi-model mean– which means, yes, it’s outside the spread. And there has been more melting.

     I haven’t looked at whether the multi-model mean of the ice is inside the estimated uncertainty intervals based on the “ice-weather noise”, but I can. (Maybe. As projection testing, it’s complicated by the fact that the forecast looks somewhat curved and I only have a graph.) I guess I’ll do that.

117. Rob:

     “Are you guys surprised that when it comes to AGW effects, that scientists after the fact are claiming “it was worse than we anticipated

     Given this is how they get more funding, no. Given that this is how they undercut their trust and respect from other scientists, yes.

     Speaking of realities, I haven’t seen any evidence you have any particular claim to have a remarkably better grasp of it than anybody else. By the way, anybody who says things like this:

     Wow. You guys sure have time to talk. Don’t you have a family and a job

     deserves no respect. I doubt it is going to make anybody think anything besides “wow this guy sure is an a**hole.”

118. You don’t have to respect him. He seems to know what he’s talking about.

119. Andrew_FL (Comment #80933)
     “Not all trends must have a “cause”. ”
     ——————————————-
     Well a simultaneously warming atmosphere, warming ocean, and melting land and sea ice do require an global energy imbalance. Otherwise, for example, the melting ice would drain thermal energy from the atmosphere and ocean. Asserting chaotic behavior does not free us from cause and effect.

120. bugs:

     You don’t have to respect him. He seems to know what he’s talking about.

     Still reserving judgement about that. He speaks a certain “group speak”, which in itself, puts him in no particular position of authority with me. Obviously he’s done more reading on this, whether he’s done the right reading remains to be seen. The fact he gets snitty when pushed to explain the science, speaks of somebody who may have reached the limits of what he actual knows and is able to defend.

     (That defensiveness is a common characteristic of True Believers of anything. They have a list of things that they believe religiously, and when these get challenged, it makes them behave very ugly.)

121. Rob Dekker #80936,

     I could not get the links you put in your post to come up (sever not responding). I will look again later.
     .
     WRT to family and other obligations: I am sure most people in fact do have these obligations, as I am sure you do. Questioning someone else’s choice of time allocation, especially someone you do not know (as you seem to be doing), strikes me as inappropriate. And that has nothing to do with who you are or what you do for a living. If Joe the plumber made the same comment, it would be just as inappropriate.
     .
     That said, I do appreciate you taking the time to technically engage people on the actual science involved. I hope you do not think this engagement is a waste of your time.

122. Lucia could you link the data if you have it for this figure?

     I’ve been looking for a link to the model predictions for a while.

     TUVM.

123. Rob,
     One more comment. You said:

     Rather that responding to red-herrings about global ice cover and the stochastics of chaos, is this kind of basic reality recognized here at Lucia’s ?

     I am not certain what to make of that question. Evaluating a measured trend in light of chaotic variation (at all time scales) seems to me a perfectly normal thing to do with noisy data. Considering the apparent discrepancy between sea ice cover changes in the Arctic versus Antarctic also strikes me as a perfectly reasonable way to try to better understand what is happening in the Arctic. I am not at all certain why you think those things qualify as ‘red herrings’.

124. Re: lucia (Comment #80943)

     I think global surface temperatures are much more important than sea ice. Moreover, given recent papers we now aren’t even sure the ice melt is entirely due to AGW. It may just be much more variable than previously thought.

     Let us not forget the Medieval Warm Period (when Greenland was actually green), which consensus climate science is determined to present as strictly a local phenomenon. To me, this seems an implicit admission that a relatively ice-free Arctic need not have anything to do with global warming, in principle.

     But I would disagree with the first part of Lucia’s statement. The global surface temperature anomaly is, after all, only an abstraction. It is not even clear to me whether it is meaningful at all to the precision that it is often quoted (say, a tenth of a degree). It is certainly not the temperature anywhere on earth; and, by itself, it does not tell you anything about, for instance, how much extra heat is, or is not, stored in the oceans–which is the critically important energy reservoir.

     Sea ice, on the other hand, is real and tangible, and its presence or absence can have important consequences. The problem is that (as argued just above) its connection to global warming in general, and greenhouse gases in particular, appears somewhat indirect, and, as the models show, it seems to be still poorly understood.

125. Owen (Comment #80947)-“Well a simultaneously warming atmosphere, warming ocean, and melting land and sea ice do require an global energy imbalance. Otherwise, for example, the melting ice would drain thermal energy from the atmosphere and ocean. Asserting chaotic behavior does not free us from cause and effect.”

     Now you are moving the goal posts. I thought we were discussing Arctic sea ice, not global changes in energy balance.

     But you still completely miss the point. Now you assume that the global energy balance is always signal + white noise. People who can’t imagine how the energy balance could vary like red noise, without any external “cause” are pretty naive.

126. Re: Andrew_FL (Aug 30 08:56),

     Chaotic is not stochastic. Saying it’s signal plus red (or white) noise doesn’t make it so. It’s all signal. All trends in a deterministic chaotic system like weather have causes. You just can’t predict them in advance or expect them to continue forever.

127. julio:

     Sea ice, on the other hand, is real and tangible, and its presence or absence can have important consequences

     Yep. Take for example what happens when the Hudson Bay freezes late (last fall). Like you said, it’s a very complex part of the system, and I’m glad to have the sense to stay away from trying to model, let alone predict in the absence of usable models.

128. DeWitt Payne (Comment #80955)-For the usual definition of “signal” yes. As usual when one is speaking CS English, those used to speaking normal English won’t get your meaning. I was speaking CS English because that is the language Owen and others like him speak.

     In CS English, “signal” is synonymous with “forced variation” or even “AGW signal”. It is not the usual definition of the term at all.

     As for the issue of causality, again I think we are speaking of different notions of causation. All weather has immediate causes (they are “caused” by prior conditions of other variables that also make up the weather.) Some weather has no true “final” cause-It’s weather patterns all the way down. In CS English, “signal” means that component of variations which does have a final cause.

129. Re: Andrew_FL (Aug 30 11:26),

     In CS English, “signal” means that component of variations which does have a final cause.

     Even so, the chaotic part of weather isn’t stochastic noise, red or otherwise. I should probably say ‘may not be’ rather than ‘isn’t’. The really nasty part of chaotic behavior is that the system response to a forcing may not be predictable either. It depends on the topology(?) of the system, which isn’t known and isn’t likely to be known as the chaos is both spatial and temporal with very high dimensionality. There can also be quasi-periodic oscillations at multiple time scales too, to make life more interesting.

     Even though it may not be stochastic noise of whatever color, the end result is the same. Variation that looks like a forced trend may not be.

130. DeWitt Payne (Comment #80958)-Heh, yes, fair points. I see that we essentially agree that:

     “Even though it may not be stochastic noise of whatever color, the end result is the same. Variation that looks like a forced trend may not be.”

     This is essentially the point I am trying to make. One needs to consider all the potential things that could confuse a person into thinking they are seeing a “forced trend” when some significant part of it, or sometimes all of it, is not. Noise of some color is just one possible source of confusion.

131. Re: Carrick (Aug 29 21:27),

     Clearly there has been a shift in the phase of the oscillation, so that when you subtract the averaged behavior from previous years, it no longer is canceling the annual component + its harmonics.

     Don’t you mean amplitude rather than phase?

     The question is: what’s the trend in the amplitude of the peak at 1/year? I’m not convinced that you have more than 3 degrees of freedom (four data points), if that, for that estimate. You have lots of points to determine the amplitude for a given year, but that doesn’t help with the trend, I think.

132. Rob Dekker,

     I finally got a look at the data you linked to in your earlier comment.
     .
     The first link http://www.whoi.edu/page.do?pid=25636 does not seem to me to show much of anything except very cold (-1.5C) water in the surface layer during pretty much the entire graphic. Perhaps you see something I do not.
     .
     The link http://www.whoi.edu/page.do?pid=41795&tid=441&cid=106814&ct=61&article=72312 shows an evolution of temperature which is not at all clear to me. Your comment “Notice the cold/fresh water at the surface at the right side of the temp graph, locking the warm saltier ‘open ocean’ water below it” may be correct, but honestly I am not sure that the graphic shows significant ‘trapping’ of heat. It does seem to show that the end of melt season and the start of freezing is complicated, with freezing-out of concentrated brine leading to a complicated profile, at least for a while.
     .
     The final link says: “In the month of August, bottom melting averaged 4 cm per day and reached maximum values of 11 cm per day in the last week of August, compared to characteristic averages of about 1 cm per day.” I think nobody disputes substantial bottom melt in August. The question is bottom melt after surface free-over in November.

133. Julio,

     Let us not forget the Medieval Warm Period (when Greenland was actually green), which consensus climate science is determined to present as strictly a local phenomenon. To me, this seems an implicit admission that a relatively ice-free Arctic need not have anything to do with global warming, in principle.

     .
     Hi Julio. The above could come from Jeff Id! 😉
     .
     You are right, of course, that there is no certainty WRT claims of GHG forcing causing all (or even a majority of) the loss of Arctic sea ice. Like most of climate science, the noise is large, the signal is not much larger, and the available data is relatively short term. I would not bet US$10,000 on such things.

134. Carrick–
     I should have added “I’d have to compare to what I can read by eyeballing the graph.” I don’t have the model data for ice extents. That image url resolves RC. Maybe Gavin knows where you can get that data.

135. Hi Steve 🙂

     Just to be clear, I am sure that AGW is a large contributor to the observed loss of Arctic ice. It’s just that I wouldn’t be surprised if “natural variability” played a large part too. This would be consistent with the relatively bad job the models are doing.

     I do view the warming of the Arctic with some concern, like any form of rapid climate change. An oscillatory component would be very good news.

     I wouldn’t bet $10,000 on anything–my wife would kill me first 🙂 But I give Rob credit for having the courage of his convictions, even though I would like for him to lose in this particular case!

136. julio

     I wouldn’t bet $10,000 on anything–my wife would kill me first 🙂 But I give Rob credit for …

     I note “the wife” aspect of julio’s comment. My question for Rob– are you single? I’ve been wondering this about people who bet $10,000 on line in general. It seems to me spouses would often find $10,000 online bets irksome.

137. Lucia,

     spouses would often find $10,000 online bets irksome

     An eloquently understated fact of life for the wedded.

138. DeWitt:

     Don’t you mean amplitude rather than phase?

     No, I was actually was thinking phase.

     The question is: what’s the trend in the amplitude of the peak at 1/year?

     Why do you want to take one point per year…. that seems to needlessly through away data. Or put another way, it starts with the conclusion you want to prove.

139. Re: Carrick (Aug 30 17:35),

     This doesn’t look like a phase shift to me.

     Why do you want to take one point per year

     I don’t want to, I don’t see any benefit from using multiple points. The monthly data for 2007 on are so noisy as to be useless. I doubt removing the large seasonal component will help all that much, since that will cost you degrees of freedom too. You’ll have to show me that you have actually gained anything compared to using a single annual figure like max-min.

140. DeWitt Payne,
     The 10 year linear trend (on NSIDC data) projects a 2011 minimum of 4.52 million km^2.

     Since we will end up very close to that next month, this means even if a ‘phase shift’ occurred in 2007, the 10 year trend line has already caught up with it. IOW, the 2007 dip was not statistically significant enough to call it a ‘phase shift’ after which one would need to wait-and-see what the trend would do.

141. DeWitt:

     This doesn’t look like a phase shift to me.

     The way you measure phase shifts isn’t by eyeballing them, but I guess you know that.

     I don’t want to, I don’t see any benefit from using multiple points

     Because there’s information contained in the frequency domain that you can’t get to from max/mins. Signal processing 101.

142. Rob:

     Since we will end up very close to that next month, this means even if a ‘phase shift’ occurred in 2007, the 10 year trend line has already caught up with it. IOW, the 2007 dip was not statistically significant enough to call it a ‘phase shift’ after which one would need to wait-and-see what the trend would do.

     Nobody is referring to the 2007 dip as a “phase shift”, just that’s where a “phase shift” may have occurred. And “caught up” and the existence of a linear (or other) trend has nothing at all to do with the discussion either.

143. julio, I wouldn’t bet $10,000 on anything–my wife would kill me first
     LOL.
     Think of it this way :
     First of all, this is a 5 year bet, so that’s a risk of only $2000/year, less than people typically spend buying stuff they don’t really need. Now calculate how many hours per year your spend on-line debating (time that you could have spent actually working for money). Divide $2000 by that number of hours, and you have a compensation per hour for your blog posts. That’s marginal right ?

     For me personal, this bet is not about the money. It is about recognizing that the Arctic is deteriorating much faster than we anticipated, and understanding the risk we are taking by ignoring the possibility that the Arctic may be much more sensitive to climate forcings than our models suggest.

     Remember that the Arctic is the Northern Hemisphere’s air conditioner. It keeps summers cool and winters warm and heavily influences weather patterns across the entire Northern Hemisphere year around. Weather patterns which may very well change drastically with an Arctic entering a state with significantly less ice, and potentially ice free in summer.

     I have publicly stated that I really hope that Connelly wins. Because if I am right, then the Arctic is in much worse shape than we thought, with potentially devastating results for Arctic wildlife and weather patterns all across the Northern Hemisphere.

144. Carrick,

     I think we’re talking past each other to some extent. Frequency domain analysis is not the only way to do time series analysis, not that I’m an expert. There are other tools. A power spectrum by itself doesn’t tell you how the different frequencies vary in amplitude with time. There may not have been much variation from 1979-2006, but that may not be true for 2007 onward. I’m also not sure that phase shift means the same thing to you as it does to me.

145. Re: Rob Dekker (Aug 30 21:53),

     The thing is, though, if you are correct, it’s way too late to do anything. It was already too late in 1988 when Hansen gave his testimony to Congressional committees.

146. DeWitt:

     I think we’re talking past each other to some extent.

     Agreed. That’s why I clarified what I was looking for.

     Frequency domain analysis is not the only way to do time series analysis, not that I’m an expert

     No, but it’s complimentary to the time-domain analysis that you are doing with your “decomposition”. Some tools are better at certain things that others. You don’t use ARIMA-based tools to analyze periodic behavior (and seasonal variability in ice extent is by any standard, periodic, even if complicated).

     You can also use a combination of Fourier + time domain (e.g., finite-window OLS fit to Fourier coefficients). E.g.,this

     It’s also important to specify the frequency range of the phenomenon you are interested in. That’s why Rob Dekker was so far off when he was trying to apply 10-year averaged slopes to something that in the end regarded sub-annual variability. It literally couldn’t have had less to do with what we were discussing.

     In practice you shouldn’t ignore either time or frequency analysis, they are both important in signal analysis.

147. DeWitt The thing is, though, if you are correct, it’s way too late to do anything

     If there is anything I learned in life, it is this :
     It can always get worse.

148. Carrick, I think we have been talking past each other too. I thought you did a Fourier analysis of Arctic sea ice, but I now see that you did this for Global Sea Ice cover.

     I agree with DeWitt : it does not look like a phase shift to me either. A phase shift of two countering sinus curves would give you multiple peaks in the Fourrier spectrum, not the single peak you are seeing. Also, a phase shift implies that peaks are off-set in time domain, which is not apparent in the two individual time series (Arctic and Antarctic).

     The appearance of the 0.25 magnitude peak at 1/year in the 2007-2011 graph is easily explained when you look at the time spectrum : The 2007 massive negative anomaly dominates the time series and with subsequent 2008 and 2009, 2010 negative anomalies forms a strong peak at 1/yr.

149. “I have publicly stated that I really hope that Connelly wins. Because if I am right, then the Arctic is in much worse shape than we thought, with potentially devastating results for Arctic wildlife and weather patterns all across the Northern Hemisphere.”

     Nice POV.

150. “the Arctic is in much worse shape than we thought”

     I see that someone has been propagandized.

     Andrew

151. Rob Dekker,

     If there is anything I learned in life, it is this :
     It can always get worse.

     .
     What I have learned in life is: things seldom turn out as badly as you fear… nor as good as you hope. These outcomes happen, but it is rare.

152. Things have changed so much in the Arctic in the last five years, that the Inuit have discovered they can paddle their whaling boats and kayaks in open water in the summer. Previously, they just dragged them across the ice and chopped holes through the ice to get to the whales and the Walrus.

     And the Polar Bears have discovered that they can swim for five days straight covering 800 kms. Previously they just paddled around in the melt ponds.

     The Narwhales, Beluga, Grey and Humpback whales have discovered what their blowholes are for. Previously they just breathed little air pockets trapped under the permanent sea ice.

     The Snow Goose and the Canada Goose, as well, have discovered that the Arctic archipelago is a great place to raise gooslings in the Spring. Previously they stayed farther south.

     The Muskox and the Arctic Fox and the Arctic Wolf has discovered that that their fur molts and changes color in the summer for the first time.

     On the other hand, the Sun sets at the north pole for six months starting Sept 22, just like it has always done for the last 4.4 billion years.

153. Bill Illis (Comment #81010)-“the Sun sets at the north pole for six months starting Sept 22, just like it has always done for the last 4.4 billion years.”

     I really don’t think the parameters of the Earth’s orbit and rotation are sufficiently stable over billions of years for a statement such as that to be made.

     Surely the changes of the axial tilt alone would be sufficient to render this statement inaccurate. The date would also shift, would it not, with precession of the equinoxes?

154. Rob, the analysis was of northern sea ice extent. The trouble with your original discussion was long-term trends are an example of a secular variation, I’m more interested in what is happening with the sub-annual variation. Nor is any discussion of a “phase shift” related to any mechanisms of why the secular variation has changed, nor whether it is “worse” or “better” than we thought.

     My original interest was seeing whether the shift in the ratio of old to new ice has affected the seasonal variation in ice, and it was partly motivated by noting the nonsubstracted variability seen at the end of the series. (Post 2007.)

     I’m confused about what “one peak” you are referring to. The last figure was a decomposition of the Fourier components of the GSFC+JAXA combined daily data. The spectra analysis of 2007-now has more than one peak in the anomalized residual. (The annual component is the largest but no surprise there. See for example the spectrum of the non-anomalized series. )

     From the Fourier component analysis I don’t see any evidence that <the variability at the end of the anomaly series can be explained simply by multiplying the seasonal variation by a constant.

     Part of the talking past each other with DeWitt and myself is he has insisted for some reason to look only at max/min for something that is intrinsically relates to a sub-annual phenomenon. His only linear regression analysis shows evidence for change in behavior too (the residuals for the last four years are all negative), he only gets four points because he is partitioning it into years

     He was asking a different question (relating to trends in the amplitude), but his approach gives the equivalent result as a more detailed analysis only if the various Fourier components retain the same amplitude and phase relationship among themselves—if you think of a sum over Fourier coefficients, where the minimum occurs in particular is a sensitive function of the relationship between amplitudes and phases of the individual components.

     So starting with the max-min is assuming that these relationships remain fixed. Otherwise max-min is telling you something different than simply the overall amplitude of the seasonal variation—and you certainly need to look at more than annual values to disambiguate these two alternatives, e.g. purely multiplicative change in the seasonal variation versus the more complex change the form of the seasonal variation itself, which is the explanation I favor.

155. Re: Rob Dekker (Comment #80995)

     Well, at least we both agree that we want you to lose the bet! 🙂

     One of the many challenges that the reality of AGW poses for the environmentalist movement is that attribution of any specific effect is dubious, and immediate fixes are impossible. This makes simple rallying cries like “save the Arctic!” nearly meaningless, no matter how heartfelt.

     I do agree with you that things can always get worse and that we should do whatever we reasonably can to keep them from getting much worse. But if we stick to “reasonable” measures, we are necessarily looking at best at a small deceleration of the current trends, whose effects will probably not be felt for decades. Better, of course, than an acceleration! But far short of what would be needed to even preserve today’s conditions, let alone pre-industrial ones.

     Given all this, there is, I think, limited value in constantly trying to scare people by confronting them with worst-case scenarios. In fact, I see all around indications that this strategy has backfired already.

     Realistically speaking, as Bill Illis just pointed out, ice or no ice, the Arctic is always going to be a cold place, and there is plenty of anecdotal evidence that Arctic species can survive relatively warmer conditions. As I think I said above, the real problem is not so much climate change itself but rapid change that does not give communities enough time to adapt. And, realistically again, our degree of control over the rapidity of the change is very limited, because there are billions of us split into hundreds of nations, and so if we move together at all, we can only move slowly.

     As for the final state of the Arctic, and the Earth as a whole, when all is said and done, it may be good to remember that a doubling of CO2 alone is only equivalent to an overall increase in insolation of about 1.6%. That would be a barely noticeable ripple in a Milankovich cycle. We are not looking at a drastically different planet 100 years from now, at least not if we manage to keep other pressing environmental challenges (deforestation, overfishing, old-fashioned pollution of everything but especially water resources) under control. This is, IMO, where resource commitment can have the largest, and certainly the most immediate payoff.

156. Re: julio (Comment #81013)

     One of the many challenges that the reality of AGW poses for the environmentalist movement is that attribution of any specific effect is dubious, and immediate fixes are impossible.

     OK, so that’s two challenges. Sorry! (I’m reminded of the Monty Python Spanish Inquisition skit…)

157. Julio,
     A very good comment!

     But if we stick to “reasonable” measures, we are necessarily looking at best at a small deceleration of the current trends, whose effects will probably not be felt for decades.

     I suspect there will always be huge disagreement about what is “reasonable”. Priorities, values, and personal inclinations are all tied up in any estimate of what is “reasonable”. My observation is that those who are most alarmed about all facets of climate change are the self-same people who believe many steps that I find very extreme, and impossible to justify in terms of costs/benefits, are perfectly reasonable. The public response to GHG driven climate change is at bottom a political rather than scientific question.
     .
     The political question is, unfortunately, even more complicated than most political questions, because the political/philosophical views of those involved in climate science lead many who do not hold similar political/philosophical views to doubt (justifiably or not!) what those climate scientists say about the magnitude and consequences future warming. I don’t think this is going to change.

158. Thanks, Steve.

     I basically agree with what you say, too, except that I do not see it as inevitable. Reasonable people should be able to talk. Distrust can be overcome; attitudes can change. The hardening of positions that you describe is unfortunate, and, of course, it parallels a similar polarization that can be seen in all areas of the political discourse these days. But it should be possible to do better than that!

159. Andrew_FL (Comment #81011)
     August 31st, 2011 at 8:34 am
     ——————

     The tilt and most of the rotation came from the collision with another body that created the Moon 4.45 billion years ago (a new number was proposed last week at 4.3 billion years ago but close enough).

     The tilt could have changed some over time – several degrees – but it has probably always been enough to create 6 months of darkness at the poles. The rotation has slowed down from 12 hours/day to 24 hours but the 1 year Earth orbit has only changed very marginally – a few days max. There is a precession in the equinox dates though and it would not have always been Sept 22 (but there were no calenders back then anyway).

     On the other hand, if there were no tilt in the Earth Axis, the Earth would be a completely frozen Iceball. The tilt creates the summers which allows most of the ice and snow to melt at the poles. Without the tilt, huge glaciers would exist at the poles – push toward the equator and the Albedo would rise to 0.7 or something. The result would be a -100C Earth.

160. Bill Illis:

     The result would be a -100C Earth.

     So there you have it: Bill has hit upon the solution to global warming. Build giant machines that torque the Earth, causing its tilt to decrease, in order to compensate for the extra CO2 forcing.

     (When the next ice age starts, they can simply increase the tilt to prevent the ice age from starting.)

161. Carrick,

     Build giant machines that torque the Earth, causing its tilt to decrease, in order to compensate for the extra CO2 forcing.

     “Give me a fulcrum, and I shall move the world….”
     I don’t know if the Earth would ice over absence axial tilt, but it was a great opportunity to quote Archimedes. 😉

162. That is a great quote, it ranks up there with my “simply increase the tilt” quote. 😉

163. The ice ages are caused primarily by a small shift in the axial tilt from 24.5 degrees to 22.1 degrees (it is declining right now and stands at 23.4 degrees).

     The tilt cycle is 23,000 years from max to min and there are other Milankovitch Cycles to consider, but a 1 degree change from today is enough to put us into the next ice age. There has not been an interglacial at 22.1 degrees in the last 2.5 million years and it is the most important cycle (although in the next minimum tilt and in the next one after that, the other Milankovitch Cycles will be operating exactly opposite to it so we should not go into an ice age for up to 50,000 years or even 130,000 years).

     So the tilt actually saves us from being an Iceball. Change the tilt to Zero degrees instead of the current 23.4 degrees and see what happens. The average temperature at the north pole is +2C in the summer and -24.5C in the winter. Take the average of that and apply it over the whole year and see what happens.

164. Bill Illis (Comment #81029)-Glaciation/deglaciation varies as the time integral of Northern Hemisphere Summer Insolation (for god only knows what reason that particular season, that particular region, and why it must integrate over time…none of this fits “global mean radiative forcing” paradigm). The obliquity alone doesn’t do it because if you integrate a 41,000 year period it just produces that same period, but with the phase shifted-which isn’t the periodicity associated with the main glacial/interglacial cycle. I think that the Eccentricity is clearly crucial to get the right periodicity but one needs to consider that phase shift. So I think one can’t really say that obliquity is the main effect.

Comments are closed.