Antarctic sea-ice models improve for the next IPCC report

University of Washington

IMAGE
Lettie Roach in her office. view more  Credit: Dave Allen

The world of climate modeling is complex, requiring an enormous amount of coordination and collaboration to produce. Models feed on mountains of different inputs to run simulations of what a future world might look like, and can be so big — in some cases, lines of code in the millions — they take days or weeks to run. Building these models can be challenging, but getting them right is critical for us to see where climate change is taking us, and importantly, what we might do about it.

A study in Geophysical Research Letters evaluates 40 recent climate models focusing on sea ice — the relatively thin layer of ice that forms on the surface of the ocean — around Antarctica. The study was coordinated and produced to inform the next Intergovernmental Panel on Climate Change report, due out in 2021.

All the models projected decreases in the aerial coverage of Antarctic sea ice over the 21st century under different greenhouse gas emission scenarios, but the amount of loss varied considerably between the emissions scenarios.

“I am really fascinated by Antarctic sea ice, which the models have struggled more with than Arctic sea ice,” said lead author Lettie Roach, a postdoctoral researcher at the University of Washington. “Not as many people are living near the Antarctic and there haven’t been as many measurements made in the Antarctic, making it hard to understand the recent changes in sea ice that we’ve observed through satellites.”

The models are known as coupled climate models, meaning they incorporate atmospheric, ocean, terrestrial and sea ice models to project what the future holds for our climate system. We are all familiar with the story of soon-to-be ice-free summers in the Arctic and the implications that may have on global trade. But what’s driving change around Antarctic sea ice and what’s expected in the future is less clear.

This study’s assessment of Antarctic sea ice in the new climate models is among the first.

“This project arose from a couple of workshops that were polar climate centered, but no one was leading an Antarctic sea ice group,” said Roach. “I put my hand up and said I would do it. The opportunity to lead something like this was fun, and I’m grateful to collaborators across many institutions for co-creating this work.”

The Antarctic is characterized by extremes. The highest winds, largest glaciers and fastest ocean currents are all found there, and getting a handle on Antarctic sea ice, which annually grows and shrinks six-fold, is critically important. To put that into perspective, that area is roughly the size of Russia.

The icy parts of our planet — known as the cryosphere — have an enormous effect on regulating the global climate. By improving the simulation of Antarctic sea ice in models, scientists can increase their understanding of the climate system globally and how it will change over time. Better sea ice models also shed light on dynamics at play in the Southern Ocean surrounding Antarctica, which is a major component of our southern hemisphere.

“The previous generation of models was released around 2012,” says Roach. “We’ve been looking at all the new models released, and we are seeing improvements overall. The new simulations compare better to observations than we have seen before. There is a tightening up of model projections between this generation and the previous, and that is very good news.”

###

Co-authors of the recent study are Cecilia Bitz at the UW; Jakob Dörr at the University of Bergen; Caroline Holmes at the British Antarctic Survey; François Massonnet at Universite Catholique de Louvain; Edward Blockley at the U.K. Met Office; Dirk Notz at the University of Hamburg; Thomas Rackow at the Alfred Wegener Institute in Germany; Marilyn Raphael at the University of California, Los Angeles; Siobhan O’Farrell at the Commonwealth Scientific and Industrial Research Organisation in Australia; and David Bailey at Hamilton College in New York state. Funders of the research included the National Science Foundation and NOAA.

From EurekAlert!

62 thoughts on “Antarctic sea-ice models improve for the next IPCC report

  1. Let me guess… they will show no decline until present but somehow will predict rapid decline for the next years. Right?

    • The icy parts of our planet — known as the cryosphere — have an enormous effect on regulating the global climate.

      Climate models grossly under estimated the 1997-2007 drop in Arctic sea ice. Having been tweaked to bet match the drop they totally failed to match the fact that 2018-2019 Arctic minimum was NO LESS than 2007 min.

      Models totally failed to match Antarctic sea ice increase at the same time as Arctic was melting. Since models are constructed such that rising CO2 is the primary driver, they are unable to produce the famous “polar see-saw”.

      The so-called “polar amplification” had to be renamed Arctic amplification, since it was not polar at all since it did not apply to both poles.

      In short they do not have the slightest idea what the primary driver of changes in sea ice at either pole is. They have ZERO skill in predicting or even matching the existing record of past change at either pole.

      Why would we give any credence to ANY claims of what may happen over the rest of the 21st century ?

      This is not science, it is a dogma. A political dogma masquerading as science in an attempt to gain unmerited authority and credibility.

      ENOUGH OF THIS BS !

    • OH FFS please fix the “ki11” filter. The word ski11 should not get banned from our lexicon because it share some letters.

      Climate models grossly under estimated the 1997-2007 drop in Arctic sea ice. Having been tweaked to bet match the drop they totally failed to match the fact that 2018-2019 Arctic minimum was NO LESS than 2007 min.

      Models totally failed to match Antarctic sea ice increase at the same time as Arctic was melting. Since models are constructed such that rising CO2 is the primary driver, they are unable to produce the famous “polar see-saw”.

      The so-called “polar amplification” had to be renamed Arctic amplification, since it was not polar at all since it did not apply to both poles.

      In short they do not have the slightest idea what the primary driver of changes in sea ice at either pole is. They have ZERO ski11 in predicting or even matching the existing record of past change at either pole.

      Why would we give any credence to ANY claims of what may happen over the rest of the 21st century ?

      This is not science, it is a dogma. A political dogma masquerading as science in an attempt to gain unmerited authority and credibility.

      ENOUGH OF THIS BS !

      • “OH FFS please fix the “ki11” filter. The word ski11 should not get banned from our lexicon because it share some letters.”

        I don’t know for sure, but I think that particular word is in the WordPress filter, which is separate from WUWT. It is definitely a pain. It must drive the moderators nuts with the extra work it requires.

        • I believe you are mistaken. Individual WP subdomains set their own key words. This is quite flexible if done properly. The problem is that using short strings like ki11 or 1iar means filtering ANYTHING with those letters in them. There is a means of specifying a leading and trailing space to ensure just those WORDS are trapped, not just anything containing those letters.

          I researched and proposed the required config to solve this years ago, posted it to “tips” page but sadly it got ignored.

          Using the numeric 1 instead of letter l works, as long as you can remember all the silly, perfectly fine words which fall into this trap.

          It’s true that it wastes everyone’s time.

  2. ” The icy parts of our planet – known as the cryosphere – have an enormous effect on regulating the global climate.” This is backwards as the global climate regulates the cryosphere. The obvious example is how, in the current Ice Age we are in, as we cycle in and out of glacial cycles, the climate has changed even where there are no advancing glaciers. Do we think that ice cubes control the functioning of a refrigerator? No wonder the global climate models are all over the place! I’m going to get some of those ice cubes and put them in a glass of special fruit juice (later), and then maybe I can understand global climate models. Stay sane and safe (local golf course opens today, I think I will be alright).

    • “The icy parts of our planet — known as the cryosphere — have an enormous effect on regulating the global climate.”

      I thought CO2 dialed the climate. I just haven’t been keeping up with these scientific advances brought to us by millions of lines of codes.

      • This claim is baseless since they have no fundamental understanding which is capable of the slightest prediction.

        They are referring, it seems, to the naive, simplistic ASSUMPTION that less ice means more sunlight means more melting: ie a +ve feedback … runaway melting.

        This is totally inconsistent with the existing satellite record of Arctic sea ice which totally disproves this trivial hypothesis.

        1980-2007 was at least consistent with the hypothesis, though that does not constitute proof it was correct. The “accelerating decline” the stopped accelerating since 2018-2019 was the same as 2007. You can NOT have a +ve f/b which goes AWOL for over a decade and still pretend the facts are consistent with you naive feedback assumptions.

        There are other factors dominating Arctic sea ice which remain totally unknown due to the obsession with linking everything to “global warming” and CO2. No one is even trying to get a working understanding because no one will get funding for anything which does not fit the dogma of CAGW.

    • The tropics totally control the Arctic and Antarctica. Everything is driven directly or non directly from the tropical region.

      Just look at the atmospheric temperature over Antarctica during the 2015 El Nino, colder for longer.

      These folks don’t even look at some of the most important data available. The stupid, it burns.

  3. GIGO as usual. Unvalidated and unverified models that cannot calculate a non-linear, multi-variate chaotic system.

    • It’s worse than non validated, they are INVALIDATED. The get everything wrong, even after post hoc adjustments are made.

      Total failure.

    • Precisely. All they have done is to sift through the ‘G’ input again with their fingers crossed.

  4. I would encourage WUWT readers to review this paper “Structure and Performance of GFDL’s CM4.0 Climate Model” by Held, et al published last year.

    https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2019MS001829

    Search on the word “polynya”. Look at section 4 – The Preindustrial Control. What does all of this tell me?
    With no anthropogenic forcings at all, the piControl shows me that a computed simulation produces wide swings in sea ice and consequent global and northern/southern hemisphere temperature trends. In other words, it shows why natural variation should not be ruled out as the cause of whatever trends the instrumented record shows, for both sea ice at both poles and for temperatures.

    • When they analyzed the pre-1979 satellite data era photos (Nimbus missions) that had been sitting around getting dusty, they found Antarctic sea ice had been outside know bounds + and – in just those few earlier years!

  5. What grieves me bitterly about nonsense like this is that it has consumed huge amounts of money and resources on computer models which could have been employed more profitably to observe and measure the physical world.

  6. “I am really fascinated by Antarctic sea ice, which the models have struggled more with than Arctic sea ice,” said lead author Lettie Roach, a postdoctoral researcher at the University of Washington. “Not as many people are living near the Antarctic and there haven’t been as many measurements made in the Antarctic, making it hard to understand the recent changes in sea ice that we’ve observed through satellites.”

    There really isn’t much global warming down there and the obsession with ice free summer in the Arctic has gone nowhere. There is no trend in sea ice in the Arctic. There was a brief 3 or 4 year decline once that had gotten climate scientists on their toes but that also went nowhere. The climate science track record on sea ice is a history of dismal failures but a strangely umdaunted determination to push on with the feedbacl warming acceleration drama that has eluded them.

    https://tambonthongchai.com/2019/09/28/sea-ice-extent-area-1979-2018/

    https://tambonthongchai.com/2019/07/02/antarctic-sea-ice-collapse-of-2019/

    • Yep, my falsifiable hypotheses is that these f-wits will, at some undetermined point in the future FOD, or will admit that this “relentless beating down of back-radiated IR” could never melt ice in the first place.

      I’m going with #1. This dog never could hunt.

    • “. . . and there haven’t been as many measurements made in the Antarctic, making it hard to understand the recent changes in sea ice that we’ve observed through satellites.”

      And here, all along, I thought that scientific, remote sensing spacecraft that were in orbits to permit monitoring the Arctic ice sheets were also able, from those same (near-polar) orbits, to monitor ice sheets in the Antarctic. Maybe I have been, as Rick (Humphrey Bogart) famously stated, “misinformed”/sarc.

      “Useful satellite data concerning sea ice began in December 1972 with the Electrically Scanning Microwave Radiometer (ESMR) instrument. However, this was not directly comparable with the later SMMR/SSMI, and so the practical record begins in late 1978 with the launch of NASA’s Scanning Multichannel Microwave Radiometer (SMMR) satellite, and continues with the Special Sensor Microwave/Imager (SSMI). Advanced Microwave Scanning Radiometer (AMSR) and Cryosat-2 provide separate records.
      “Since 1979, satellites have provided a consistent continuous record of sea ice. However, the record relies on stitching together measurements from a series of different satellite-borne instruments, which can lead to errors associated with intercalibration across the sensor changes. . . . The first instruments provided approximately 25 kilometers by 25 kilometers resolution; later instruments higher. Algorithms examine the microwave emissions, and their vertical and horizontal polarisations, and estimate the ice area.
      “Sea ice may be considered in terms of total volume, or in terms of areal coverage. Volume is harder, because it requires a knowledge of the ice thickness, which is hard to measure directly; efforts such as PIOMAS [8] use a combination of observations and modelling to estimate total volume.
      “There are two ways to express the total polar ice cover: ice area and ice extent. To estimate ice area, scientists calculate the percentage of sea ice in each pixel, multiply by the pixel area, and total the amounts. To estimate ice extent, scientists set a threshold percentage, and count every pixel meeting or exceeding that threshold as ‘ice-covered’. The common threshold is 15 percent.”— source of foregoing quoted text: https://en.wikipedia.org/wiki/Measurement_of_sea_ice

      So 2020-1979 represents a period of more than 40 years of obtaining satellite data of Antarctica ice. I have no idea what Ms. Roach implies by “there haven’t been as many measurements made in the Antarctic”.

      BTW, there are currently 70 permanent research stations scattered across the continent of Antarctica, which represent 29 countries from every continent on Earth. The southernmost scientific station, the Amundsen–Scott South Pole Station located at 89°59′51″S, has been continuously occupied since 1956 (although it has been rebuilt, demolished, expanded, and upgraded several times since then). And the McMurdo Station, located at 77°51’S on the coast of the Antarctic continent, was also established in 1956 for conducting scientific observations and has been continuously manned since then.

      So, 2020-1956 represents a period of more that 60 years of obtaining ground data of Antarctic ice.

      • It just happens that 1979 is a convenient starting point for the pseudo-experts in Arctic ice. Using that year they keep telling us the melting is accelerating when it has effectively stopped decreasing over 10 years ago.

  7. All the models projected decreases in the aerial coverage of Antarctic sea ice over the 21st century under different greenhouse gas emission scenarios, but the amount of loss varied considerably between the emissions scenarios.

    I have a feeling that was a requirement to be included. If your model didn’t show ice loss, it was considered defective.

    • Max,

      Error in the original paper – “aerial coverage” should read “areal coverage”. The author is not talking about airborne coverage of the Antarctic continent and surrounding ocean. She just can’t spell.

  8. “The new simulations compare better to observations than we have seen before.” That’s new, they actually looked at observations…..

  9. Nobody writes millions of lines of code anymore….so I wish reporters would quit using that term to indicate “complexity”.

    Seriously..

    As for the models, they are only as good as the input used to run the functions. IF that input is garbage, you get garbage out.

    I was taught that modeling was a useful tool to illuminate what may or may NOT be happening, not as an absolute. It’s like a tool maker that makes the worlds best welding torch but has no idea how to use it properly and the weld still fails.

    • Agreed. During my years of training, one of our pieces of old-timer dogma was “Only a fool extrapolates.” and “If you must extrapolate, don’t count on your results.”

    • Just Jenn posted: “Nobody writes millions of lines of code anymore….so I wish reporters would quit using that term to indicate “complexity”.

      Well, check this out:
      “A global climate model typically contains enough computer code to fill 18,000 pages of printed text . . . Scientists translate each of these physical principles into equations that make up line after line of computer code – often running to more than a million lines for a global climate model.”—source: https://www.carbonbrief.org/qa-how-do-climate-models-work

      Your wish will not be granted as long as climate change™ exists.

      • Gordon: “Scientists translate each of these physical principles into equations that make up line after line of code”

        I know……it’s so sad that THIS is what is used to persuade the public on how “important” those models are. Its the equivalent of building a completely useless over the highway walkway, complete with ramp and stairs at each end in the middle of a desert. It may be useful, but not where it is located and regardless of how many man hours were put into building it, how much money was spent, you are still left with a useless walkway for jackrabbits.

        Perhaps the report should have questioned WHAT principles those “scientists” were translating into WHAT equations.

  10. “The new simulations compare better to observations than we have seen before.” Is that because they are better predictors or is it because they are over-tuned to the recent data? The scientific method requires a training set (if you have to have one) and a testing set. if you use all the data as your training set, then the model is a correlation and likely useless for prediction. It also implies you don’t have the physics correct yet.

    I do agree with her that these models require a great deal of coordination and collaboration, but I call it collusion.

  11. “and we are seeing improvements overall.”

    In other words, they are less useless than before.

  12. Fascinating that by a raise of a hand a study is granted. Not saying the lead is not qualified but raise your hand and off you go. Opportunities abound in climate modeling.

  13. Obviously, a word was omitted in the above article’s first sentence. I have corrected it here using the appropriate acronym (no charge):
    “The world of climate modeling is complex, requiring an enormous amount of coordination and collaboration to produce GIGO.”

  14. Erl Happ: Climate changes – oh so naturally

    This is a tour de force, pulling together the different strands of climate knowledge and weather lore Erl has been building up over the years. Hi ideas fit well with those of Marcel Leroux, who worked out that climate change is largely driven by longer term changes in the polar oscillations. Erl believes these are largely due to ozone changes caused by solar variation which drive the global air flows via consequent surface pressure changes. As Hans Jelbring tells us: Wind controls climate. As Nikolov and Zeller tell us, surface pressure and insolation control temperature. Erl delves into the underlying causes of those polar variations, and connects the levels and latitudes of the atmosphere for us in a novel, logical and interesting way.

    While looking for Leroux and Antarctic and these Ploar Highs I came across this article.

  15. Millions of lines of code, and nary a regression test in sight. If we tried that where I work, we would get sacked very quickly. Even our test tools have tests. Process models written by software amateurs are not fit for the purpose of creating projections to be used by governments for making long term policy decisions that could cost the world trillions of dollars and millions of lives.

    • “Even our test tools have tests.” Yes, of course, but how did you get your management to pay for development of tools to test the tools? “What verify and validate the lab – are you crazy?” It is often seen as an unnecessary cost.

      Or does your customer require it? Or your insurance company?

      • Jim,
        In safety-critical systems it is imperative to reduce errors as much as possible (keeping in mind that you still have to produce a useful product within a meaningful time frame). In addition, many of these systems (like medical devices, avionics, etc.) are covered by government regulations that stipulate the types of development processes which can be used, and testing processes fall under these regulations. If you are working for a government agency, they may add on additional requirements as well. This is the environment I have spent most of my professional life in, so I’ve seen the good, the bad, and the ugly.

  16. Imagine if these people were running a business that had to make a profit to survive the medium to long term, and to pay real dividends to financial backers.

    All they do is take, their poor results protected by the endless supply of cash.

    The IPCC and affiliates is the equivalent of Enron, except the cash keeps coming.

  17. Just Jenn
    ” they are only as good as the input used to run the functions”.

    I’d posit the models are even worse than supposed. NONE of them use actual mathematical simulations of the processes involved. Due to a massive lack of computing power the models simplify complex, millimeter scale turbulent processes with multi-kilometer averages into nonlinear functions. Turbulence, so far, simply cannot be modeled because because it is a random process- from the millimeter scale of micro eddies in the sea surface(which fade into Brownian motion around 1mm) to meter scale waves and eddies.

    Engineers in both aerospace and marine design have doe extremely well to develop models to predict how plane and ship hulls behave and can optimize, to a degree, the efficiency of the plane and hull shapes. They still have problems with turbulence that have to be resolved using wind tunnels or water flow tanks to properly shape small details.

    The best example I know of this is the F-18 fighter jet.The first block of planes sometimes suffered severe vibrations in the rear of the plane, especially in landing config. The vibration tore off onr of the rudder fins causinf crashes, some fatal. The turbulence came from the joint between the front of the wing at the fuselage. Gaps there, for control of the airflow on the inner part of the wing could cause vortices that straifgt slammed into the fins. The fins were strengthened an on later models little vertical fins on the top of the plane deflected the vortices so they hit the rudders consistently on the inner sides.

    The computer simulations didn’t work. The runaway calculations simply wouldn’t finish. A wind tunnel was used to determine the exact position an functioning of the fins.

    • Please note that those little fins on top of the wing root aren’t on all versions of the F/A-18 – the engineers fixed the problem. But you are correct about the problem, and the initial solution. You are also right about using computerized fluid dynamics models in some flight regimes. There is still a need for human brains, and testing.

  18. “All the models projected decreases in the aerial coverage of Antarctic sea ice over the 21st century under different greenhouse gas emission scenarios, …” Apparently, they didn’t just use RCP 8.5. I wonder if they tells us which RCP gave the results closest to reality?

  19. “Building these models can be challenging, but getting them right is critical for us to see where climate change is taking us, and importantly, what we might do about it.”

    Too bad they haven’t come close yet.

  20. ““This project arose from a couple of workshops that were polar climate centered, but no one was leading an Antarctic sea ice group,” said Roach. “I put my hand up and said I would do it.”

    “put my hand up”, as in ‘teacher, may I?”.
    That is, of course, the assumption that critical team leaderships go begging for want of qualified leaders?
    No application for employment.
    No list of recommendations.
    No support from officials or people in power.
    No haranguing decision makers.
    Riiigght…

    “We are all familiar with the story of soon-to-be ice-free summers in the Arctic and the implications that may have on global trade. But what’s driving change around Antarctic sea ice and what’s expected in the future is less clear.”

    “what’s driving change around Antarctic sea ice and what’s expected in the future is less clear”, well, that is a good admission that one does not know.

    However, “The models are known as coupled climate models, meaning they incorporate atmospheric, ocean, terrestrial and sea ice models to project what the future holds for our climate system. as an overriding statement fails to recognize that the snow and ice levels predicted by coupled climate models are not their only failing.
    That ‘I don’t know’ extends to far more and greater atmosphere relationships.

  21. “The opportunity to lead something like this was fun.” And I will continue to get funding provided I get the “right” answers.

  22. The current Antarctic Sea Ice Extent since the 26th May 2020 has been larger than the 1979-1990 Average. Many recent years have been better than the 1979-1990 Average despite 2016-2018 being lower. The 2001-2010 and the 2011-2019 averages are both better than the 1979-1990 Average. It recovered nicely from the 2017 record low.

    The Arctic Sea Ice Extent has generally declined over the last 30 years or more. The earths tilt/wobble can create the see-saw of a cooling Antarctic when the Arctic is warmer. Air & sea weather patterns could be localised or part of the see-saw. The orbit, sun activity and global forcings (eg. SO2, Ozone, CO2?), if any, would affect both N&S poles to a similar amount. Geological and volcanic activity is generally localised so Arctic volcanoes wouldn’t affect the Antarctic (and Antarctic volcanoes don’t affect the Arctic). So the difference in results and behaviour of the 2 poles need to be explained with less influence from CO2. Why is it so? Blindly changing model parameters until you like the result is a lottery driven by bias not science, so it proves nothing. Changing climate observations based on climate models is very risky (eg. land temps, ocean temps, Satellite measured MSL, tree rings). GIGO.

    What can be certain is the climate models do not sufficiently explain the observations but many scientists don’t like to admit mistakes or inadequacies of research. Reanalysis, changing reference periods and not comparing long range forecasts with real data makes it harder to evaluate the quality of the models. Making more models and averaging the results is not being more scientific. It would be slightly better to use the models that are closer to reality. It would be best to find empirical data to verify the parameter values and their uncertainty.

    https://nsidc.org/arcticseaicenews/charctic-interactive-sea-ice-graph/

Comments are closed.