What do the GCM models get wrong about sea ice?

[anonymous]

During the EGU 2013 Twitter bombarded me with tweets declaring an ice free Arctic before 2050. Here is a link to the presentations: http://meetingorganizer.copernicus.org/EGU2013/posters/12141

Apparently models are used to analyse a wide range of sea ice features. However, to my knowledge there is not a single GCM model capable to reflect recent development, neither with extent as a metric nor with volume. Here is an interesting paragraph discussing parameters useful to select skillful models:

Our selection is based on relationships between the baseline sea ice state and the year at which SSIE crosses a given threshold. Our analysis suggests that CMIP5 models tend to reach a given summer sea ice extent earlier when (1) they have smaller initial September sea ice extent , (2) the amplitude of their climatological cycle of sea ice extent is larger, (3) they have a smaller ice volume in the annual mean, (4) the extent of thin (<0.5 m) ice is larger in September, and (5) they lose ice at higher rates now.

These results can, in addition, be interpreted in light of simple physical mechanisms, resp. (1 and 3) models with a larger initial volume of ice need more energy, and thus time, to melt ice and reach a given extent, (2) the seasonal cycle of sea ice extent is a proxy for the model sensitivity to the seasonal cycle of incoming solar radiation, (4) the ice is more susceptible to melt away in areas where it is thin, and (5) the most sensitive models now are likely to reach ice-free conditions earlier under future warming.

It is also important to stress that these criteria are not fully independent. For example, the amplitude of the 1979–2010 mean seasonal cycle of sea ice extent correlates significantly (p < 0.001) at 0.67 with the 1979–2010 mean September thin ice extent in the CMIP5 models. [Constraining projections of summer Arctic sea ice, F. Massonnet et al., 2012]

Which other physical parameters do you think might be hard to model according to observations?

[Jim Hunt]

Hi Arctic.io,

I've been discussing this sort of stuff with some of the names on your list recently, in conjunction with The Distributed Arctic Sea Ice Model

The most relevant quotes seem to me to be these.

Rear Admiral Jon White, re sea ice models:

We need more data, and another ten years

Currently unattributable, re GCMs:

The CICE model is much more sensitive to forcing data than to internal model parameters.

[anonymous]

Thanks Jim, 'We need more data, and another ten years' is obvious. Point is this makes 20 years in total and leads to a possibly more accurate question: What data is needed? And only slightly unrelated, who actually defines data needs?

Or simply what wrong with this picture?



Looking at this again, I think I have to turn it into a blog...

[Ice Cool Kim]

Which other physical parameters do you think might be hard to model according to observations?

I suspect one thing that is not being modelled correctly is angular dependence of reflectivity.

I have asked two climate researchers, one field researcher measuring radiation in the Arctic, another a modeller in CICE group. Both chose not the reply to the question, despite having replied initially.

I can't find anything in the public CICE code, though it's quite a volume so I may have missed it. In any case it's out of date.

Two requests to get read access to svn trunk have simply been ignored.


Much attention has been given to melt pools recently, that was specifically what both the field researcher and modeller were working on.

Applying fixed "albedo" figures to snow if fine but it is not fine for water. Since , for much of the melting season incident light is very low significant proportions will not be absorbed but reflected.

This will have a significant cumulative effect over the season.

If anyone has any info on how I could check that I'd be interested.

[Ice Cool Kim]

Also over horizon scatter/refraction.

It is well known that the sun is still visible when it is geometrically below the horizon. This is due to refraction by the atmosphere.

This only for a few minutes at dawn and dusk in temperate latitudes but will be the case for days maybe weeks at high latitudes when the transition around the equinoxes is very gradual.

In a similar way to the oceanic tides there are atmospheric tides. These tend to be much simpler because they are not constrained and reflected by the continents in the same way as oceanic tides.

I recently showed the correlation between AO and ice coverage persists up until 2010 (when there seems to be a net divergence).

As part of that investigation I noted the circa 14d cyclic variations around max/min periods.

Such atmospheric tidal waves could be affecting the length of the polar "dawn and dusk" and thus the onset and end of melting/freezing seasons.

[Jim Hunt]

Hi Kim,

My own requests to "modeller(s) in CICE group" have not "simply been ignored". Quite the reverse in fact.

Perhaps you're going about it the wrong way?

[Ice Cool Kim]

Possibly. What kind of requests? Did you get access to trunk?

[Neven]

2011, Rampal et al.:

IPCC climate models do not capture Arctic sea ice drift acceleration:
Consequences in terms of projected sea ice thinning and decline

[ChrisReynolds]

Models overestimate the strength of the atmospheric inversion.

Medeiros, B., C. Deser, R. Tomas, and J. Kay, Arctic inversion strength in climate models, Journal of Climate, 24 , 4733–4740, 2011.

Zhang, Y., D. Siedel, J.-C. Golaz, C. Deser, and R. Tomas, Climatological characteristics of Arctic and Antarctic surface-based inversions, Journal of Climate, 24 , 5167–5186, 2011.

Boe et al, 2009, "Current GCMs’ Unrealistic Negative Feedback in the Arctic." http://www.atmos.ucla.edu/csrl/publications/Hall/boe_et_al_published.pdf

This leads to excessive heat loss from the surface, and ice both forms too thick, and recovers too vigorously from perturbations.

[Ice Cool Kim]

Models that far back did not include melt ponds.

[Jim Hunt]

Requests for additional information Kim, and no I haven't gained access to the CICE 5.0 SubVersion trunk yet. Like you, I do however have access to version 4.1, which is the latest stable release, as I pointed out to you quite some time ago.

What do you suppose it takes to get access to "the unstable trunk" of "a coordinated international scientific project"?

[Jim Hunt]

What's wrong with this picture?

There aren't a whole lot of pretty coloured dots north of the Arctic Circle.

I guess that Rear Admiral White may well have access to some thickness data that the rest of us are not privy to? Nonetheless he seems to think more dots up there would be a very good idea.

[Ice Cool Kim]

Like you, I do however have access to version 4.1, which is the latest stable release, as I pointed out to you quite some time ago.

Man, that codes from 2010. I'd say (hope) they've come along way since then. The point was, if I want to suggest something or that something may be missing, it would be pretty pointless basing that on old code. Neither am I going to scratch my head too long unless I know that they have not already modelled it.

What do you suppose it takes to get access to "the unstable trunk" of "a coordinated international scientific project"?

Well the CICE site doc says something like "if you require access to trunk, please contact ....."

That does suggest that access is possible and they don't indicate there are any particular restrictions.

Then you have to dig pretty damn hard to find an address for the person named. An when you do email them you don't even get a refusal. Nada.

I was not asking for commit rights, just svn checkout access to read the code.

I think it's great that v4.1 is open-sourced but it seems the openness is limited.

[Ice Cool Kim]

Boe et al, 2009, "Current GCMs’ Unrealistic Negative Feedback in the Arctic." http://www.atmos.ucla.edu/csrl/publications/Hall/boe_et_al_published.pdf

Rather than -ve f/b being unrealistic it may be lack of a +ve one.

Melt ponds do provide quite a bit of +ve feedback. In fact annual swing then gets too big.

Flocco 2010 adds lid of refreezing ice and it gets a bit closer.

Here I showed that there was strong acceleration that fitted such a +ve f/b behaviour but crandles thought it was "absurdly" strong.

http://forum.arctic-sea-ice.net/index.php/topic,232.0.html#top

The modelled strong neg. maybe correct, they were just missing the +ve one.

My post here shows behaviour that would be in agreement with a strong neg. f/b from the large open water post 2007.

http://forum.arctic-sea-ice.net/index.php/topic,234.0.html

[Jim Hunt]

I think it's great that v4.1 is open-sourced but it seems the openness is limited.

Have you ever been involved in a collaborative software development project, either open or closed source?

Have you ever been a "climate scientist" on the wrong end of some "bad publicity"?

Do you ever click any of the links I provide in answer to your questions?

Does that answer your question?

[Vergent]

The problem is simply that we have never mastered the second law of thermodynamics. We know it is true, but we have never generated a general model of it.

Consider a beach in August. The sand is hot. the water cool. The second law says that the heat is going to get to the water and warm it. But the model says there will be an onshore wind.

V

[ivica]

What about turbulence ?

[Ice Cool Kim]

Have you ever been involved in a collaborative software development project, either open or closed source?

Both. Many times.

Do you ever click any of the links I provide in answer to your questions?

I did indeed follow the link when you originally posted it . Most helpful, thanks. That was the site that I got the doc from that I referred to and where I got the contact person's name.

Does that answer your question?

I was not asking a question, I said I didn't get a reply.

[Ice Cool Kim]

What about turbulence ?

Yes, big problem.

I actually think trying to model the Earth from first principals is a bit , shall we say, ambitious.

Thinking that current GCMs are capable of prediction, is foolishness.

[Ice Cool Kim]

Consider a beach in August. The sand is hot. the water cool. The second law says that the heat is going to get to the water and warm it.

No it doesn't.

It says heat will flow from a hotter body to a cooler body if the two are in thermal contact.

There is nothing in your example that goes against that principal.

[anonymous]

Thinking that current GCMs are capable of prediction, is foolishness.

Boy, really, please, make up your mind and try to stay on-topic. Whether turbulence is included or not has nothing to do with the skill of models at a global level, conservation of energy and forces still applies.

And this is not the thread to discuss this, open another one or use PM!

[Ice Cool Kim]

Which other physical parameters do you think might be hard to model according to observations?

Atmospheric turbulence, cloud formation and precipitation. All of which are needed for the inputs to ice modelling.

ego: Thinking that current GCMs are capable of prediction, is foolishness.

[anonymous]

I guess that Rear Admiral White may well have access to some thickness data that the rest of us are not privy to?

I don't think so, there is a lot of draft data from submarine uplooking sonars and moorings at the NSIDC and WHOI. The thickness data from IceSat and CryoSat is in the literature. And data from current and archived IMB buoys is online too. But, it seems the only guy having analysed thoroughly all of this is Peter Wadhams. And his projections are rather dire. I wish someone could convince Tamino doing this in public again.

[Ice Cool Kim]

and CryoSat is in the literature.

How much? More that 7 dots per year? If so please point.

[Ice Cool Kim]

Peter Wadhams. And his projections are rather dire.

Not as frightening as his solutions.

Anyone who proposes "engineering" a system that we do not understand is either insane, irresponsible of both.

Hubris, dear professor. Do not confuse yourself with The Almighty.
Go and play with someone else's planet. Don't touch mine.

Sadly, the current state of hysteria means that some well-meaning people will be prepared to listen to that kind sorcerer's apprentice idiocy.

 

[Ice Cool Kim]

Has anyone seen what Arctic ice looks like in HadGEM3 ?

[Neven]

I would urge folks to stay on-topic. There are other places to bash models (like I did here for instance).

[anonymous]

and CryoSat is in the literature.

How much? More that 7 dots per year? If so please point.

Go ahead: how-to-access-cryosat-data

[Ice Cool Kim]

Go ahead: how-to-access-cryosat-data

He-he. Yes I'd seen that. Even the specification doc is a masterpiece of european bureaucracy.

I was not planning of reprocessing every blip from the radar altimeter.

You said "in the literature" , what I thought you meant is that there was some published processed data. All I have seen in that sense was something from the late Dr Giles but that just gave a graph with 7 points per year and even that was not available in numerical format.

Anyone aware of CryoSat2 derived data >7 dots and < 7 terabytes ? 

[Neven]

This is costing me energy I don't have. Any more off-topic derailing stuff and I'm turning into Nero. Or Caligula. Whichever you prefer.

[frankendoodle]

2011, Rampal et al.:

IPCC climate models do not capture Arctic sea ice drift acceleration:
Consequences in terms of projected sea ice thinning and decline

While I believe transport is a big factor in ASI reduction, I question why the rate of North Atlantic icebergs doesn't follow the same patterns as ASI if transport through the Fram Striate is such a big cause: http://www.navcen.uscg.gov/?pageName=IIPIcebergCounts

Melt ponds are more likely the culprit. While GCMs included CO2 from industrial and natural sources, did they include the lowering of albedo from soot? And even if they did, did they factor in the sudden jump last decade in wildfire activity in the Northern Hemisphere? Or the increase in industrial output? 

[ChrisReynolds]

Frankendoodle,

The ice bergs from that will mainly be MYI and Greenland calving, FYI melts out rapidly in the Fram Strait export. The 1990s peak may be due to the exiting of a lot of MYI and from Greenland.

Except for specific experiments GCMs do not usual include albedo effects of soot. PIOMAS definitely does not factor in that effect, nor does it factor in the effect of increased melt pond formation due to more FYI.

[Vergent]

I think that it would be easy to tweak these models so that they would track the recent losses. You only need a few watts/m^2. I think the question should be:

"Why haven't models that track the recent ice losses been published?"

and the answer is:

Because, if they do track the ice losses, they crash to zero ice in short order.

V

[Ice Cool Kim]

What are you suggesting there Vergent.

That if the result was too alarmist they would not get published ?!

Maybe I have not been watching closely enough, but that has never been an issue so far.


Read Flocco 2010 on CICE.  Several recent attempts at adding melt melt pools resulted in too much annual swing. That paper was about adding a further refinement where a thin "lid" of ice forms on top of the melting pools. This brings it closer to reality but it's still showing too much variation.

That could indicate that they are getting closer to an accurate model. But it could also be a case of tweaking things until they fit the calibration period in a way that fits, without being a correct model.  I'm not saying which is the case but both are valid interpretations. I hope they are getting close.

This was the case with global temps. Speculative feedbacks were applied to CO2, volcanoes were exaggerated and it was reasonably close to the 1960-1990 calibration period. It did not fit so well to the pre-1940 warming period nor to the post-Pinatubo period with negligible volcanoes.

The natural experiment that Earth has provided us, with a sustained non volcanic period, gives new information that allows us to compare the two and "tweak" the models.

It seems that HadGEM3 is the first of a new generation of models that get closer to modelling the more recent period.

That was why I was interested in whether any output for arctic ice coverage from that model was available. It's not just the ice model that matters, it's also what the rest of the model feeds to the ice model as input that matters.


HadGEM3 , with more accurate surface temperature results and a more +ve f/b in the form of melting pools maybe the best estimate so far.

Is anyone aware of any such output?

[anonymous]

There are several providers of CMIP output data, start here esgf-data.dkrz.de or choose one from the list of nodes (middle of the page, left side).

A very good discussion of model short comings (includig PIOMAS) is the Hadley Centre Technical Note 91:

In this paper, we refer to a range of climate models used in CMIP3 (IPCC AR4) and CMIP5 (IPCC AR5) model intercomparison projects including results  obtained by examining an ensemble of climate models to span the range of  structural uncertainty (i.e., uncertainty due to model differences, although in practice, initial conditions are also important). Due to the availability of extensive data and detailed diagnostics, here we also focus on results from models developed at the Met Office Hadley Centre. HadCM3 (Gordon et al., 2000), HadGEM1 (Johns et al., 2006; McLaren et al., 2006) and HadGEM2-ES (Martin et al., 2011; Collins et al., 2011). Although not referred to extensively here, HadGEM3 (Hewitt et al., 2011) is the latest coupled atmosphere-ocean-sea iceland model under development and is currently used in the GloSea4 system (Arribas et al., 2011) which is used for seasonal forecasting. [Assessment of Possibility and Impact of  Rapid Climate Change in the Arctic, August 2012]

[Jim Hunt]

Both. Many times.

Great. Perhaps now we can continue the numerous off topic parts of this conversation over on the DASIM thread?

I was not asking a question, I said I didn't get a reply.

The implicit question was "Why didn't I get a reply?"

Re CICE/GCMs/HadGEM3 clicking the following link may prove to be helpful:

http://oceans11.lanl.gov/trac/CICE/raw-attachment/wiki/BoundaryLayerWorkshop/

See part 10 in particular.

[Ice Cool Kim]

actico: thanks , useful link , tho still no signs of HadGEM3 ice data.

Arctic sea ice extent has declined at an annual rate of over 4% per decade since
satellite records began in 1979. This rate is faster in the summer season and
there is evidence that the rate of loss has increased over the latter half of the
satellite period. There is also evidence that the ice has thinned at a rate of
approximately 60cm per decade. However, the heating required to melt the ice at
this rate is very small - just 1 W/m2 representing only 2% of the magnitude of the
seasonal cycle, implying that observing and modelling the mechanisms
underlying these changes will be challenging.

1 W/m2 is tiny. Confirms a comment someone made above. Chances of modelling to that accuracy are not high. Chances getting something about right by the wrong parameters is high. Like they "challenging".

Rather confirms some of my earlier comments.

"Arctic sea ice extent has declined at an annual rate of over 4% per decade since satellite records began in 1979. "

Helps to put it in perspective. Easy to get the impression it's 40% sometimes.

Thanks for the link.

[Ice Cool Kim]

Jim, thanks for link:

• SHEBA (Beaufort Sea/Chukchi Sea 1997/8):
• Max diurnal cycle in near-sfc T = 3.3 C
• Max diurnal cycle in incoming SW = 470 W/m2

Modelling 470 W/m2 to better than +/- 1 Ouch!

Nothing is impossible given enough effort but it's going to take a generation to get to that level of accuracy.

[Jim Hunt]

Hopefully getting back on topic, it seems "A-Team" hasn't explored this neck of the woods as yet, so to reiterate/requote once again:

VP isotropic rheology cannot and does not capture observed shear, vorticity, divergence, thickness distributions, trends in velocity and export, or deformation-driven ice production.

Jerome Weiss. Drift, Deformation and Fracture of Sea Ice SpringerBriefs in Earth Sciences DOI: 10.1007/978-94-0007-6202-2.5

http://192.95.52.196/wordpress/wp-content/uploads/2013/02/Paper5EarlyView.pdf

[anonymous]

actico: thanks , useful link , tho still no signs of HadGEM3 ice data.

Choose the UK node, enter 'sea ice' in the search box, wait for results, filter by model on the left hand side, wait for results, scroll down list, click 'add to cart' somewhere, open the 'Data Cart' tab, click 'expand', click 'http', login, done.

Here a list of UK data sets available without login: http://badc.nerc.ac.uk/data/dataset_index/

[anonymous]

ICK, HadGEM3, sorry, was sure it is already included, but I have seen it, googling....