HYCOM vs ASMR2 Imagery

[Jim Hunt]

I suppose they can't make pronouncements on the cleavage, etc.

I've asked a supplementary question, but I didn't phrase it quite like that!

[Neven]

[Jim Hunt]

My supplementary question on the HYCOM forum has received the following answer:

I attach the ACNFS plot you referenced and the corresponding plot from our next global real-time system (GOFS 3.1), which will likely replace our current global analysis (and ACNFS) once NAVGEM 1.4 is operational (it is running here with NAVGEM 1.3, like ACNFS).  They assimilate the same observations, but ACNFS only does so near the ice edge while GOFS 3.1 assimilates sea ice concentration everywhere (but with higher error bounds in the ice interior).  The reason for ACNFS only assimilating near the edge is in part due to that being the most important area for navigation but also because SSMI satelite ice concentrations tend to "over saturate" in the summer.  We switched GOFS 3.1 to "believe" the observations because they are generally better in recent years.

The two are quite different in the interior of the sea ice and GOFS 3.1 is certainly better there.  We did not see the low concentrations near the North Pole in ACNFS this time last year, so something has happened to make the free running model produce low concentrations this summer.  We are looking into it.

[Neven]

Thanks for spending time to confirm something we already suspected, Jim. I hope that puts an end to the usage of ACNFS maps that are out there somewhere on their own.

For this year. 

[seaicesailor]

Thanks for spending time to confirm something we already suspected, Jim. I hope that puts an end to the usage of ACNFS maps that are out there somewhere on their own.

For this year. 

I know it is very much questionable but I find ice drift gives reasonable information, if one reads them with caution, and away from the divergence centres and squall lines where the absurd holes appear.
Note that when the ACNFS predicts a hole, real ice concentration in CAB eventually diminishes in fact, the problem is you don't know where, and never it does to that exaggeration.
So even when I criticize it so much, one can get certain indications from it
The ice edge dispersion or compaction is also predicted. Again, maybe with questionable accuracy, but at least you know when to expect extent stalls or losses.

The validity of the SST and SSS maps out of the ice is still not challenged, nor supported here.

[A-Team]

hope that puts an end to the usage of ACNFS maps

I like Hycom myself. Nobody in their right mind ever single-sources information to begin with. We have a plethora of independent resources at our fingertips to synthesize and check for consistency.

It is a very decent project in some ways. For example, where else do you see a responsive export forum like Hycom (just look at the technical sophistication of some of their answers)? Answer: other sites never get back to you, never fix valid bugs you report, never implement easy suggested improvements, never fix broken files, or have a PR rep pat you on the head and pretend to forward your question on to a specialist whose contact information you are never given.

Where else do you see ongoing product development? The Hycom products are imperfect but they are very active in trying to improve them. I mostly see projects on autopilot, ghost ships that sail the same old seas long after the PI has passed on. Meagre documentation and no one left who could write it.

Mostly though I like the visionary aspects of the animations. They've built a very nice container so it's already in place to hold improved product. Even if the Hycom data is ultimately wrong, their representation of it is still very effective, whereas science communicated poorly is just those trees falling silently in the forest.

Where else is anyone even making a stab at predicting 2D sea ice conditions? Nowhere, they are all looking at today or analyzing the rear view mirror. It's from venturing predictions that you find out where and why your theories go wrong.

Why do we have fix so many maps and make so many animations? Because so many projects are too damn lazy or incompetent to offer and archive them. People just going thru the motions at work: they won't spend ten minutes making a fix if the time can't be billed to a grant.

I was looking around today for some basic July-Sept 2012. Modis has it but never got around to reprocessing it to the consensus projection, meaning there's no land mask or co-registration. AMSR2 3k started up in 2013. But there's Hycom with everything imaginable archived and indexed. And some of it, like the ice edge position and off-ice surface salinity and temperature, is   decent enough.

[Neven]

I don't have a problem with ACNFS per se (on the contrary),  but with a certain kind of usage of ACNFS. And then the persistence of that usage, and all the misunderstandings and conflicts this then produces. It sucks energy out of the Forum.

[FishOutofWater]

HYCOM is teaching us all about the live ongoing development of scientific models. Some people are learning the hard way about how to use a model in the process of being tested and improved but I think it is a great way to learn about science

[Jim Hunt]

Bumping this thread to point out that the ACNFS/NOGAPS section of the US Navy's web site seems to have disappeared. However ACNFS/NAVGEM can still be found at:

http://www7320.nrlssc.navy.mil/hycomARC/arctic.html

GOFS 3.0 is at: https://www7320.nrlssc.navy.mil/GLBhycom1-12/POLAR.html

and the current GOFS 3.1 is at: http://www7320.nrlssc.navy.mil/GLBhycomcice1-12/POLAR.html

The GOFS 3.1 reanalysis goes back to 2014.

Comparing MODIS, AMSR2 and GOFS 3.1 for yesterday reveals:

[jdallen]

Bumping this thread to point out that the ACNFS/NOGAPS section of the US Navy's web site seems to have disappeared. However ACNFS/NAVGEM can still be found at:

http://www7320.nrlssc.navy.mil/hycomARC/arctic.html

GOFS 3.0 is at: https://www7320.nrlssc.navy.mil/GLBhycom1-12/POLAR.html

and the current GOFS 3.1 is at: http://www7320.nrlssc.navy.mil/GLBhycomcice1-12/POLAR.html

The GOFS 3.1 reanalysis goes back to 2014.

Comparing MODIS, AMSR2 and GOFS 3.1 for yesterday reveals:

Thank you, Jim.  I haven't been spending much time in the navy models, recently, but may start again. GOFS seems a significant improvement over the older.  I'm updating my links.

[interstitial]

Just a comparison for this time of year.

[interstitial]

and the rest. It should be noted that 2014 image was later in the year than the rest.

[interstitial]

and January 2019 to June 2020

[OffTheGrid]

 
Mind reader! Lols
Would you believe I just downloaded those seven years for 25 June and was about to post them? At least one of us must be psychic!
Perhaps certain people repeatedly claiming that this year is not in the running for any records got to you too.
This month animation of Beaufort ice compressive strength is interesting. Seems to be getting very mushy.
Might as well post the latest month animation too, out to July 2.
As might be expected with a week of hot offshore winds, the ocean side of the CAA looks like Its detaching like it did last year.

The most simular year to this one at this point seems like 2017. Not sure if the velocity of export to the Beaufort and Fram were as high as this year has produced. Particularly how the thickest ice has been pouring out of the inner basin down the outer CAA, and keeping the Beaufort and Chukchi full, as fast as they try to melt it.

[interstitial]

I think the hycom model is much better than any of the others yet I don't see it on the forum very often. I am not sure why.

[OffTheGrid]

Maybe because Its got "experimental" stated in big friendly letters? Perhaps people think that piomass and the various extent and area metrics are actually  measured, when all of them are just as much "experimental" models, if not more so?

[oren]

I don't tend to rely much on Hycom, I doubt its veracity, but that's just my intuition as I've never done a rigorous verification.

[blumenkraft]

Here is my problem with HYCOM. One day you have 4m thick ice. Then the ice cracks and HYCOM drops the thickness to 2m immediately. No, it didn't lose half of its thickness in a day.

[interstitial]

I like the feedback on the hycom model especially from experienced persons. I don't have much experience because until now I couldn't get consistant acess. When I compare daily pictures I don't notice any jumps in thickness from day to day. Especially not from 4 meters to 2 meters. Maybe the 3.1 model fixed that? Right now there is so little above 3 meters it would be hard to spot. I guess what I am asking is if you have based that on the newest model or on an older one? I do like the higher resolution but if the model is junk that is not as important.

[blumenkraft]

Especially not from 4 meters to 2 meters.

Well, i exaggerated there a little but that's the gist.

[interstitial]

Especially not from 4 meters to 2 meters.

Well, i exaggerated there a little but that's the gist.

Who hasn't but the last time you looked was it the older model?

[oren]

Wht not compare Hycom to the Cryosat +SMOS thickness map from April 15th? That map should be quite reliable. This way you can see if Hycom is in the right ballpark, at least for that date before melting begins.

[johnm33]

Hycom. Mostly the older model was preferred, when they ran side by side, myself i preferred the current one. Bear in mind that there's some very expensive kit under the ice and whilst whole areas may be designated a particular depth for safety reasons it may be that there's an anticipated max for keels thats indicated rather than actual thickness, thus wave action passing through may cause much of the thicker ridged ice to simply keel over into created 'voids' changing the safety code. Lets say perhaps 15% of 4m ice floes will have a 12m keel depth and as waves [swells] pass the thinnest ice clumps together in the troughs freeing space for the keels to flip, so no real change in volume but huge change in geometry.

[interstitial]

Wht not compare Hycom to the Cryosat +SMOS thickness map from April 15th? That map should be quite reliable. This way you can see if Hycom is in the right ballpark, at least for that date before melting begins.

I looked. While their were some differences they looked about the same to me. The piomass with its lower resolutions blurs some features  also the major color changes happen at different thresholds. Those color thresholds made one model look like it was diverging from the other but looking at it I can see that would occur with the same volume estimates. Did you see something different? Did you expect me to see something different?

[oren]

I have not looked. I would appreciate a posting of Hycom Apr 15th and Cryosat Apr 15th side by side somewhere, maybe in the new Hycom thread. I would look for differences where the anomalies were supposed to be - near Svalbard, in the Beaufort, and in the ESS.