DMI Volume

[Tom_Mazanec]

I would ask that discussion of the DMI volume (based on Hycom-CICE model) and its comparison with PIOMAS and with actual measurement (Cryosat, SMOS) take place in a separate new thread - and not the existing one that compares PIOMAS vs. Cryosat.

Here you are, oren.

[Jim Hunt]

Here you are Tom.

I've got no idea what the presumed previous conversation consisted of, but does this look "realistic" to you?

P.S. I just took a look at the "Melting Season" thread!

[oren]

Here's an old quote from Neven:

There's observations from CryoSat, which is a satellite. But as it observes only swathes of the Arctic, a couple of miles across (I forgot how much), I'm not sure if it's picking up this event. Plenty of freeboard, though. I don't know if they're doing Operation IceBridge this year (measuring with planes), but if they are, they're not doing it now.

I still like PIOMAS best overall, but whatever you pick, it's best to compare interannually, and not compare models with each other (like: this one says it's 3 metres there, and the other one says it's 1.5).

The problem with ACNFS, or CICE, or Hycom, or whatever the name is, is that there's different versions with different inconsistencies. It's also run for and by the government/military, not for scientific purposes, which makes it more difficult to check what they do exactly. I don't have the time or inclination to constantly check how accurate it is, but I do notice it's occasionally used by both alarmists and climate risk deniers, meaning it can sometimes show what people want to see.

I'm not sure about the DMI or Mercator/TOPAZ models either. But hey, it's really difficult to model sea ice thickness. I think PIOMAS gives a good general idea of what's going on. It has proven itself over the years, as far as I'm concerned.

[oren]

what's the difference between PIOMAS volume and DMI volume?

DMI is an operational forecast model.
PIOMAS is a total volume measurement model.

They both model the same physics in much the same way, but they make different compromises because they are built to answer different questions.

Because the physics is the same they are each capable of doing the job the other one does, but they are likely to be worse at it.

If you want an operational forecast (what will the ice conditions be where I want to travel in the next couple of days) use DMI. If you want to know what the volume was in the past, use PIOMAS.

Some features of the physics (in particular the way the melting point changes with salinity) make the equations tricky to solve without numerical instabilities, and this particularly plagues the operational forecast models. Consequently they make rather more severe compromises with the physics than PIOMAS does and are fairly hopeless compared to PIOMAS when they step outside the narrow range of questions they are built to answer. PIOMAS crushes them in seasonal forecasting even though they are forecast models and PIOMAS is a measurement model.

[Wherestheice]

I contacted an expert from PIOMAS in regards to the differences between DMI and PIOMAS. Here is what he had to say

"There are some factors behind the different results. The key factors are that these two models differ in model physics, parameterization, resolution, and forcing. The different results are generally caused by these model differences. You can consider this as model uncertainty. Hope this answer your question.
Thanks for the interest and best wishes"

Jinlun Zhang

[oren]

Cross-posting from the questions thread:

I've noticed that PIOMAS and DMI are in disagreement over volume numbers. The DMI is claiming that 2020 is a record low for volume for May 5th, while PIOMAS claims volume is currently around the 2010's average and 2017 was significantly lower than any other year on this day.

How do the two insitutions differ in the way they measure sea ice volume, and what does a difference this dramatic between the different volume measurements indicate?

[oren]

I believe DMI is wrong, though it is hard to prove.
The best proof of wrongness can be achieved by comparing the DMI thickness map (and other thickness maps, such as PIOMAS) to actual buoy measurements in certain locations. Even this would not be hard proof, as volume is made up of an average of different ice floes, old ice mixed with new ice and even pressure ridges, which make up a significant percentage of the volume in the Arctic ocean. But it could give clues to wrong thickness and wrong rates of thickening.
My best layman indicator that something is very wrong with DMI is those years where the volume has bottomed and started growing before the end of August. Bottom melt in the Arctic ocean continues in September, and volume growth can only begin when initial refreeze happens, which is on Sep 10th-20th in the central Arctic, and even later in the adjacent seas in the Inner Basin. So I get the feeling something is simply not modeled properly in DMI. Perhaps they over-rely on low negative air temps and somehow disregard ocean interaction below the ice.
In addition, I have a strong suspicion thickening physics are not modeled properly either. Looking at an animation of the past month, available at http://polarportal.dk/en/sea-ice-and-icebergs/sea-ice-thickness-and-volume/, barely shows any thickening in the Central Arctic during the month, even near the North Pole, while temperatures were minus 10-15^oC and lower, and all indicators show thickening still continues at this time of year, albeit a a lower rate. Compare this to PIOMAS animations by Wipneus which do show continuing thickening, which I believe is the right behavior.

[grixm]

Cryosat agrees more with DMI at this time in that it shows that the volume is near or at an all time low, unlike PIOMAS. http://greatwhitecon.info/wp-content/uploads/2020/03/NRT-Vol-2020-04-15B.png

Whether DMI hits the top or bottom at the correct time is less interesting than the model's difference between years. If DMI shows that volume bottom one year was lower than another year, that is valuable information regardless of whether that bottom was in August or September. Because even if it is biased, it is biased in the same way, canceling the bias out.

Therefore, it is important information that DMI shows that the current volume is lower than any other year at this time. It is not infallible, no single model is, but it is a piece of the puzzle that should not be ignored. And doubly so since, as I said, Cryosat also says that the volume is near an all-time low.

[kassy]

CryoSat uses a synthetic aperture radar over areas of sea ice to measure a much smaller footprint than previous satellites. This provides the resolution to identify sea ice floe regions, and open ocean lead regions between floes, and to measure their surface elevations. Over sea ice, CryoSat echoes are assumed to scatter from the interface between the ice surface and the layer of overlying snow [Beaven et al., 1995; Laxon et al., 2013], and we can therefore measure the sea ice freeboard (the height of sea ice above water).

http://www.cpom.ucl.ac.uk/csopr/science.html

vs

The 3D ocean model HYCOM and the sea-ice model CICE is developed at the University of Miami and Los Alamos National Laboratory. The models are fully coupled at each time step. Output are the surface variables sea level and ice conditions (concentration, thickness, velocity, convergense, strength, etc.) and 3-dimensional maps of current, temperature and salinity at sigma levels.

Model set-up
The DMI HYCOM-CICE set-up covers the Atlantic, north of about 20°S and the Arctic Ocean, with a horizontal resolution of about 10 km. Model forcing is ECMWF weather forecasts. A 144 hour forecast is produces twice daily, at 00 and 12 UTC.

http://ocean.dmi.dk/models/hycom.uk.php

vs

PIOMAS Ice Volume Validation and Uncertainty
PIOMAS ice thickness and volume results  have  undergone substantial validation via comparison with ice thickness data from  submarines, moorings, airborne electromagnetic induction (EM) measurements and ice thickness retrievals from ICESat.  In addition, model sensitivity studies were conducted to assess the impact of model parameters on ice volume anomaly trends.

The problem of validating sea ice volume
It is difficult to validate total Arctic sea ice volume directly.  There are no true measurements of total ice volume that can be compared to model-derived estimates.  Validation is best achieved through a comparison with in situ and satellite-derived ice thickness observations.   The observations used here are collected at the Unified Sea Ice Thickness Climate Data Record.  We can also try to estimate how much the model-derived volume and trends change when we change model parameters that are not well know.   The combination of this allows us to make estimates of the uncertainty in the ice volume and trends from PIOMAS.

http://psc.apl.uw.edu/research/projects/arctic-sea-ice-volume-anomaly/validation/

http://psc.apl.washington.edu/sea_ice_cdr/

[Richard Rathbone]

Whether DMI hits the top or bottom at the correct time is less interesting than the model's difference between years. If DMI shows that volume bottom one year was lower than another year, that is valuable information regardless of whether that bottom was in August or September. Because even if it is biased, it is biased in the same way, canceling the bias out.

The DMI model changes from year to year. Year to year comparison with DMI is apples to oranges because the models used are different. DMI bias does not cancel.

[grixm]

Whether DMI hits the top or bottom at the correct time is less interesting than the model's difference between years. If DMI shows that volume bottom one year was lower than another year, that is valuable information regardless of whether that bottom was in August or September. Because even if it is biased, it is biased in the same way, canceling the bias out.

The DMI model changes from year to year. Year to year comparison with DMI is apples to oranges because the models used are different. DMI bias does not cancel.

Source for that? Last time the model was updated AFAIK was 2016. And in the note about this update they say this:

However, the trend between years is almost unchanged. Thereby, a year with a large sea-ice volume in the old setup also has a large volume in the new setup, and similar for years with low sea-ice volume.

http://ocean.dmi.dk/arctic/icethickness/icetext.uk.php

[Richard Rathbone]

They use CICE,  the current version of which was released on March 16th, 2020.

https://github.com/CICE-Consortium/CICE/wiki/CICE-Version-Index

In the absence of any statement to the contrary, I assume that they keep their code up to date.