SMOS

[Hyperion]

Actually this thread was started by Neven. I can't accept the credit. 

Here's how the CMOS microwave maps look for the past 40 days.

Downloaded from: https://seaice.uni-bremen.de/data/smos/png/
...
There is a definite trend from beige to other colours: dry -> wet

I played around a bit with those SMOS images.  I wrote a script to download the daily SMOS images for June 2010-2018 and to count the number of beige pixels in each image:




Average for the first 28 days of June:

(For what it's worth...)

Great effort Steve. But let's make sure we remember that we are/have transitioned to a predominantly bottom melt regime, rather than the meltpond surface melt regime historically. The salinity gradient and porosity of the young ice combined with the disintegrating pycnocline and more mobile ice and turbulent ocean surface mean, paradoxically that SMOS thin ice thickness maps are mostly more accurate than the were in past years where melt ponding was the norm for melt initiation. And therefore less beige Pixels were present due to melt ponds eliminating them. I love SMOS for its raw data feeds unmuddied by fiddling due to preconceived opinions of what should be presented, and what not. But the beige cutoff at levels where the error gets high, and the preemptive statement that the thickness scale should be ignored during melt season due to melt ponding making it inaccurate makes inter year comparison in a fast changing Arctic difficult.
The previous years in your chart continuing a steep descent while 2017, and more so this year not, most probably speaks more about less melt ponding and more bottom melt rather than any kind of slowdown.

[Tetra]

??

Is this thread really necessary? You've already posted the same comment onto the main melting season thread.

And you are specifically talking about 2018 and it's melting season in comparison to past years.

Also "fiddling"??

Plus, you do know SMOS is just like any other tool. With all the melting shenanigans going on this year, it probably has a lot of errors.

It shouldn't be used as a holy grail to support wide scale hypothesis's about the Arctic.

[Jim Hunt]

Is this thread really necessary?

A good question!

Plus, you do know SMOS is just like any other tool. With all the melting shenanigans going on this year, it probably has a lot of errors.

If memory serves, and much like CryoSat-2, SMOS maps used not to be published during the "melting season".

[Neven]

Is this thread really necessary? You've already posted the same comment onto the main melting season thread.

It's better to have any discussions as to what SMOS actually means during summer over here than on the melting season thread. But I agree one of Hyperion's comments is redundant, and so I've removed the copy on the melting season thread.

[Rob Dekker]

Regarding what SMOS means in summer, I found this :

The emissivity is mainly a function of bulk ice temperature, salinity and thickness. The bulk ice temperature is estimated from surface air parameters of JRA-25 reanalysis data and a zero-dimensional thermodynamic model [3]. The bulk ice salinity is estimated from a sea surface salinity weekly climatology. The sea ice thickness is corrected for the influence of the thickness distribution function to account for the invalid assumption of a uniform plane ice layer [3]. The latter correction leads to an apparent deeper penetration depth than previously reported [1]. We apply no correction for the influence of ice concentration. Thus, one can expect an underestimation of ice thickness when there is open water within the SMOS footprint. We account only for the thermodynamic influence of a parameterized snow layer on the bulk ice temperature but do not account for the direct radiative effect [4].

https://icdc.cen.uni-hamburg.de/1/daten/cryosphere/l3c-smos-sit.html

This suggests that SMOS is affected by open water between and on the ice.
More open water will trigger SMOS to report thin ice.
That means that ice concentration and SMOS in summer should have overlap (and correlate).
Low ice concentration (lots of surface melt) should thus correlate with low SMOS 'thickness' measurements.

That is consistent with the finding that SMOS in summer correlates well with low Sept extent.

[Steven]

That is consistent with the finding that SMOS in summer correlates well with low Sept extent.

For what it's worth:  using the average number of beige pixels in the SMOS images in June as a predictor of the NSIDC September extent, the estimate for the September 2018 extent would be  4.74 +/- 0.68 million km2  (95% prediction interval). 

This should probably be taken with a grain of salt, since the SMOS dataset isn't supposed to be used during summer and there are only 9 years of data (2010-2018) and I only looked at the beige pixels and ignored the other colors.

For the 2018 JAXA extent daily minimum the estimate would be 4.39 +/- 0.71 million km2 (95% prediction).

[Hyperion]

Nine days to the 4th.
Almost no beige pixels at all now, but for a blob compacted at the Fram entrance against Greenland.

Starting to Look a lot more like 2012. Same nine days for comparison.
And July 4th 2016, 2017.

Of course 2012 had just experienced a June Cliff, and had soggy ice and extensive melt ponding, particularly in the Pacific/ American side.

This year the signature is more likely to be from a mobile pack, with gaps and waterlogged fragments with no freeboard between the floes. It will be interesting to see what happens with the forecast out to the 9th on GFS showing an unprecedented heat and steamhosing from particularly the Russian side initially, but all directions. Perhaps some compaction will occur initially, but the cyclonic pattern taking over the whole basin is liable to cause dispersion. US Navy melt forecasts are looking extremely ominous.

[Hyperion]

Here's ,3,4,5 July. And an animation of 2-7 July surface winds and temperatures. Does look to me like compaction/dispersal correlates well. Temperatures?

[Cid_Yama]

30 cm is 1 ft.  So unless it's red or purple it's less than a foot thick.   

[oren]

30 cm is 1 ft.  So unless it's red or purple it's less than a foot thick.

Not really, as SMOS can't measure thickness in summer

[Hyperion]

Unless the ice in the the area happens to be dry and meltpond free of course. Which can happen if the air temp is below zero, and/or bottom melting in high salinity is chilling it to below the surface freezing point. But what you are getting then is the average thickness per 40x40 km area of ocean. So if its 50% concentration, then the ice in the area would be twice as thick on average as is indicated.
 The projection is conic equidistant if anyone feels like writing a script to recolor and rekey on this basis. Cryosat thickness is not available in summer for similar reasons. If surface melt charts were used to flag areas as "surface melting, thickness unknown" then actually a quite informative and plausibly accurate product could result.
At present SMOS is best for day to day comparison for information on ice movement and melting progress.

[Sciguy]

DMI is still showing most of the pack at over a meter thick.  Visit their website here:  http://polarportal.dk/en/sea-ice-and-icebergs/sea-ice-thickness-and-volume/

US Navy also has most of the pack at over a meter thick:

https://www7320.nrlssc.navy.mil/GLBhycomcice1-12/navo/arcticictn/nowcast/ictn2018070312_2018070400_930_arcticictn.001.gif

You can get their daily maps here:  https://www7320.nrlssc.navy.mil/GLBhycomcice1-12/navo/arc_list_arcticictn.html

[Hyperion]

And their melt animations

[Cid_Yama]

30 cm is 1 ft.  So unless it's red or purple it's less than a foot thick.

Not really, as SMOS can't measure thickness in summer

This is one of those common wisdoms that is actually untrue.  SMOS has high uncertainty for thick ice during the summer, underestimating the thickness of thick ice.  Uncertainty for thin ice is much lower.

New methodology has been devised in the processing of SMOS data in recent years, that has reduced the uncertainties and is now providing a useable data set in the summer.

http://meetingorganizer.copernicus.org/EGU2017/EGU2017-4452.pdf

Alfred-Wegener has also been looking at using a synergistic interpolation of SMOS and Cryosat-2 data to reduce uncertainty as SMOS has low uncertainty with thin ice and Cryosat-2 has low uncertainty with thick ice.

https://www.ecmwf.int/sites/default/files/elibrary/2018/17987-merged-sea-ice-thickness-product-complementary-l-band-and-altimetry-information.pdf

[Rob Dekker]

That is consistent with the finding that SMOS in summer correlates well with low Sept extent.

For what it's worth:  using the average number of beige pixels in the SMOS images in June as a predictor of the NSIDC September extent, the estimate for the September 2018 extent would be  4.74 +/- 0.68 million km2  (95% prediction interval). 

That's impressive, Steven.
A +/- 0.68 million km2 suggests a standard deviation of 340 k km2, which would be at the low end of the predictions at SIPN.

Did you consider submitting an entry this year based on this SMOS data ?

https://www.arcus.org/sipn/sea-ice-outlook/2018/june/call

Since SMOS is a unique, new, variable, your contribution would be very much appreciated.

[Hyperion]

30 cm is 1 ft.  So unless it's red or purple it's less than a foot thick.

Not really, as SMOS can't measure thickness in summer

This is one of those common wisdoms that is actually untrue.  SMOS has high uncertainty for thick ice during the summer, underestimating the thickness of thick ice.  Uncertainty for thin ice is much lower.

New methodology has been devised in the processing of SMOS data in recent years, that has reduced the uncertainties and is now providing a useable data set in the summer.

Very interesting Cid. I wonder if their archive has been reprocessed with the new algorithm. Or are comparisons with past years now made more difficult.
Anyhow. Freshly loaded to the Bremen servers. Here's the latest. As predicted a beige spot north of Siberia from the offshore winds causing compaction.
Don't expect it to last. The monster low in place in the Beaufort is going to disperse it back into the ESAS and Chukchi kill zones. The beige clump compacted against mid CAA is being fried and blown out to sea right now. And if the forecast for the next five days holds true, that low is going to grow into a monster covering the whole Arctic and spend the week winching in the whole north sea, with winds blowing all the way from Florida, and filling the whole gap from Greenland to Europe. Waves will be spectacular, as will the surface surge of hot water into the Atlantic killzone.

[Steven]

That's impressive, Steven.
A +/- 0.68 million km2 suggests a standard deviation of 340 k km2, which would be at the low end of the predictions at SIPN.

Did you consider submitting an entry this year based on this SMOS data ?

https://www.arcus.org/sipn/sea-ice-outlook/2018/june/call

Since SMOS is a unique, new, variable, your contribution would be very much appreciated.

Thanks, Rob.  But since I used only 8 years of data in the regression analysis (2010-2017), I'm not sure how seriously those numbers should be taken.

Actually I have more confidence in NSIDC sea ice concentration as a predictor of the minimum.  That gives an estimate of 5.0 +/- 0.8 million km2 for the NSIDC September extent, as I posted in the poll thread last Sunday.

[Steven]

Not really, as SMOS can't measure thickness in summer

This is one of those common wisdoms that is actually untrue.  SMOS has high uncertainty for thick ice during the summer, underestimating the thickness of thick ice.  Uncertainty for thin ice is much lower.

New methodology has been devised in the processing of SMOS data in recent years, that has reduced the uncertainties and is now providing a useable data set in the summer.

http://meetingorganizer.copernicus.org/EGU2017/EGU2017-4452.pdf

It's obvious that SMOS grossly underestimates sea ice thickness in summer.  Nothing in your link suggests otherwise.  Note that they are talking about determining sea ice concentration from SMOS data in summer, not thickness.

[Hyperion]

My guess is they have reprocessed the historic data all of the early files were modified end of October 2017.
Steven, if you have 1% of an area of ocean covered by small floes averaging 2m thick, with sub 200m gaps between them, should you say this is 100% area, 100% extent, and 100% concentration of 2m thick ice?
This is what we are being asked to swallow on every chart and metric, but SMOS.
Its meltponds SMOS has problems with in summer. With thin, young, porous, fragmented, and highly salinity stratified ice,  meltponding is improbable now. But far more prevalent in 2012. Less so every year.

[Hyperion]

Today's SMOS is in,  and results are as expected.

On the seventh the Beaufort cyclone has already detached the blob of beige from mid CAA. And compacted the pack in the north Beaufort through dispersing floes in the direction of the pole.

The winds blowing out of Fram strait and around the top of Greenland have shifted the beige blob off the northeast tip of Greenland and compacted floes in the mouth of Nares.

Winds out of Bering straight are withering Pago's bridge in the Alaskan coastal Chukchi- Beaufort border.

In general, melt is progressing at pace around the whole periphery of the pack.

[Hyperion]

Today's latest.
That cyclone mixing is wreaking havoc. Obviously not meltponding confusing SMOS over where the sea level height is going on here. Waves, warm, wet, windy whupping under cloud and fog veil huh?

[Hyperion]

Today's SMOS. It seems all the central basin ice is being expelled towards kill zones where either incoming ocean and Atmospheric heat is being soldly accelerated into, or Ocean heat and waves already exist enough to finish the deal.

[Hyperion]

It near as dammit is a blue ocean event already Neven. I bet my left nut that anything near the periphery, extending now to a Chukchi to Barents strip right across the pole where meltponding is being reported is actually open water, and satellites being fooled by wave action.
And what I cannot stand is bullying. Which is what you supported.
If it wasn't for the south Asian low pressure zone acting as a giant cross equator Hadley cell with the southern hemisphere whole Indian ocean high, it would be all over already. But that looks to be changing as the Arctic low pressure is merging with it
 And a single high pressure covering north Atlantic, America, and Pacific combined with that is a dipole covering half the planet, and the Arctic sea ice the meat in the sandwich.
I am only posting here because its not fair that Nevens inability to distinguish between deliberate concern troll attacks and actual naivety deprive others of factual reports.

[Hyperion]

https://www7320.nrlssc.navy.mil/GLBhycomcice1-12/navo/beaufortictn_nowcast_anim30d.gif

https://seaice.uni-bremen.de/data/smos/png/?C=M;O=D

[Neven]

It near as dammit is a blue ocean event already Neven. I bet my left nut that anything near the periphery, extending now to a Chukchi to Barents strip right across the pole where meltponding is being reported is actually open water, and satellites being fooled by wave action.
And what I cannot stand is bullying. Which is what you supported.

As said, if there's a BOE event this year (it's already practically there according to you), I will apologize. Conversely, if there's no BOE, you'll get banned from this forum.

[RikW]

https://www7320.nrlssc.navy.mil/GLBhycomcice1-12/navo/beaufortictn_nowcast_anim30d.gif

How reliable is this animation? Because if this is true almost all ice outside 80 degree is below 1m, thus doomed. Or there is a lot of open water, which lowers the ice thickness, which also isn't very good news I'd say.

Too bad we have a lot of clouds on Worldview. Any suggestions for layers/overlays to increase the visibility? I've been checking worldview and animated last month for some parts, I can only say for sure that Kara sea is doomed, which isn't a big surprise and I think the beaufort isn't an ice sheet but just a lot of floes, most of them much smaller than I would like.

For example 2 pictures of beaufort/cab 4 days ago, low on clouds, doesn't look that good. I think it's among the worst years for this region around this date (2015 and 2016 beging the other 2 worst years) but I think the 'bad' region extends closer to the pole (but those clouds...)

[Peter Ellis]

It near as dammit is a blue ocean event already Neven. I bet my left nut that anything near the periphery, extending now to a Chukchi to Barents strip right across the pole where meltponding is being reported is actually open water, and satellites being fooled by wave action.

Blue is ice, white is clouds.  Why even bother to post something so easily disproven? The people that built and run SMOS say that during the summer it doesn't accurately measure ice thickness, and that is the only factual report here.
https://go.nasa.gov/2NKVaF7

[oren]

It near as dammit is a blue ocean event already Neven. I bet my left nut that anything near the periphery, extending now to a Chukchi to Barents strip right across the pole where meltponding is being reported is actually open water, and satellites being fooled by wave action.

Seriously Hyperion, this is utter nonsense. A blue ocean event is defined in these parts of the world as <1 million square km of sea ice extent. Do you claim that we are there or very near that? Just note that JAXA is currently reporting >8.5 million sqkm of extent. It's not a BOE and will not be a BOE by September either.
About the "satellites being fooled by wave action", why not just open worldview and take a look for yourself? The white or bluish white of ice is easily distinguishable from dark waves. Your bet doesn't seem to be a good one.

[echoughton]

Me thinks Hypernion is going to lose a precious jewel. 

[Pmt111500]

Ah, SMOS, new ice product? What does it measure? Remaining thickness of first-year ice? The numbers are anyway quite a lot off of some of the other products. And they also differ from each other. Not so long ago there was a thickness map that looked like it measured the thickest bits remaining, maybe this does the same for thinnest bits.

[Hyperion]

I tried to post this gift on the melting forum but Nevz just deleted it.  These are entering the Arctic Basin as over hundred kmph over 22 second over 600m wavelength Waves. They are not showing up on Nullschool or other wave charts because of higher,  short period waves that they show automatically instead. But these have come fro the southern ocean, are being accelerated and lengthened by strong cross equatorial winds blowing north out of the south Atlantic, and the strong  north Atlantic southerlies blowing into the Arctic from off the US coast. The massive highs pressure systems covering the Atlantic basins are culpable.
These waves are about a meter high entering the ice, and should be easily capable of penetrating right through to the Chukchi. The reflection off shelves and refraction around shallows like Svalbard causes complex crossing and reinforcing patterns. The Pyramid waves I have mentioned before. No question that waterlogged  and fragmented icefields, with sea showing through  gaps, holes and puddles connected to the Ocean raised relative to the troughs nearby would fool radar Altimetry into thinking there was more free board than there is, more thickness and meltponds on the ice.
Sure, blue ocean already is a bit far measuring by jaxa extent definition. But on actual area, If I'm right about this it could be getting close.

[RikW]

It measures thickness in a reliable way if there are no melt ponds and/or open water is absent. So underestimation in the summer. Although it gives an indication if melt is happening, because when thickness decreases fast during summer, there are either a lot of melt ponds, a lot of open water between floes or thickness is really decreasing.

So based on the satellite images you can safely state the ice is in a bad condition

[Pmt111500]

It measures thickness in a reliable way if there are no melt ponds and/or open water is absent. So underestimation in the summer. Although it gives an indication if melt is happening, because when thickness decreases fast during summer, there are either a lot of melt ponds, a lot of open water between floes or thickness is really decreasing.

So based on the satellite images you can safely state the ice is in a bad condition

Thanks, so to calibrate it's position in the scale of true values, we'd need to check the numbers against other products during winter. Granted currently the compactness is pretty bad according to some other measures so this would underestimate thickness of remaining ice quite a bit. Say 50% open ocean on a grid cell could then mean the SMOS thickness is 50% smaller than the renaining ice on a grid cell? Anyway compactness is low currently. Interesting product this SMOS, looks almost designed to state the precariousness of the Arctic Ice

[Hyperion]

The advantage SMOS has on thin and fragmented ice is its penetration of freshwater is good. Salinity in the water decreases penetration I the xband microwave frequency markedly. So clouds and fog have little influence, unlike with Amsr2. But the saltier meltponds that can cause it to see the sea level a fraction higher than it is.
But the influence of wave tilt, porous first year ice etc is likely to make the overestimation of sea level quite small. Especially with the surface  being pegged to below freezing point by the high salinity water melting the floes from below. These charts have shown little or no sign of exaggerated thinness from hot and wet coastal atmospheric influx to my eye.
I think it would be fairer to call this a summer volume product than thickness as it is showing average thickness over over a mix of ocean and ice. If we were to plot a new chart that took the US Navy thickness and SMOS volume, and used them to colour pixels by the percentage difference, then we might get a more accurate area chart than presently presented. The US navy has the best under ice sonar resources so like SMOS their charts are unlikely to be as badly compromised by the swell influence I've pointed out. Which BTW IMO is the main reason piomass has been overestimating thickness, particularly north of Svalbard. And over the outer ESAS.
So if SMOS says 20cm and US navy 1m, then we might have around 20% actual area.  If this was the average Arctic wide then Area would be really only 20% of extent. Thus my point about being darn close to a blue ocean event right now.

[Hyperion]

Today's SMOS, and the most famous pyramid wave images in the world. I strongly recommend reading the following paper, for help in understanding wave, land and wind interaction relevant to today's arctic.

http://rsnr.royalsocietypublishing.org/content/63/2/119

WHAT KIND OF A WAVE IS HOKUSAI’S GREAT WAVE OFF KANAGAWA?
by
JULYAN H. E. CARTWRIGHT
1,*AND HISAMI NAKAMURA
2,*
1Instituto Andaluz de Ciencias de la Tierra, CSIC–Universidad de Granada,
Campus Fuentenueva, E-18071 Granada, Spain
2Chuo University, 742-1 Higashi Nakano, Hachioji, Tokyo 192-0393, Japan

The great wave off Kanagawa by Katsushika Hokusai is probably the most famous image in
Japanese art. It depicts three boats in heavy seas on the point of encountering the eponymous
wave, while Mount Fuji is glimpsed in the distance.

[Hyperion]

And the 13th.

[Wherestheice]

And the 13th.

SMOS is a better way to look at the state of the Arctic instead of extent....

[Neven]

An even better way is to look at them both. And then compare, compare, compare to what has happened in the recent past.

That SMOS graph doesn't mean much if it isn't compared to the ones from 2007-2017.

[Hyperion]

An even better way is to look at them both. And then compare, compare, compare to what has happened in the recent past.

That SMOS graph doesn't mean much if it isn't compared to the ones from 2007-2017.

I agree about the recent past. But if we are going to compare with previous years back to 2010 when SMOS started, then we have to remember that since 2013 the ice has been increasingly younger and more porous. Meaning years like 2012 would have far more meltponds, hence SMOS would have overestimated sea level, and underestimated ice volume. It should be far more accurate now.

But lets compare SMASS directly with US Navy thickness for the 13th of July.

It actually looks to me like the SMASUSnavthik ratio would give Area estimate closer to 10% coverage over most of the Pacific-American side of the pole.
Hmm. I'll pm Steven. If I resize these better and reduce the number of colours it should be easy to get accurate    smasusnathik figures and plots from them.

Edit: attached a few hours work with an unfamiliar gimp. May be usable.

[Hyperion]

Here's 12, 13, 14 July of both SMASS and USNavThik.
In unadulterated, and reduced to 16 colour plots. These can be split into frames at ezgif.com
This looks like a better way to do this than gimps over sophisticated posterising stuff. If I hadn't had four computers stolen in the last 12 months, it would be a walk in the park to write a pixel counting Smasusnavthik plotter. Anyone who wants to try should have no problems. Just finish the edit by cropping at the 70 degree north line, as per top left corner, and resizing to match.

[Wherestheice]

can someone upload a link to find the SMOS data?

[Random_Weather]

https://seaice.uni-bremen.de/data/smos/

[slow wing]

The maps are here:
https://seaice.uni-bremen.de/data/smos/png/

I've used the ones ending in _hvnorth_rfi_l1c.png, though I don't know why they differ from the _hvnorth_l1c.png ones, nor do I know if there is any reason to prefer one over the other.

By inspection, they look similar but not identical. I'm guessing the "rfi" ones have some extra processing but that's just a guess. Does anyone know?

[Wherestheice]

Thank you both! Very helpful

[Hyperion]

The maps are here:
https://seaice.uni-bremen.de/data/smos/png/

I've used the ones ending in _hvnorth_rfi_l1c.png, though I don't know why they differ from the _hvnorth_l1c.png ones, nor do I know if there is any reason to prefer one over the other.

By inspection, they look similar but not identical. I'm guessing the "rfi" ones have some extra processing but that's just a guess. Does anyone know?

The rfi version has the radio frequency interference from high powered radar systems processed out of it. I've been posting the plain unadulterated version. Call me quirky, but I prefer raw data feeds as little processed as possible. It seems to be outside the pack that gets any and only occasional effect anyway. And I guess I kind of like seeing when and where these probably mainly military radar systems are being used.

The last seven days. With a funky fade between frames and a little enlargement, seems to help visually track the changes.

[RikW]

So massive thickening in the CAB, so I presume melt season somehow ended already?

[Wherestheice]

So massive thickening in the CAB, so I presume melt season somehow ended already?

i think its just the ice compacting.

[Hyperion]

So massive thickening in the CAB, so I presume melt season somehow ended already?

i think its just the ice compacting.

Yes. There has been strong winds and heat coming in on the Pacific side for a few days and a ring of small cyclones around the Russian coast and a larger one in Chukchi-Beaufort.
The big area of blue at the eastern end of the Chukchi is where the remnant Beaufort circulation is. That's quickly going poof. So where the compaction is happening isn't at all great for the ice surviving. Its been shifted to where warm salty Atlantic water is available at shallow depth and its also vulnerable to wave mixing and breaking from the Atlantic.
And lined up for the Fram, Nares, and CAA export and kill zone.

[Hyperion]

Those were 16,17,18,19 1000hpa winds and temperature.
This is today.

[Tor Bejnar]

from July 1 (above)

SMOS is affected by open water between and on the ice.

I think what is useful is the beige areas which show 'dry' ice.  I presume this is local freeze, new snow or compaction.