02 December 2009

Fake ice

The ice is real ice, but the satellites are being fooled.  Actually, not even that.  The satellites are correctly reporting what they're seeing, but the humans have been making an assumption that no longer holds true due to change in the Arctic.  A friend pointed to an interview (audio only) of David Barber, on Quirks and Quarks on CBC.  Early on, he mentioned how the satellites were being fooled.  That isn't exactly what's up, but was the right answer for the circumstance.  I'll take a bit more time to discuss the details of what is happening on this part of the story. 

Our standard method for observing sea ice from space is to use satellites to measure the microwave energy emitted from the earth's surface.  Sea water is a very bad emitter, so has a very low brightness temperature.  Sea ice is a pretty good emitter, so has a much warmer brightness temperature.  You go through a couple of elaborations on this, and out pops the fraction of the surface that is sea ice instead of sea water.

You can also be a little more demanding than that.  Particularly in the Arctic, this makes sense.  The elaboration comes from the fact that not all ice emits microwaves equally well.  Salty ice is a better emitter than fresher ice.  In the summer time, when ice floes do some melting (but not enough to get rid of the whole floe for the ones we're interested in), it is the saltier parts of the floe that melt away first.  This is the same thing happening when you salt the sidewalk -- the salt lowers the melting point, and the salty parts melt first.  When you get to this time of year, and the melting stops, what is left is a relatively fresh ice floe.  It is also called 'multiyear' ice, since it's now in its second winter.  With a more detailed analysis of the microwaves, you can try to distinguish between the first year ice (saltier and a better emitter) from the multiyear ice (less salty, but still a better emitter than sea water) from the sea water (very poor emitter).

Analogy for the remote sensing: The satellite is listening to the surface.  Sea ice is much louder than sea water, so you can start by just checking how loud things are.  Louder = more sea ice.  You can then listen a little more carefully.  Multiyear ice is a little quieter than first year ice, and a little different pitch.  So there's a choir of sea ice, and the multiyear ice is, say, the sopranos singing a little quiter than the basses (first year ice), but both are much louder than the baritones (sea water).  If you've listened to a choir, a band, or just a room of people talking (and deciding how many people were in it, and how many of them were men vs. women), you've done the same sort of discrimination that we're doing with the satellite observations.

Now for the spot where we were fooled. 

Up until basically this year, it was a standard assumption that multiyear ice was thick ice.  It made sense.  Ice that was thick enough to survive the summer can't be terribly thin (we figured).  And in the next winter it has a chance to pile up even more thickness either by freezing more ice on, or by getting piled up as the circulation jammed ice floes on top of each other.  This had been confirmed many times in field expeditions starting long before the satellites were flying.

What Barber found instead was that the ice breaker was cruising along at 13 knots (25 kph) through what the satellite said was multiyear ice.  He was amazed.  That is about the speed the breaker would go through open ocean!  What he realized had happened, after going up to the bridge, was that the satellite was seeing just a thin scum of the fresh multiyear ice, that was on top of also a thin layer of first year ice.  There was only just enough multiyear ice for the satellite to see that.  It is no longer the case that we can assume that the sound of multiyear ice means thick ice.  There's still ice, so methods for deciding total ice coverage are ok.  But we can't use the multiyear ice signature to mean 'thick' any more.  Probably (my opinion) this has become a progressively less accurate thing to do over the last few years, and it's only now that the difference has become as drastic as David found.

This is only a small part of the interview, and the whole thing is well worth listening to.


Scruffy Dan said...

I listened to the interview just as I was finishing up a post on arctic ice, partially stating that I may have over reacted with the 2007 melt season, and also on the news that an albedo tipping point may not be as close as some had feared*.

This complicates things...

* http://www.gps.caltech.edu/~ian/publications/Eisenman-Wettlaufer-2009-incl-SI.pdf

carrot eater said...

"thin scum of the fresh multiyear ice, that was on top of also a thin layer of first year ice."

How would this come to happen?

Jesús R. said...

May it also work the other way around? i.e. satellites seeing just a thin scum of saltier first year ice, that is on top of a thick layer of multiyear ice?


John Mashey said...

Great article.
This is useful example (akin to the recent tree-ring growth issue).

Given X = f(Y) = g(Z), there may be straightforward relationships f() and g() across some range of values for Z, in which case, one can get reasonable values for X.

But, if Z takes on new values not seen before, f() or (g) may simply change.

As a technology example, Moore's Law has rolled long enough that some people started taking for granted, an nice exponential cure.

Of course, others knew better: we still have a few more rounds to go in CMOS for density, but we also use to get predictable clockrate improvements as well (smaller transistors switch faster), but that has essentially stopped. That's why there has been a bit phase change in microprocessors to multiple cores,rather than advertising big GHz improvements like they used to.

Jill said...

probably the term rotten sea ice would be better than a scum. So we had a multi year ice that was not complete, and the first year ice then froze in and around and even under some of this rotten multi year ice.

Hank Roberts said...

How about a pointer to how sea ice forms year after year? I've read several descriptions, but I don't know which to recommend.

Nothing I've read suggests any way salty first year ice could grow on top of multiyear ice to any significant extent though.

jg said...

This is a great post, and I regret that I only discovered your blog last October for I missed your Earth Temperature 1 post you linked to. Did you ever do an Earth Temperature 2? by another name?

I've been struggling for two days to articulate questions over these two posts. Can I send you an illustration that will show my questions (and confusion)?

Robert Grumbine said...

A little more description of the ice. Pelto prefers 'rotten' as a description, which could also work. It isn't how I envision it, but I've also not gone out in to the field to see the sea ice myself. I rely on photos taken by friends, and quite a few of the AIDJEX and CRREL reports on ice structure. I believe the former are now available in full, and the latter are often available online.

For 'rotten' ice, envision swiss cheese. The saltier parts of the floe melt out first, leaving behind holes throughout the ice floe. Come winter freeze up, then you can get new ice freezing in to the spaces.

The vision I was carrying is more that the swiss-cheese ice floe breaks up (it collides with something else and shatters). The bits then float on top of the sea.

The new ice freezes on to the bottom, rather than the top -- almost always. The top of the ice floe, remember, is above sea level, so the freezing adds to the bottom. The almost is that if you have enough snow fall, the weight of the snow pushes the top of the floe below sea level, sea water rushes in, and then freezes. This is the only obvious-to-me way of getting the salty signature on top of multi-year ice. This is mostly a concern for the Antarctic, where there is more snow than the Arctic.

jg: Sure, send the figure.

No, earth temperature 2 hasn't appeared yet. It will, some day. There are so many temperatures that meteorologists use, I'm sure I'll be back on the topic.

carrot eater said...

Speaking of brightness temperature, I have some questions about the work of Harries et al (Nature 2001; Journal of Climate 2007). Perhaps I'll save it for the eventual appearance of "Earth Temperature 2". If that's going to get into how you can finagle lower trop, mid trop and stratosphere from the satellite data, that's going to be a good contribution.

EliRabett said...

If the ice freezes from the bottom (let's leave snow out of this for a while) then you can easily have thin layers of multiyear ice over new ice.

The interesting thing is whether the satellite measurements in September have been correlated against aircraft radar and optical observations recently. AKA sky truth

Gareth said...

See also: University of Manitoba page, with video of Barber.

And Jeff Masters has an interesting recent post on how Arctic climate is changing.