songster

My understanding is that there's always a zone of a few hundred metres of very high bottom melt rates at the edge of the pack. As long as the water is <i>any</i> amount above freezing, that's all it takes, because the specific heat capacity of water is so high. It really depends just how close those buoys are to the edge - and we're talking a resolution of 1 or 2 pixels on MODIS at most. If you look at buoy I, you can see the rate accelerating, presumably as the ice edge catches up to it. http://imb.crrel.usace.army.mil/2012I.htm

They're nowhere near the centre of the main ice mass, that's why. They're in the fringes, right on the edge of what's picked up by Bremen (~= various grades of pack ice), and a little in from the edge as seen be IMS (~= open drift ice).

Same to you. Both these assertions are rubbish. Ice bridges form in narrow straits: there has never been a bridge between Greenland and Iceland. The Odden was a region of first year frazil / pancake ice, not thick ice.

Let's not. You're entitled to your own opinions, but not your own facts. Ice area on Jan 6th 1982: 17.3492889 million sq km Ice area on Jan 6th 2012: 16.8366909 million sq km http://arctic.atmos.uiuc.edu/cryosphere/timeseries.global.anom.1979-2008 Do yourself a favour and look things up before you make yourself look stupid again.

Indeed. Subtitle for this melt season appears to be "The Year It Just Didn't Stop"

I think it might be simplest to drop the earliest years and just dot in averages, much as IJIS do here. http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm I think they go a bit too far though - perhaps 5-yearly dotted averages from 1980-84 through to 2000-04, and then individual years from 2005 on?

It's thinner and the Earth's hotter.

I explained that. To take the past week as a concrete example, in that part of the Laptev, the historical average temp. for August 16th is hotter than the historical average temperature for August 9th. This is because historically, this is the time of year when that region melts out, so it has a higher ice coverage on the 9th than on the 16th. More open water on the 16th than on the 9th = higher surface temperatures on the 16th than on the 9th. This year, the ice went long ago, and temperatures have remained more or less steady in that region over the last week. When you subtract out the historical average, that then means that this year's anomaly has dropped over the last week.

Look at the actual temperatures - nothing out of the ordinary. The reason the anomaly is going down is because the historical average is rising. Historically, ice would just be starting to disappear from the Laptev, which produces a sharp upward jump in temperatures as the (zero) ice melts out to reveal the water a couple of degrees above zero. Thus when you compare today's comparatively static temperatures to the historical average, you get a sharp drop in the anomaly. You can see the opposite of this effect going on right now, nearer the centre of the basin. Over the last ~3 days, a vast area of white (zero) has been replaced by 3-degree water (= +3 degrees anomaly compared to the historical average which is still zero). Simple version: anomalies do Weird onions near the ice edge, because you're comparing open water to ice.

You've already got one - dunno if I qualify as "level" though. In fact, I'd not be surprised to find an overlap of half a dozen or more under differing names. 'Tis the way of the Internet.

I was giving you the best chance! If the ice is under 2m, then the "edge area" is an even smaller proportion of the total surface area. My point is that there is a common trope ("fragmented ice has a higher surface area to allow for melt") which makes very little physical sense in that the exposed "edge" of each floe is a negligible proportion of the total surface unless the ice is fragmented into pieces only a few metres across - something we'll never be able to see on the MODIS images. We're largely in agreement here - the Navy makes maps predominantly to get the ice edge correct for surface ships. They don't really care about accurately modelling thickness except in as much as that affects the ice edge model. No submarine will use an ice thickness map as a gauge of where/when to surface. I think the only point of difference is that you say they'll deliberately err on the "safe side" and overstate thickness in the central pack, whereas I'm saying I don't think they worry much about it one way or the other.

I don't think the Navy thickness maps deliberately err on the thick side, they simply don't bother to try and model it well. They're produced for surface shipping, not submarines, and so the ice edge is what they're concerned with. If a sub wants to surface, it'll check the thickness by sonar. Again, scale. Pixel size for ACNFS is a minimum of 3.5km per side. For a sub wanting to surface, what possible use is it to know that the average thickness of the surrounding 12.25 km^2 is 1.5m? You need the thickness of the bit directly above you!

Oh, come on. Think scale. The ice is a maximum of a few metres thick. The pixels are a minimum of 250m across. Any lump of ice big enough to see on the MODIS image - even a single pixel - is big enough that the "edge" area is negligible in comparison to the flat area. Leads visible by MODIS are relevant in terms of increasing solar absorbance, but irrelevant in terms of increasing the surface area available for melt. It's a minimum of two orders of magnitude difference, mostly three or four orders. Edit: possibly a bit dogmatic there. When the pack is consolidated and near 100% concentration, then the leads are important, because melting is pretty much only occurring at the edges. However, when you're looking at lower concentration ice in the marginal zone, then the difference between one floe of 2.5km diameter (10 pixels) and 100 floes of 250m diameter (1 pixel) is pretty much negligible. The flat area is 4,908,738 m^2, while the edge area is 15,708 or 157,078 m^2 assuming 2m thickness. What that means is that shattering the pack into single pixels only increases the exposed surface area by 3%.

http://nsidc.org/dat...w_ice_analysis/ Key section "This is a manually created product which uses multiple images to map the snow/ice regions. Surface data is also made available to the analysts to aid with real-time quality control. Regions covered by cloud during the 24-hour analysis period are generally mapped as persistence, taking lower resolution passive microwave data and surface observations into account where possible." What this means is that IMS is primarily based on visible wavelengths. This has two major implications: i) It's much more sensitive to low concentration thin ice - you can verify this yourself: just look at MODIS images in regions near the ice edge where the passive microwave sensors show nothing. In general this would mean that IMS will lag behind the microwave imagery during melt season (it continues "seeing" the ice for longer), and lead it during the re-freeze (it "sees" the new thin ice before the microwave sensors do). ii) It's also much more susceptible to cloud interference. Although there is cross-checking with the microwave data, cloud covered areas are in the main only updated when the clouds clear - i.e. the previous state is carried forward by "persistence". This means that the IMS product will lag reality when there's a lot of cloud around. The combination of these two factors is why IMS tends to read higher during the melt season, and also show a few days' lag in responding to specific events such as the recent storm. If I remember from last year, IMS was ahead of the microwave products during the re-freeze. This indicates that (i) is more of a factor than (ii). TLDR version: IMS is probably a better indication of reality most of the time, but can be misled when it's cloudy. Also, much of the ice cover in the IMS product is incredibly thin. Edit: fixed punctuation :-)

Let's not over-egg the export factor. Fram strait is about 500km wide. Assuming we would get a new record of 4 million without any export at all, then to get below 3 million requires an extra 1 million loss through Fram in 6 weeks. Fram is 500km wide, so that requires the ice to drift 2000 km in 6 weeks, or ~45km/day. That comes out to 50 cm/s, which is about ten times the usual maximum drift speed for central basin ice. Even the current storm is only pushing ~20 cm/s drift rates. It's just not physically plausible to export that much that fast.

Interestingly, the drop not showing up on the NSIDC or Uni Bremen graphs yet. They're both 5-day averages, which may partly explain it. Also, both use a 15% threshold, so it may be there's a load of ice that's just dropped below 30%. Consistent with that, IJIS is dropping fast, but not as fast. There may also be differences in how the various algorithms deal with cloud, water swells overtopping the ice, etc.

Hmm, in citation research, impact factor is assessed by looking at papers published in the last two years, using that as a metric of immediate interest. So I think it's fair enough to call March 2011 "new research". It's new enough that it'll still be spreading within the research community, and there won't have been much substantive follow-up publication to date.

Summer temperatures in the Arctic are critical, agreed. The DMI model does not tell you the relevant temperatures. It tells you the temperature at 2m from the ground, i.e. right next to the melting ice. That temperature is forced by the physics of the situation to hover near zero. During the summer melt, there is an Arctic-wide temperature inversion, with the surface being at (almost exactly) zero, with warmer air above. To know how much heat is being transferred into the ice, you need to know the atmospheric temperature above the inversion layer. A good proxy for this is the 850mb temperature, which BFTV regularly posts for us along with useful commentary. Edit: Saying the Arctic ice is fine because the DMI temp is nice and steady makes as much sense as saying a naked person in the Arctic is fine because the temperature of the air inside their mouth is still at 37 degrees. That temperature won't change until the patient's dead. Likewise, the DMI temperature won't go up until after the ice is gone.

Different wavelength I think

Um, this isn't stuff that requires peer review, it's very very basic science. Open water during the melt season increases heat absorption, because water is darker than ice. This is not a controversial fact! This is what GW was talking about in post #283. To state, as 4wd seemed to, that increased open water leads to summer heat loss is a logical impossibility. You can't say that the Arctic is simultaneously gaining heat (i.e. melting more ice and exposing more water) and losing it (radiating more heat from the now-open water). I presume this is why GW was quite so incredulous about being contradicted on it, and he's quite justified in that! The flip side is that open water during the re-freeze increases heat loss, because ice is an insulator. This again is not controversial. It's an open question as to which effect is stronger overall - but note that this relates more to loss of winter ice cover than summer ice cover. That is, it's a question of whether the increased ice formation during the re-freeze is sufficient to offset the increased ice loss in the summer. In fact, it's precisely this extra heat input from the open ocean to the atmosphere above that may (note cautious phrasing) be affecting autumn and winter weather patterns. This last part is not settled yet - and note that GW posed this as a question in #283. Yes, there are real unknowns and scientific controversies related to Arctic ice melt. The fact that in sunlight, exposed water absorbs more heat than ice is not one of them!

Those are supraglacial melt lakes - consider the scale. A single pixel is 250m, so those features are kilometres across. That would be one heck of a moulin...

You can't put a trend on those without removing the seasonal cycle first (e.g. by calculating the daily anomaly before doing the trend analysis. Otherwise your trendline gets misled by the fact that the data series is starting just before a peak and ending just after a trough.

http://tamino.files.wordpress.com/2012/03/bymonth.jpeg NSIDC ice data: complete series plotted for each month. As you can see, there is a clear downward trend for every single month of the year, although the trend is more pronounced in summer months. The Antarctic has an upward trend, which is much smaller, and is (if I recall correctly) not significant for most months.

That's a remarkably teleological argument. "Supposed" by whom/what - was the polar ice cap designed? Self-evidently not. But if you strip away the spurious imputation of purpose, what are you left with? Are you just stating that when things get hotter, ice melts, and thus absorbs some of the heat? Thanks, but I think we all know that already. Attributing purpose to physical law is a stretch too far. The question is when the downward trend is going to stop, and how. After all, using your false language of purpose, the ice cap is "supposed" to get bigger again when the climate gets cooler. What's going to bring that about? It doesn't. This too is self-evident. You just have to think about what the language of "heat escaping" actually means in physical terms. Heat doesn't escape during the summer - that's the melt season when there is heat input into the ice. Heat escapes during the re-freeze. It's true that increased ice loss (lower summer minimum) leads to increased heat escape (more ice formed during re-freeze). This means that the annual cycle gradually gets wider, with more ice loss in the summer followed by more ice re-forming in winter. But these effects are not equal and opposite. If they were, then the winter maximum would stay constant. It hasn't: there is a downward trend in the maximum as well as the minimum - for volume in particular as MYI gets replaced by first-year ice, but also for extent and area.

I don't think the "blue ice" phenomenon should affect IJIS unless it's being seen as under 15% concentration. Looking at the images here, IJIS counts everything other than open water when calculating extent - you can see the East Siberian and Leptev seas are still counting as fully ice-covered except for the polynyas north and west of the New Siberian Islands (which are clearly visible on the MODIS mosaic). So it looks to me as though the losses are real. Looking at the high-res images, you can clearly see the blue ice beginning to melt through to open water in any case - the whole region is peppered with darker blue dots. http://www.ijis.iarc.uaf.edu/cgi-bin/seaice-monitor.cgi?lang=e http://lance-modis.eosdis.nasa.gov/imagery/subsets/?subset=Arctic_r05c04.2012162.aqua.1km http://lance-modis.eosdis.nasa.gov/imagery/subsets/?subset=Arctic_r04c05.2012163.terra