The first question at a "demise" talk might be 'what are you predicting for summer 2013'. Wipneus has probably given the best overall answer to that -- the geospatial cell-by-cell exponential downtrend in volume in coming years -- but a temporary September minimum area uptick will still perplex a non-specialist audience. So what is going to happen -- can we find a set of proxies today that predicts a record -- or at least respectable -- melt season? Craig M sees the 'Beaufort getting interesting real quick'. I also think that likely based on open water and melt on both sides of the Bering Strait. I've been keeping an eye on three other developments -- the recent gyrations in the Kara Sea, anomalous temperature brightnesses in upper Nares Strait, and the collapsing arc between Banks Island and Cape Bathurst -- based on newly calibrated 89V Ghz Jaxa imagery which we have not utilized previously on this site. That is most easily obtained by rooting around in their file system, starting here: http://suzaku.eorc.jaxa.jp/GCOM_W/JASMES_daily/data/AMSR2/ The temperature brightness scale here is Kelvin. It is related to but not quite the same as near-surface ice thermometer temperatures which we can track better from thermistors in the buoys or the AVHRR (often useless these days because increasing cloud temperatures (to -0.6 ºC today) have become opaque to the satellite's infrared sensors). The Kara Sea area lies northeast of the tip of that elongated Siberian island Novaya Zemlya whose Holocene sea level / isostatic rebound after Weichselian glaciatian is reconstructed here). After enhancement and tiling of the Modis 1 km swath r03c05 on 20 May 2013 (day 140), you can see the region consists of open water, floes, and thin gray ice binding them together. The far lower resolution Jaxa microwave animation shows a broader region over the last month; it is rotated 90º relative to Modis still, with the last day 21 May 13 paused. You can see some outright ice disappearance on the northenmost Barents Sea side and possibly portentious churning in the upper Kara. However this region has had an on-and-off history since March, it is a difficult proxy to develop for overall melt season trend or inter-year comparisons but early open water there would have consequences. The Northern Sea Route -- Vilkitsky Strait and all that -- and its inconsequential future signifcance for shipping as recounted at wikipedia -- makes interesting reading vis-a-vis glowing press accounts. The ice arc bridging the Amundsen Gulf between Cape Bathurst and Banks Island collapsed dramatically last year on 27 May 12, if you recall that animation from "On the Move". Initial collapse this year occurred on 19 May 13 as the ice pack shifted and may progress further in the next few days. Ice deeper within the Gulf is typically 2 m thick. Despite being a week earlier than record year 2012, the region is again a problematic guide to progress of the overall Arctic melt season.

The front row seats to the demise of Arctic sea ice just got better. May 17 was a big day for Shizuku (droplet, not water 'mizu' as you might think from the W in GCOM-W1), our favorite satellite with its 100 minute polar orbit and 6x2 channels of cloud-translucent passive microwave emissions measured onboard by the AMSR2 instrument -- Jaxa began offering ground and radiosonde calibrated brightness temperatures (TBs themselves from 25 Jan 13 on) interpreted for you as sea ice concentration, sea surface winds, snow depth and so forth. Each of these requires a journal article algorithm to get there. Those for sea ice concentration include the Nasa-Team (NT2, Markus 2000), Artist (ASI, Spreen 2008), and AMSR Bootstap (ABA, Comiso 2003). My current understanding is that NSIDC uses a different meteorological satellite F-17 and overlapping sensor sequence SSM/I, SMMR, SSM/I and SSMIS with NT/ABA, Bremen uses ASI and the 89 Ghz channel only, and Jaxa the emissivities (from temperature brightnesses) of the 6.9, 18.7 and 36.5 Ghz channels in ABA, with sea ice extent and sea ice area being further derivative products of sea ice concentration. While these are midway between misleading and useless from mid-September freeze-up to June melt season onset, this data could provide an albedo product, yielding daily and seasonal changes in the radiative heat transfer balance -- the reason what happens in the Arctic doesn't stay in the Arctic. Here the solar surface radiation is coming from a 6000 K blackbody whereas the emissivity is that of a 273 K graybody, so according to Stefan-Bolzmann (T to the 4th), we're toast once the summer ice melts but still freezes over in fall. Although a lot of imagery has long been offered directly at their overlay facility (even better resolution from drilling into their file system), anyone can register for custom data. The reason for registration being they need your email to send the url to the data after your request has been processed. It's all explained in great detail in their documentation. Here is what they are providing in terms of product, ground pixel resolution, pixel area, and channels used in the respective algorithms: Below I've attached an animation takikng 36v, 36H, 18V channels as RGB from 01 April 13 to today 21 May 13. It is at native resolution, which means for the relevant part of the Arctic Basin, some 90,000 pixels or 60% of the pre-masked image.

Jim, I don't see any topology operative here. Perhaps refrain from citing Thom's catastrophe theory unless you have a Ph.D in geometric topology and 10 years of related research experience -- otherwise you have not and cannot read the papers. Higher dimensionality of the classification makes it utterly inapplicable to the physical sciences as far as anybody knows -- proposed applications to predicting prison riots and when dogs will bark are simply preposterous. New Age buzzwords are not appropriate in a scientific blog. Metastable states across many fields of science are commonly described by potential wells. The potential could be gravitational, covalent bond strength, gibbs free energy of protein denaturation, etc etc etc. Because of Occam's Razor, nothing more is warranted. The figure below shows three wells labelled A, B, and C. If the system is originally in Well A and natural variation is only enough to bounce it around withing the region indicated by the arrows, the system is stuck in its well (here called an equable climate). Should Arctic sea ice in conjunction with unusual but natural variation weather push the system 'over the top', it could come to rest in the next metastable state, Well B. Metastable states B and C need not exist. According to the precautionary principle, we should not invoke them without basis in regards to northern hemisphere weather subsequent to the coming loss of summer Arctic sea ice. It is quite feasible for the system simply to tank, as it would here after leaving Well C.

Patrick says, "After an absence from blogging I'm back, as someone or other famously said." Well, that would be the Terminator himself, Arnold Schwarzenegger. http://www.youtube.com/watch?v=WgPePk3kGZk I'm back too, though like a T-800 droid at the end of the movie, somewhat the worse for wear and not fully operational. The smoking gun points at open water on the 15 May 13 Jaxa image. And I do mean open water all-liquid-phase watery wet water. It's real easy to measure its daily area -- and more nuanced compositions -- with a click of the mouse so we hardly need an external product for that. In fact, I have a long list of grievances with 'extent' and whether we should even be lending credence to the concept with blog mention. First up, why do we have a daily albedo product for Greenland but not one for the Arctic Basin -- what's preventing us from cloning over the method of Jason Box? That is, extent gives nothing but abuse after freeze-up yet serves as a mediocre proxy for heat balance during isolation season. Second, it's better to focus on just the Arctic Ocean and put aside the ice east of Kamchatka and at mouth of Saint Lawrence River (same latitude as Paris) etc. Check out the mid-winter NSIDC extent map to see what all they're including. I'll put out a couple of options for defining Arctic Ocean boundaries shortly (as masks for the common types of satellite imagery). Third, I have the distinct impression that very few blogging away on sea ice extent know how extent is operationally determined, in the sense of being able to replicate its computation. By replication I do not mean writing someone for the code, compiling it on your platform, and repeating the calculation on the same data. Replication means starting with how 'extent' is defined, writing an algorithm appropriate to that from scratch, applying it to your favorite applicable satellite imagery, and coming up with more or less the same answer. We can't just take other people's products at face value -- we have to pick them apart like Wipneus and Chris did with Piomas. The tricky part is the "15% ice" rounded up to 100% (because NSIDC says it "might" be ice covered by water) that differentiates extent from area. What exactly does that mean when a single structureless NSIDC extent pixel is 25 km on a side? While 625 sq km of area is suitably small relative to the 14 million sq km freeze-out (note the Arctic Ocean relevant to climate change is much smaller), it's still a really big pixel -- 11 times the size of Manhattan Island (58.79 sq km) -- relative to intrinsic dimensions of sea ice features such as ridges, fractures and floes. Unreplicable products are the curse of climate science; after some time goes by, no one has the slightest idea any more what physics was considered or wasn't, what simplifying assumptions were made and why, whether the product is still applicable at the end stages of ice loss. Yes, if you had full text of all the citations in all the citations of all the citations of all the initial citations, this all might be spelled out somewhere but no one has time or text access to routinely pursue that. For example, the 15% was intended for ice edge, yet perimeter will become vastly more extensive in coming near-terminal melt years -- is the TB (passive microwave emission brightness temperature) prescription still applicable? In summary, I favor migrating out of products we cannot replicate into in-house products on a programmatic (blogwide) basis. We've seen from buoy data that it is rather easy to do this for volume, while extent and area are just mouse clicks and albedo is like Greenland. If 38V is all they're using for Greenland daily melt, we can three channels on the sea ice. Some products we can do faster and others better. Better because we are free of the constraint of tying back into 1979 ice: we can work with modern imagery that need not go too far back. 1979 is flogging a dead horse to begin with -- you can't radically change the chemical composition of your planet's atmosphere without there being consequences, not when it starts filling an infrared band gap. So let's get out of reactive mode and look more towards predicting the future.

Actually dust storms in the Taklamakan, though peaking in April, can occur at any time of year -- attached is a notable 02 Nov 12 event as caught by Modis. Since we can flip easily through the rapid response imagery, the Greenland dust event putatively associated with the Dec-Jan SSW is accessible -- its plume track, if any, is can be followed with satellite CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) and/or numerical simulation. On the Greenland side, it's important not to conflate snow pits with ice cores. The former represent melt down of some 150 kg of snow per horizion and only sample a decade or so back in time, attaining a time resolution of only two months; most studies have saved only the fines. Cores can go back 125,000 years but an annual layer at depth has a millionth the starting mass to work from -- a whole decade may need be melted to get the milligrams of dust required for strontium 87/86 and neodynium 143/144 analysis -- there's no seasonal resolution at all. Back at the Taklamakan, because the peaks surrounding the basin run to 5,000 meters, raised dust often stays within the basin. But if not, the dust launched higher in the atmosphere implies a higher energy weather event. This dust has to move north and stay high to deposit at 3,000 m core sites in northern Greenland. Since everyone agrees dust in northern Greenland comes mostly from the Taklamakan, the question is how much of it, particularly the larger particles harder to get aloft, is associated with SSW events. That is going to depend very much on the definition of SSW and whether they, as proposed, originate from the influence of the unique topography of the Talim Basin on weather patterns. If we now start to slowly dial back the definition of SSW, what are we left with? Is there in fact a distinction between the mildest admissible SSW and a conventional weather pattern strong enough to lift dust out of the Takim Basin to an upper troposphere atmospheric elevation? I would suggest not -- the mechanisms under discussion here for strong events may moderate down to long-known Chinese weather features giving rise to weak to moderate out-of-basin dust storms, described in a large literature. (Ironically China itself is minimally affected, as dust from these events largely fails to descend onto their network of 146 surrounding long-term meteorlogical stations.) About a dozen out-of-basin Taklamakan dust storms occur per year, a far higher frequency frequent than SSWs, depending on how narrowly these are defined, eg if rapid warming the top of the troposphere is not deemed good enough and disruption of the Polar Vortex is required. Meanwhile, an upper troposphere plume originating from the Talim Basin in 2009 took 13 days to circle the globe, which it then did a second time . "The dust ball was formed when a wind storm ripped across the desert, kicking up the dust, and trapping it against the mountains of the Tibetan plateau. The scientists said [in Nature Geoscience] the 800,000 tons of dust was forced higher and higher into the air, until it reached an altitude of around 16,250ft. A warm convection flow then lofted it further to between 26,000 ft and 32,500 ft. The dust was then trapped in the polar jetstream, a fast-flowing air current that lies just under the stratosphere, and began its journey around the world." [news clip] "All dust storms in China are associated with cold air outbreaks, which result in the occurence of frontal systems and the Mongolian cyclonic depressions. The routes of cold air can be divided into north, northwest and west ...The dust storms in the Taklimakin Desert are associated mainly with the western route of cold air outbreaks.... Only the strong upwards wind created by highly unstable synoptic systems during the spring may provide favorable conditions for the dust to easily escape the basin." "As the cold air from the northwest broke into the Taklamakan Desert from its northeastern entrance, it provided strong motion to mobilize the dust. The northeasterly wind was presumed to lift the dust in the Taklamakan Low toward the southwest, but because of high mountains, the air masses were elevated to around 5 km agl and merged into the conveyer belt between the Taklamakan Low and the Tibetan High, which is clearly indicated by the back trajectories and the NAAPS model. [ref 3] "Indeed, dust from the TM is entrained to elevations >5000 m and often takes a northern route before being transported in the westerlies... Dust is first transported to the north, reaching 50°N, before moving eastward over Lake Baikal and northeastern China which makes it more plausible for the Taklamakan dust to be transported towards higher latitudes, eventually reaching Greenland." [[Sun et a 2001 and ref 4] http://onlinelibrary.wiley.com.oca.ucsc.edu/doi/10.1029/2000JD900665/pdf https://www.jstage.jst.go.jp/article/sola/8/0/8_2012-025/_pdf http://onlinelibrary.wiley.com/doi/10.1029/2002GL016446/pdf http://www.atmos-chem-phys.net/10/2615/2010/acp-10-2615-2010.pdf

I looked for the ultimate basin map, an oblique shaded relief map of the Taklamakan and surrounding mtns with exaggerated vertical scale, no luck on that. Here are a couple of on-topic papers from 2006. Pasting the titles in google scholar gives 39 newer articles citing the first and the 116 citing the second. Sounds like the dust plume from the 2013 event might be traceable from the satellite record ... spare us that trip to Greenland. Late Quaternary environmental changes in the Taklamakan Desert, western China, inferred from OSL-dated lacustrine and aeolian deposits Sediment records from the Tarim Basin of western China are of great importance for understanding Late Quaternary climatic variability in Central Asia. A chronology of aeolian and lacustrine deposits from the centre and southern margin of the Taklamakan Desert, central Tarim Basin, has been established using optical dating methods. Distinct variations in humidity during the last 40,000 a in this extremely arid inland basin have been identified. Lacustrine sediments were deposited in the centre of the Taklamakan during two periods of wetter than present day conditions at around 2000 and 30,000 years ago.... Sedimentological evidence for a late Holocene humid period are consistent with records in ancient Chinese literature. Wetter environmental conditions in the past within the Taklamakan, as indicated by the presence of lacustrine deposits, are also supported by data from adjacent regions. It is assumed that changes of global westerlies and of the mobile polar high triggered the fluctuations of precipitation in the study area. However, variations in temperature in the Taklamakan Desert are presumed to be mainly controlled by the intensity of the winter monsoon. http://www.sciencedirect.com/science/article/pii/S0277379105001976 A review on East Asian dust storm climate, modelling and monitoring In arid and semi-arid area of Asia, dust storms occur frequently. Asian dust storms have a major impact on the air quality of the densely populated areas of China, Korea and Japan, and are important to the global dust cycle... Much progress has been made in the development of integrated dust storm monitoring and modeling systems by making use of advanced numerical models, satellite remote sensing and GIS data. In this paper, we summarize the recent achievements in Asian dust storm research with an emphasis on dust climatology, modeling and satellite monitoring. The concept of integrated dust storm monitoring and modeling system is described and a summary of the developments in key research areas is given, including new dust models and techniques in satellite remote sensing and system integration... http://www.sciencedirect.com/science/article/pii/S0921818106000324

I'm wondering if Jason Box's croudsourced trip to Greenland will recover dust with Taklamakan's signature from the large SSW event you are describing, rather than (or in addition to) soot from the nearby Lapland fire that they are looking for. Here I recall they are melting down this year's snow column from some five sites along a southern Greenland east-west transect to concentrate particulate matter and its chemical constituents. At the much higher latitude core sites, such as NEEM, they've also melted out recent snowfall as a sideline to main drilling location. Unless the station had people over-wintering again, this winter's SSW would not have been studied. However it is no big deal to go out from Thule on a snowmobile. In discussing dust, every published study makes a fuss over particle size -- the bigger ones are harder to get high enough in the atmosphere long enough to be transported any distance but are easier to count and analyze. I'm recalling that they get just about straight Taklamakan dust at higher latitudes at the larger particle sizes. On the other hand, if you collect dust of all sizes out of Qaqortoq in extreme southern Greenland, with ordinary westerlies overhead, don't expect every mote will originate from Taklamakan. On the isotopic signature, this depends on geologists having sufficiently sampled potential source sites all over the world. Basically, they do a scatter plot of isotopic ratios and draw tight circles about all the samples from a given desert (or a sphere if a 3D plot). Over the last 15 years, as far as I know, the northern Greenland dust has always landed unambiguously within the Taklamakan circle, specifically the Takim sub-circle. While there is always some lack of finality -- maybe not all source deserts are represented, or those represented have unsampled sub-regions with different composition, it is rather late in the day for this type of objection -- geologists have been over the ground quite thoroughly by now. In terms of Holocene frequency of large SSW events, taken as Taklamakan fall-out layers in northern Greenland ice cores, that's already available from the published scan line record. The very first things done with a new core are from non-destructive sampling, here just optical scattering off the bubbles and larger dust (which are readily distinguished). Only the larger depositions provide enough material to source them isotopically and not every one that could be sourced would be sourced. So -- before looking at this data of putative frequency of large SSW Holocene events -- you might write up what your mechanism predicts here, are there specific periods of the Holocene during which SSWs would be more or less frequent?

Fantastic post! Just a couple quick comments -- I'm off on vacation in a couple hours: I'm ok with mercator projection -- it's a common choice for global display and we all learned in school that it distorts things at the poles. Conventionally, Mercator's cylinder makes contact with the earth's surface around the equator but here you are displaying temperature anomalies high in the atmosphere, at 10 HPa, a fixed pressure surface not a fixed height above the earth (ie not a concentric sphere). Thus there is a second projection from this colored surface down to the conventional cylinder, perpendicular to its axis, that gives the display as overlaid on continental outlines. People outside of meteorology will struggle with this aspect. Viewers can also walk away with the wrong impression because of the ratio of the frame speed of the animation to actual elapsed time. Here the event happened over 30 days (11 Dec 12 to 09 Jan 13) but is packed into 8 seconds (33 frame animation at 250 ms frame rate). It looks 'explosive' because of this 324,000-fold time compression. The SSW feature is right-sized already -- increasing map scale is not particularly helpful. Right now, the earth is given 394 x 220 pixels of the 523 x 305 pixel animation rectangle, which cannot then display properly on typepad (maximal width 415), dropping visible animation width down to 313 pixels (= 394 x 415 / 523) which is three-quarters of that attainable by turning the anomaly scale sideways and moving it under the map. The animation could be supplemented by a more intuitive view, an animated meridional slice through a quarter-earth and its atmosphere. It looks to me however like the event has a natural central track of development (see below) that is more of a rhumb line than a meridian (except for penultimate frames where it bends northwesterly), though some choice of great circle (rhumb lines are not geodesics) would work ok for the quarter-earth plane. Somewhere in the explanation of this SSW, the direction and speed of central anomaly propagation might be explained, eg is it the same for every event originating in the Taklamakan or influenced by other atmospheric or rotational considerations. Can the animation be extended to include more of its aftermath, the descent talked about so much in the text? This might be done better separately in polar stereographic projection since it is largely a polar vortex and NH story with the Taklamakan now out of the picture. Finally, minor point, the title of the animation is rather cryptic. For example, CDAS -- after-the-fact weather reanalysis, Climate Data Assimilation System -- could use an explanatory wikipedia link. Temp Anoms could be spelled out. And 11d rm is a mystery -- the acronym rm has 133 uses already, none applicable here (http://acronyms.thefreedictionary.com/Registered+Midwife).

Internal consistency considerations show this buoy data has nothing whatsoever to do with ridging, over-riding, or slabbing. I would call these a mix of traveller anecdotes about inconsequential areas, uncritically repeated received wisdom from the old days, and model fantasizing -- the track data proves them completely irrelevant over the long life span of buoy 2012L. The pixel resolution on the best satellite imagery we have access to is over a kilometer on a side. This is still small relative to dimensions characteristic of ice thickness changes. Buoy position is known far better. So what? -- toss the excess. The first thing to do with this precious daily buoy ice thickness data is overlay the buoy track on satellite images of the same date. You might wonder why nobody is putting out these daily products. I sure do. An overlay requires a different rotation, offset of the north pole, and rescaling for each satellite. The main thing here is to avoid anything that degrades (dithers) the precious pixels of the satellite image. To get negligable alignment error in the case of Jaxa radar, try: ... rotate Jaxa 180º ... scale buoy map by 52.13% ... offset buoy pole by (x,y) = (-1,-61) Below, I took a 3x3 pixel average about the point where the buoy sits today and made the reference color square in the upper right. If the subtleties of color space voxels in Jaxa radar correspond to ice thickness, then all the other pixels on the image of the same color will be within a narrow thickness range of the buoy, which has been indistinguishable for many months from 2.13 meters. The pale green overlay shows this, with a radius of 10 in rgb color space about the reference color. This procedure seems to work quite well vis-a-vis other ice thickness products -- could this be a coincidence? No. The only question is whether buoys lead to something better, which they surely do as a daily product. However once calibrated, the satellite data alone could suffice.

Speaking of the 2013 reality check, let's talk about real-time Arctic Ocean buoy data. These devices -- some just installed in April -- are embedded into and under the ice and report daily ice thickness to the nearest centimeter by upward directed sonar as their floe drifts along, so are not to be confused with bottom-anchored moorings watching the ice pass overhead, for which stored data is retrieved only annually. The above-ice parts of the device measure snow depth as well as air temperature and pressure. Awaiting clarification, but it appears the internal temperature of ice and sea water underneath are measured with thermistors at 45 different locations. It's easy to get the data because the Army has something better than a grudgingly open data policy: "We encourage the use of all data on this web site," only asking for a citation to the url below rather than future paper way down the road. The lead author, DJ Perovich, has been out on the ice since forever; his data was the subject of my very first post (which nobody wanted, this being an ice theory blog back then). Start here: http://imb.crrel.usace.army.mil/newdata.htm Perovich, D., J. Richter-Menge, B. Elder, T. Arbetter, K. Claffey, and C. Polashenski, Observing and understanding climate change: Monitoring the mass balance, motion, and thickness of Arctic sea ice, http://IMB.crrel.usace.army.mil 2013. I look at the downloads for my favorite buoys, 2012L and the goat's head buoy 2012J. The first is formatted as a flatflle database with 5,656 rows and 55 columns. The data is read out once an hour (once a day for the thermistor string), meaning each calendar day takes up 24 rows, only one of which has the profile temperatures. It covers the buoy's track from 27 Aug 12 to 20 Apr 13. To reduce file size, you could average out each day, index hourlys into a relational database, keep only the thermister line for each day, allowing concatenation of all buoys. Although file size here is not an excel-killer, a basic flatfile database has less baggage, allowing faster sorting and summarizing. The issue really is who has the most elegant graphing capability. The latitude and longitude of the buoys are given with incredible GPS accuracy -- on the 20th, it was at latitude 74.9399 and longitude -147.7217 -- and I don't think they're kidding about significant digits. This is far greater resolution than that of any of our ice models or imagery, the point being -- since the first order of business is to overlay real ice thickness and ice temperatures onto the 3 ice-penetrating radars, the infrared imagery, the Modis visible, Piomas cells in wipneus polar stereographic projection, and the Navy ice thickness animation -- we can pull out the properties of the exact pixel corresponding to the buoy. That's assuming we could locate lat-long accurately on these images. In practise, not one of them provides product map scale and no two are the same, not to mention rotated away from Greenwich by variable amounts. However I can work all these out easily enough from the sharp buoy mask provided. Just to whet your appetite, buoy 2012L is not reporting a single centimeter of ice gain or loss this whole time. It's stayed 3.35m thick. If this holds for a lot of other buoys, I'm planning to hit the trash button on a whole lot of model papers, products and posts. I suspect this observation will provide a very useful ice thickness calibration for radar -- if it stays consistent. The air temperature at the buoy ... that we want to compare to the near-surface infrared legend. And so on.

Good question, Espen. It looks to me like the minor crack has to be an artefact of some kind, as it continues right through the Precambrian rock and not even along a plausible route there. The major crack terminates at its top at a swath line. These images are composites from different satellite passages. So this requires very rapid formation of the crack at lower swath time, but after the upper swath has been taken. If artifacts, I have no idea what could have given rise to them, as they don't appear to be individual scan line malfunctions. However the 250m resolution is a stretch -- I find it a bit pixellated, so not worth the file size pain.

Here is yet another very serious miscommunication from Stroeve on 17 April 2013, picked up in typepad's news link column. This is not someone mis-speaking, but part of a very consistent pattern of public statements over many months. "Scientists aren't certain how long it may be before all of the Arctic sea ice disappears in the summer months. But most scientists "think it's going to be in the next two to three decades," said Stroeve. http://www.insidescience.org/content/arctic-sea-ice-leaves-record-small-footprint/984 It's correct on its face: sure, if you wait until 'all' the ice is gone in June (a 'summer month') and poll scientists who are 'mostly' modellers who aren't about to admit their life work was a colossal blunder, you can arrive at 2033-2044. The public and policy makers won't see the extreme spin here. What they hear is, NSIDC says is not our problem, it's 25 years off and might not even happen -- it's just a distant extrapolation, not even a consensus on 2040. The fact is, this statement grossly understates the timing and climate change implications of Arctic melt-off and the consequent policy urgency -- the Arctic is already failing as planetary heat moderator. By 2015, the full impacts of this loss will be almost entirely upon us. Setting the goal post so high -- every last ice cube must be gone by the first of June or it doesn't count (01 June is defined as meteorlogical summer http://en.wikipedia.org/wiki/Summer) is to live another 25 years in serious denial. And then wake up to a truly hopeless situation. I wonder what drives this criminal evasiveness in so many climate scientists. To be frank, I think it is all about covering their butts. They screwed up royally, it is all on the public record, and now they're trying to put off the day of reckoning as long as possible, walking back their 2100, their 2080, their 2050, their 2040 a decade at a time, while they still can.

Wayne writes, "of previous 4 seasons, 2013 has earliest underside melt. Will add evidence this weekend. The camera height and image quality is better, smooth surface first year ice. Now for photoshopers like A-team who know refraction technique http://eh2r.blogspot.ca/, horizon height vary because of height of camera, hopefully a good solid mount. This variance: state the ice is, melting, freezing or steady, with time, variance measurements how thick it is captures ice radiation budget." You have gotten onto something quite interesting here measuring refraction, Wayne. Image processing software could make four small improvements here, objectively (consistently) locate the horizon to fractional pixel level and do so under somewhat cloudy/foggy conditons on a large volume of imagery. This depends on a rock-solid, windproof camera mount. Below I expanded your split horizon from the Redondo Beach web cam to pixellation. It would be slightly better to keep the initial camera file format (png or tiff) rather than lossy compression (ie irreversible) to jpg. It would be fantastic if your horizon included a ice mass balance buoy. These measure and report daily on temperature, pressure, snow and ice thickness variations. The best buoy in the vicinity of Resolute is 2013A off Grise Fjord, deployed on 22 Jan 13. Although not moored, it hasn't budged since being set out. It is currently at 76.39 N, 82.89 W at 26.11º C, 1008.42 mb, no snow cover, on 1.34m thick ice. http://imb.crrel.usace.army.mil/newdata.htm

Paul, papers take forever and a day. Only one paper in four submitted to Science is even sent out for review and only one in three of those is ever published. Behind their paywall. Where it doesn't move the needle. Not that it would anyway. That takes a title like "Life after SSWs: will there still be a cannabis crop?" Last week, I had a paper (hopefully my last) accepted for May 2013 publication in PNAS. I finished my end of it in mid-July 2012. And this is with 2 of 3 reviewers saying it represented a major breakthrough unifying two large academic disciplines. I favor real-time open-source publishing. For the last ten years, I've just posted daily work-in-progress on the local campus implementation of wikipedia. That's a great collaborative environment (that you control as originator) and best of all, has infinite undoes, is free of length constraints, but provides no deleting. Knowing I'm going irrevocably on the record that day induces me to get my data and exposition halfway together as it will show up forever on google search. You might think some jerk -- like Watson and Crick -- would come along and steal the ideas before I could write it up for a journal. Hasn't happened -- most scientists have more integrity than that. Mostly I have met collaborators who shared my interests. Whether the other millions of visitors got useful information or quickly realized their mistake, I couldn't say. Journals have not taken this as pre-publication but instead have allowed it as direct supplemental material. The bottom line however is someone has to sit down and create real content. Having a bunch of mice sitting around waiting for a volunteer to bell the cat just doesn't go anywhere. Good news and bad news: The bad news is now I will have a lot more time for climate blogging. Since that amounts to typepad haiku, I've been laying down bits and pieces the last few months so future posts can be more in-depth, just citing older ones for background. The good news is from Monday on, I will be offline camping in southern Utah until mid May, with motel internet only during bad weather. So post away, no adult supervision.

Scott, that's right, ice melts at 0ºC on the celsius scale so the infrared satellite images of the ice pack will always be in minus territory. Actually, zero is defined as the triple point of Vienna Standard Mean Ocean Water (VSMOW) -- Andy Celsius himself set 0ºC as the boiling point of tap water and 100ºC as the freezing point. VSMOW has a defined isotopic compostion that is very important to evaporation, precipitation and interpretation of paleo ice cores and the overall geochemical record, namely deuterium at 156 ppm, oxygen 18 at 2005 ppm and oxygen 17 at 380 ppm. The ice covering the Arctic Ocean is a long ways from VSMOW, primarily because its salinity -- which varies both geospatially and vertically according to the extent to which brine exclusion (aging) has proceeded -- lowers the melting point. Although salts have no place in an ice crystal lattice, the ice pack is hardly one big ice crystal, so the salinity of summer meltwater ponds on the ice will vary according to source, stage and deposition from waves and wind. Dan Fahrenheit's scale -- which the US will use until hell freezes over -- was actually more useful: 0ºF is the freezing point of brine (water saturated with salt, 26% ). In other words, liquid water can stick around until −40 °C, as it might in a tight protected brine channel. The camera on the AVHRR satellite measures near-surface temperatures of the ice, or failing that, top temperatures of an intervening cloud. It does not see the intervening temperatures of the air column, that is determined by other means. The temperature and salinity of the seawater just underneath the ice can be measured by sensors embedded in the ice. In winter, there might be a few dozen of these still working, spread out over 13 million sq km. Careful geodesy plus modeling plus historic data can also get at density and so temperature. I have no idea what daily SST products offer during wall-to-wall mid-winter ice. Next up is the heat equation, as with subsea methane clathrates. That determines the evolution of the temperature gradient of the ice slab, sandwiched as it is above the relatively warm water and below the much colder winter air warming in the spring. This is complicated by just where heat from the sun is adsorbed -- reflected from a dry snow cover, scattered about and eventually captured within the ice, transmitted down to photosynthetic diatoms living in bottom brine channels, or down a few tens of meters to pelagic plankton in the water. Using the grayscale legend on the photos, I showed earlier how to color the whole Beaufort Sea ice by its near-surface temperature. That could easily be animated. More simply, take a recognizable floe -- say that one resembling Bill Clinton in profile I started a couple months back -- and just color its temperature over the season using the ever-changing legend as it swings around the Beaufort Gyre.

Sure. Note I posted this back on 29 Jan 13. It is just taken out of GISP2, NGRIP and NEEM ice core publications and less relevantly Dye-3 in southern Greenland. They look at strontium 87/86, neodynium 143/144, uranium 238/235 and less commonly hafnium 176/177 isotope ratios. Sahara dust (which reaches the US east coast) and deserts of the American southwest seem alot 'closer' to Greenland than the Tarim Basin, yet to get to northern Greenland latitudes of GISP2, NGRIP and NEEM, only the Taklimakin dust gets wafted high enough (ie into the upper troposphere, ref 1 below). In these ice cores, the loess quartz dust is primarily from Luochuan in north central China. Dated tephra is almost entirely attributed to eruptions of nearby Jan Mayen and Iceland volcanoes with rarely input from Alaskan. Prevailing westerlies do not notably bring in tephra from the western US or massive Ring of Fire events as the latitudes of origin give them downwind trajectories that don't deposit significant amounts in northern Greenland Note these cores go back to the Eemian, so the dust events history covers some 125,000 years, though sometimes melt pooled from several years is required to get enough for analysis. So to associate them consistently with SSW events requires that the latter's geographical generating mechanism be operative for many millennium. Conceivably, the core dust profile could provide a long term record of major SSW events, at least the ones significant enough to descend on Greenland. You could start with these and their internal literature reviews: https://www.jstage.jst.go.jp/article/sola/8/0/8_2012-025/_pdf http://www.climate.unibe.ch/~stocker/papers/lupker10epsl.pdf http://web.missouri.edu/~jg2kc/clay/Svensson%20et%20al.,%202000.pdf

Jim and wanderer, I see what you mean, now the cracks are cracking. Looks like concentric shear lines are developing from today's Beaufort shot. It means the motion of the ice pack has changed recently -- and dramatically -- in this region. Meanwhile, over in the oldest of the oldest ice, the safe-haven triangle above the Nares Strait has continued fracturing way landward of the common coastal arc fracture. The other thing to watch is the ice surface temperature creeping upward (provided in the upper right of the avhrr imagery). Beaufort is now showing 12.8ºC, Ellesmere 15.9ºC. It's warming by the day. Remember this is not your grandfather's heat diffusion equation any more -- the ice is not a slab between cold sea water and thin dry poorly heat transfering but even colder air above because the sunlight is trapped and warming the ice from within.

I've been appalled by the number of allied sites that are propagating a very serious misunderstanding about this cracking. The qualitative view is not terribly alarming: yes, the ice has cracked before. The quantitative view paints a very different picture: no, it has never cracked this much, this early, this pervasively in the Beaufort, and not ever to this extent in the extreme multi-year non-coastal ice. I posted earlier a simple method for year-to-year comparison of crack density (the linear kilometers of cracks of 1 km or greater width per unit area) between any two dates since the first of February. This is tantamount to counting future floes, measuring mean floe size, evaluating total future water perimter. This takes perhaps 5 minutes -- mostly locating the files -- and reveals a colossal discrepancy between 2013 and 2012 (or any other year in the satellite record), any way you slice it. The Arctic Basin this year has hundreds of thousands of kilometers of cracks -- as far as we know, not remotely with precedent in human history (the ice was way thicker beyond the satellite observational record). The ice has crossed a major threshold into the structure failure domain relative to the same old forces it used to resist quite well. It's like a beer can -- 20 years ago, only a circus strongman could crush one, now a child can. For cracks, infrared is far better than Modis visible -- these sites are terribly naive to fall back on Modis in this situation. I suspect the moderators are unacquainted with the light spectrum.

Thanks, Wipneus. Below i binned (posterized) Ascat radar every 4th day from 01 Jan 13 to 17 Apr 13 that was masked to the Arctic Ocean. The number of grayscale bins was taken as 10, which enough to put the multi-year ice into 3 categories of declining backscatter which were chosen to just capture the goat's head feature over the full range of dates. I then entered the pixel count for each class into a spreadsheet. The fourth category is first and second year ice; the ice lying outside these groups were consolidated into a minor 5th category. Pixel counts are readily converted into sq km since the known area of the basin corresponds to 90,118 pixels for Ascat native resolution and the masking chosen. Total multi-year ice, the sum of the first three categories, peaked in early February and has been declining about 1% per week ever since (last frame), modulo a bump around Apr 10th that may reflect lateral thinning as warming multi-year ice can spread out. The ice is notably more mobile than in past years. It is running at about 83% of the multi-year ice of the same date in 2012. Years could also be lined up by date lag, ie when in 2012 (Feb, Mar?) did we have the same amount of multi-year ice? Here I may revisit the masking to exclude more of the ice north of Svalbard or restrict to contiguous bin areas going north from the CAA. I have not yet varied bin number to check the sensitivity to that choice. Posterizing also needs to work consistently from 2010-2013, the years that Ascar data is available. While pixel counts are nominally just areas, ice-penetrating radar is actually measuring something proportional to thickness. So once IceBridge or Piomass or on-the-ice data have provided the calibration, we discard them and just use Ascat for daily thickness and its trends (rather than some lagging indicator), at least for the rest of the spring.

After so much weather talk limited to proximate causes if that, I'm looking forward to a deeper explanation by which "nearly every SSW event in south central Asia arises from very specfic geographic trigger points where topographic lifting create upwardly directed Rossby wave". In effect, this would be saying SSW result from the collision of India with Asia. (I'm guessing here that the trigger points are by the Tibetan plateau.) I wonder if SSW arising there could explain the fact almost all dust in Greenland ice cores originating from Tarim Basin in the Taklimakan Desert of Mongolia (based on isotope ratios). A lot of climate is driven by land distribution -- the 45 meter sill at the Bering Strait comes to mind. (Sill depth hasn't been significantly affected by isostatic rebound or erosion.) Adapted from http://www.globalwarmingart.com/ Ocean vs atmosphere driving weather? ... I wonder if a clever extra-terrestrial with reams of reanalysis data but only globe diameter, tilt, angular velocities and solar inputs could deduce our current land distribution and topography (including ocean sills) to some degree of resolution. Or vice versa.

Just to follow up a bit on R. Gates, I found WorldView (as discovered for us by Jim H) quite convenient for animated comparisons to past years, in this case collapse of the Amundsen Gulf ice shelf west of Banks Island. In recent years, that has been the first to go along the CAA, before the Mackenzie warm water plume kicks into the Beaufort. If fracture and melt observables of 2013 continue to run 5-6 weeks ahead of 2012 -- as they have since February -- then we might expect the Amundsen to disintegrate quite a bit earlier than the 27 May 12 of last year. Five weeks earlier would be 22 Apr 13. Indeed, already on 17 Apr 13, the Amundsen looks 'on the verge' on infrared. Even the Mackenzie is shelving a bit. (Because of variable cloud cover and image quality, the picture below is a composite of 5 separate shots, with the most useful part of the most recent showing and the rest made transparent.) Jaxa radar confirms the precarious state of this ice. We are looking at a prodigous, unprecedented melt season in 2013 if this pace continues (or picks up).

Lieber Wipneus, could you please add a few explanatory details as to what you have plotted on concentration vs thickness? I'm not quite understanding the use of individual symbol variation, how June lumps daily sea ice concentration, statistical definition of the blue line, where the individual 5 years are contributing, the sharp slanting line on the left and so on -- thanks! https://sites.google.com/site/arctischepinguin/home/piomas/grf/pm_thick_conc_6_31-35.png If I am following your thinking on "thinner ice in PIOMAS comes with lower concentrations", the lion's share of Piomas output is largely anticipatable from concentration. Since sea ice concentration is little more than Ascat radar, we might be better off just using radar directly (with one-time calibration from IceBridge, Piomass, embedded thermister trains, etc). Then the simplest way to get accurate daily ice thickness (and the forward trend line) at the most interesting time, winter and spring, is just to mask the Arctic basin, posterize (bin grayscales, possibly non-linearly for thinner multi-year ice slices) to the desired number of ice thickness classes, count pixels and plot. Four years of Ascat data -- very manageable. However there is a big assumption here that has to be checked, namely that satellite, instrument and signal processing nitty-gritty leading to the posted daily Ascat image does not drift or get renormalized from day to day or year to year (like AVHRR). And Ascat ice thickness is only the bridge to even better Jaxa ice thickness. There's much more information in 3 channels than 1, but increasing complexity of analysis comes with that. Beyond that, we actually have 5 combinable channels including the 89ghz. That gets us into optimal dimensional reduction, a well-trodden path in remote sensing. Piomass grid cells seem to include all sorts of ice irrelevant to main basin evolution -- in the CAA channels, Hudson Bay, Fram, seas of northern Europe with wildly fluctuating conditions and so forth. I don't know if the single volume number excludes these, which are nothing but (loud) noise. If not, more sensible masking of the radar would get at the 2013 melt issue much more sensitively. Not to mention far better ground resolution. Finally, I should say Wayne is doing excellent work with his optical horizon monitoring -- he is getting actual local melt season onset, and quantitatively too.

Greg writes, "a couple of years ago the trouble with PIOMAS for "civilians" was that it was only gridded - it was popular demand that led them to put out a daily single summary number. To summarize, (1) radar backscatter is used to estimate sea ice concentration/area/extent, (2) which Piomas takes as its primary empirical input (not being an ab initio calculator of anything) and (3) outputs a gridded data table of ice thickness which understandably proves too geeky for end users so (4) only releases the summary single volume number on a monthly basis and (5) the gridded data on an annual bases and (6) never once in 108 months of operation ever converts gridded thickness data out of its goofy coordinate system back into the world standard polar stereographic coordinate system of the original radar image so (7) when Wipneus and Chris first did this, (8) making the first sea ice thickness class colored maps, it represented the first time that the ice thickness output map could be directly visually compared to the original rader input map which (9) lead within ten minutes to my initial discrepancy maps between the high resolution radar input and necessarily somewhat blurry Piomass output which (10) are measuring slightly different things, radar signal return and ice thickness but (11) could very well lead to a very accurate correlation function between the two which (12) would allow direct read-off of better resolution ice thickness directly off the radar (13) and a daily ice thickness report for seasonal prediction purposes without waiting for or relying on the eventual lower resolution Piomas image which has served its purpose once it has provided radar imagery thickness calibration. Too bad we're just realizing this in April 2013 because the shelf life of ice volume prediction-ology, like all that we do, is going to be very limited with the multi-year ice tanking so fast. Mid-September is only five months off, so predictions don't add much value to the wait-and-see option if this is the last year (ie the heat budget disruption is about maxed out). On the bright side, the main significance of what we are doing overall with sea ice may come into play (decades or centuries later) in monitoring early stages of sea ice recovery. So maybe burn a CD of the blog periodically in the hope that the dominant life form at that time (cockroaches?) will be technologically savvy and find value in what we did.

R Gates notes "huge water temperature anomalies in the Arctic at large river discharges, which is understandable, as the entire permafrost understructure of the .." Important point. I was just puzzling the other day (to no effect) about what drives much earlier breakup on both sides of Banks Island, seems like the mouth of the MacKenzie mouth is more 'logical' (see that earlier animation). The MacKenzie drains one-fifth of the total land area of Canada -- and 75% of the basin is underlain by permafrost so we can expect some major effects on the Arctic Ocean as that melts down. In addition to the excellent link you provided, AK Betts et al 2003 write that "streamflow at the mouth of the Mackenzie remains very low until May and peaks in June" though their graphic shows some flow beginning April 1st. I'm not seeing any attributable MacKenzie outflow effect on Modis or AVHRR as of Apr 13, the weather has finally cleared a little.

Case in point: NSIDC (to their credit) forthrightly retracted their Greenland melt day story on 18 Mar 13. Read the article and be astonished at the utter lack of ground-truthing. In 2013? Melt estimates are based on little more than someone sitting in their cubicle counting pixels from a single channel of the Jaxa radar, the 37H. I had long suspected this from the visual match of their map to the imagery. Meanwhile, the really horrific news got lost in the shuffle -- melt water in the snow from last summer never refroze. What happens with that when the summer comes on? http://nsidc.org/greenland-today/