« Oh, and BTW, the Passages are open | Main | IJIS SIE: 5 million km2 mark passed »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Bfraser

Phil and Lucia,

Under "Recent Comments", if you click on the name of the commenter, it takes you to the corresponding comment.

(If you click on the title of the article, it takes you to the main article with the first page of comments loaded -- the same as if you clicked on the title under "Recent Posts")

Hope this helps,

bill

Lucia (The Blackboard)

Bfraser--Thanks!

Werther

Someone put on a great blink comparison 2007-2011 yesterday. It illustrated clearly both years are colliding on extent. I still think a new record isn’t likely anymore. But conditions are extreme nowadays. Paul K is right in assuming ‘out of control’ response of the ice. It’s probably the spread of all accumulated rubble into very high SST’s , combined with eddies bringing up warmer, saline water right under the central pack. Even under dropping 2m temps the loss of extent remains high.
For 30 august, CT comparison map +30% shows 2011 504K under 2010 and 42K under 2007. The extent difference of 181K IJIS is just about the 15-30% concentration ice. And that counts for a mere 140 km³ of thin, spread out ice. The Healy pics confirm more and more that the actual volume in the remaining high concentration pack isn’t much above 3000 km³.
From a whale’s point of view, the Arctic Ocean is almost seasonally ice free. That is, the whale’s senses might be more focused on the volume (and the polynia’s for breathing), whereas we stick to 2D data.
We’re not seeing right, a bit like Don Quichote, you know, the Spanish tragic hero, who took windmills to be malicious giants.
The comparison is flawed, though, because we take the malicious giant (meltdown) to be profitable windmills (the golden opportunity to get to Arctic resources).
Some took Cervantes’ hero to be mad. What does it make us look like?

Ned Ward

Ned perhaps you can clarify for me. From the documentation I read MASIE and IJIS have two entirely different data sources. Or more precisely MASIE uses more data sources than IJIS to create it's daily product. What am I missing? Thanks in advance your comments have always been most helpful.

Thank you for the compliment. There are probably other people who can summarize this better than me, but I'll do my best.

Uni Bremen distributes a sea ice extent product but does not currently provide a simple online table of daily SIE values. Their analysis is based on the AMSR-E passive microwave radiometer, one of the instruments on Aqua. They map SIE on a 6.25 km grid.

IJIS-JAXA also has a SIE product derived from AMSR-E (it's a joint US/Japanese project, IIRC) mapped at 12.5 km and posted on the IJIS website. This is what most people refer to because the processed results are readily available.

The advantage of these products is that AMSR-E is a very good sensor, but unfortunately the data only go back to the launch of Aqua in mid-2002.

Meanwhile, NSIDC distributes several different SIE data products. The "NSIDC" graphs that people usually look at on the web are from their Sea Ice Index product, which is based solely on passive microwave radiometer data from SSMI, which has been carried by a bunch of DMSP satellites for many years. These are coarser resolution and less reliable than AMSR-E, but they give a much longer historical record. Because of the worse signal-to-noise ratio, they use a 5-day averaging process in preparing the data for the graphs on the web. They post the monthly data -- this is what SEARCH is trying to predict -- but do not post the daily numbers, not even in the five-day-averaged format.

There actually seem to be two different SSMI derived ice extent products at NSIDC -- one based on a local algorithm ("bootstrap") and one that they inherited from researchers at Goddard (the "NASA Team" data set). Historical daily data for these can be found from the 1970s through 2004 or 2007 on the NSIDC FTP site. At the moment I can't recall which one, if either, is equivalent to the data now shown as "Sea Ice Index" on their website.

The MASIE product is a completely different SIE product from the other ones distributed at NSIDC. It's based on the National Ice Center's Interactive Multisensor Snow and Ice Mapping System (IMS). Instead of using just SSMI microwave radiometer data, IMS (and thus MASIE) combine data from a whole bunch of different types of sensors (optical sensors on GOES and AVHRR, microwave radiometers, and I don't know what else).

This is all terribly confusing. If nobody else has a better explanation, maybe over the weekend I'll try to put together a systematic compare-and-contrast of all the various SIE data providers.

Steve Bloom

Phil, in Windoze (and I'm sure there's an equivalent Mac command) ctrl-end brings you instantly to the end of a long page, so even where there's no shortcut available you an at least avoid the scrolling. Ctrl-home brings you to the top, BTW.

Bob Wallace

Michael -

Jacobson is not saying that GHGs have no warming effect, he is saying that we are paying too little attention to the role of soot. His work finds that 1.7C of the 2.5C warming in the Arctic is due to soot.

"Plus, why did the Arctic cool in the 1940s-50s when increased aerosols caused global temperatures to stagnate (generally following NH temperatures, note that SH temperatures didn't cool or stagnate...."

My understanding is that following WWII the northern hemisphere engaged in great industrial activity, fueled mostly by 'dirty' coal. We released large amounts of SO2 which worked as a partial sun block and created a period which is referred to as the Global Dimming. Once we (largely) cleaned up smokestack emissions global temperatures resumed their climb.

(Remember, putting large amounts of SO2 into the atmosphere is a commonly suggested method of geo-engineering. And a return to the days of acid rain.)

"Sea ice also doesn't last long (especially now) so I'd think that soot wouldn't be able to build up much (except for Greenland)...."

Jacobson isn't saying that soot on top of the ice is the large contributor, it's soot in the air...

"Black carbon has an especially potent warming effect over the Arctic. When black carbon is present in the air over snow or ice, sunlight can hit the black carbon on its way toward Earth and also hit it as light reflects off the ice and heads back toward space."

Take away the soot and first year ice might hang on longer into the season and slow the melting process.

Here's a much better write up of Jacobson's paper...

http://news.stanford.edu/news/2010/july/soot-emissions-ice-072810.html

r w Langford

Interesting conclusion. I must be behind the times as I thought soot in the atmosphere was responsible for about one degree C cooling rather than warming. It was known as the dimming effect of solar radiation. Soot on snow increases adsorption of heat though. Any experts out there?

r w Langford

phil, Command, up arrow for top of page or command down arrow for bottom on a Mac.

Bob Wallace

I think it was the sulfur dioxide (SO2) that gave us the cooling, not the soot.

Jacobson's point is that soot suspended in the atmosphere intercepts sunlight as it comes in and turns it to heat and also intercepts reflected light which would otherwise escape back into space, producing more heat. It wipes out albedo effect from the ice and snow which could explain why the Arctic is heating faster than the rest of the globe.

Bob Wallace

" I still think a new record isn’t likely anymore."

If you compare the recent slope of 2007 and 2011 on the AMSR-E graph you can see that by now 2007 had leveled out some in relation to the melt rate of 2011.

http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm

Assuming that conditions which cause a greater rate of melting this year, at this point in time, aren't going to suddenly change then it seems too early to write off a new record.

Current melt is not due to some unusual storm working on the ice, but due to residual air and water temperature working on smaller-than-2007 volumes. I'd guess that even if the current slope begins to flatten in the next day or two it will be a more gradual deceleration which could easily carry things past where 2007 bottomed.

We've got three more weeks, or more, of melting a new type of ice. I'm making no guesses....

Pete Dunkelberg

Bob Wallace, thanks a lot for your remarks and link on Jacobson!

Adam Schwartz

On how to easily get where you left off on the comments section. What I have done is right click on the date/time of the last comment I have read and then select Bookmark This Link. Then when I come back later I just select the Bookmark and it takes me to where I left over.

Kevin O'Neill

On August 22 Paul K. said:

Ned Ward, your reasoning using statistical trend extrapolation walks right into a major trap when the underlying system changes. When the system undergoes a major change, prior data collected on the previous system isn't as useful or accurate as you believe in forecasting the response (output) of the new system. This is a common failure mode for statisticians who don't understand the system. To quote Dr. W. Edwards Deming, who pioneered the use of statistics for improving manufacturing systems, "There is no substitute for knowledge!" and he emphasized that one should study and understand, and model, the system generating the output. He was scornful of statisticians who simply extrapolated previous system output, and didn't observe system changes right in front of their eyes.

I was unable to expand upon what Paul said at the time, but I think it's worth revisiting. Like Ned I have an Excel sheet that uses previous year's data to 'predict' this year's SIE. Unlike Ned, I was not very happy with it.

The first thing I noticed is that the final result is dependent on your choice of start date. I tried every start date from March 1st to May 31st. The average error (based on IJIS data for August 31) is 12.51%. Best case is 6.00% and worst case 17.12%. Standard deviation is 3%. Using a start date of July 1st -- nearly 2/3 of the way through the arctic sun season -- yields an error of 5.99%.

I tried numerous mathematical tricks with the numbers, but I didn't find any significant improvement until I thought about the system. What I realized is that 2007 signified a system change. Based on that belief I threw out perfectly good data; I based my Excel only on 2007 to 2010 IJIS numbers. The improvement is significant.

Using every start date from March 1st to May 31st the average error is 2.05%. Best case is -2.23% and worst case 5.36%. Standard deviation is 1.96%. Using a start date of July 1st yields an error of 1.19%.

At first glance this seems impressive, but as Paul knows, it's not. It really has very little predictive power -- unless the system doesn't change. Do any of us believe the arctic is going to be status quo going forward? No.

The real goal is to predict the NEXT system change. The years of data accumulated are only valuable if they lead to a better understanding of the system - they are not an end in themselves. The better the system is understood the easier it is to predict its behavior.

Kevin O'Neill

The perils of cut & paste; obviously if the average error is 2.05% the 'Best Case' can't be -2.23%. -2.23% is the Excel MIN value for the series. I was using that to show the 2nd method has errors both positive and negative (removing the systemic error apparent in the first method where the MIN value was +6.%).

The actual 'Best Case' for is 0.07% - which showed up using several different start dates, the earliest of which is 3/14.

Kevin O'Neill

The perils of cut & paste; obviously if the average error is 2.05% the 'Best Case' can't be -2.23%. -2.23% is the Excel MIN value for the series. I was using that to show the 2nd method has errors both positive and negative (removing the systemic error apparent in the first method where the MIN value was +6.%).

The actual 'Best Case' for is 0.07% - which showed up using several different start dates, the earliest of which is 3/14.

Bob Wallace

The next significant system change is likely to occur with the first summer melt-out.

After that we will be looking at the melt rate of an Arctic which has no multi-year ice. If the 'first year ice' Central Arctic starts melting at the same rate as other regions then it's going to be unlike any conditions previously observed by man.

A September first time total melt-out could easily be followed by routine July melt-outs.

Paul Klemencic

Kevin O'Neill, you got idea. I couldn't have said it any better. The End Zone post last by Neven last week shows some really interesting patterns that point toward a different system yet again. The 2007 pattern of melt driven from the Chukchi and E. Siberian repeated a bit in 2008, but in 2009 the pattern of melt attack on the central Arctic basin changed. And 2010 shows very much the same pattern as 2011, only this year melted out the Kara and Laptev earlier.

This new pattern could be repeated over and over in coming years, and may be repeatable. The 2007 pattern needs a persistent Arctic Dipole.

Just one more indicator of systemic changes afoot (afloat?) in the Arctic.

Stevemosher.wordpress.com

Thanks Ned, That was most helpful. When I finish the R package I'm working on i'll try to do one for Ice data, i'll let you guys know when its done.

Paul Klemencic

Final Bremen map comments:
Yesterday was a big melt day, with some really big pullbacks in the Chukchi and Beaufort regions, a shift and pullback in the E. Siberian, and significant receding ice pack edge along the east side.

But today the pack moved the other way, with a lot of ice spreading toward the Chukchi, and the east side doing a side step up toward the Laptev. The spread in the Chukchi opened up some weak patches of ice with some open water several places along 80N, and above 80N along the 165W line of longitude. These patches seemed to be located where earlier patches of green (less than 70% concentration ice) showed up earlier. Between 150 w and 165W, the ice pack moved and spread out toward the Chukchi sea, leaving a lot of weak spots opening up behind the front. This area could develop into an area of strong open sea incursion into the central pack. Although not much time left, this area has been showing weakness for some time.

The ice also was pushed in the Canadian Archipelago, so there was some extent gained there today as well. The E. Siberian moved in a different direction today as well, with some spreading and some green areas showing up.

Only on the east side did the ice edge pull back significantly, with the edge of the pack now about 250 km north of Svalbard.

All in all a neutral, or even extent gain day. Yesterday's loss should show up as really big, when we get the final number, it will probably be over 80K.

Phil263

A very small extent drop of 16k reported by JAXA for Sept 1.We are now at 4,727,813 about 20k above the 2008 minimum.

Bob Wallace

But a ~20k gain on 2007 as 9/1/07 returned a 3.9k extent gain. 9/2/07 gained 6.1k so small melts for a couple days get helped with small gains in 2007.

Chris Biscan

Jaxa had a small melt day but Bremen had a large drop. Which is the 2nd in Germany but also on a later reporting scale. So we may see jaxa have a bigger like 35-40km2 drop tomorrow.

The models again in day 2-3 show favorable conditions for a rapid recede in the laptev between 135E and 105E.

The Beaufort goes under compacting winds as well. We may see a couple 50-65km2 drops taking us to 4,575,000 by the 5th.

After that things get colder enough for maybe some freeze in the ESB.


Gonna check the Euro now.

Phil

Thanks BFraser and others on how to get to end comment. Bfraser solution of clicking on the NAME of the last commenter not the article, does exactly what I've been looking for, Ta :)

Ned Ward

Paul wrote: Ned, since Julianne Strove at the NSIDC commented at WUWT that the NSIDC sea ice extent was 4.66 million on August 30, we should have a very good confirmation of the MASIE delay in posting SIE on Monday (but since that is a holiday in USA) likely on Tuesday.

The number Julienne was quoting (4.66) is from the NSIDC's regular "Sea Ice Index" product, not from MASIE. They are two different data sets, from different sensors and different methods.

Kevin McKinney

"(Remember, putting large amounts of SO2 into the atmosphere is a commonly suggested method of geo-engineering. And a return to the days of acid rain.)"

To be picky, supposedly not, as the geo-engineering version relies upon injection into the stratosphere where it doesn't rain out.

(Of course, it does eventually clear, but I think--rather vaguely, I'm afraid--that various bits of atmospheric chemistry transforming the SO2 come into play, rather than a straight descent into the troposphere followed by cloud nucleation and acid rain to kill our forests. Though I still feel leery of the whole idea.)

BTW, thanks to various folks for the tips on navigating the pages more conveniently. I'd have never tried clicking on the commenter's name, as I'm used to the Wordpress format, where that takes you offsite to the user's page.

Paul Klemencic

Ned, thanks for the review of the different ice extent measurement systems. I didn't know all the differences between the systems.

Here is the part about MASIE (pronounce like the flower 'daisy'):
The MASIE product is a completely different SIE product from the other ones distributed at NSIDC. It's based on the National Ice Center's Interactive Multisensor Snow and Ice Mapping System (IMS). Instead of using just SSMI microwave radiometer data, IMS (and thus MASIE) combine data from a whole bunch of different types of sensors (optical sensors on GOES and AVHRR, microwave radiometers, and I don't know what else).

This is all terribly confusing.

I agree its confusing; in fact very confusing. But you left out some pretty important information about MASIE, in particular this passage from their "About MASIE" page:

...However, it relies more on visible imagery than on passive microwave data, so the ice edge position will generally be more accurate than that of the Sea Ice Index. The input is the daily 4-km sea ice component of the National Ice Center (NIC) Interactive Multisensor Snow and Ice Mapping System (IMS) product ... (goes on about links to sources of data)

So of the products you listed, MASIE is the one attempting the highest resolution (4 km), and on a daily basis, with the information available in each region. For matching daily ice maps, this product should work better than the other products.

But since it does look at the passive microwave data, the result should agree well, unless the visible imagery contradict the microwave data.

And because my neophyte and rather embryonic ice system model attempts to classify the ice pack by region; and within each region identify the vulnerable ice, estimating possibly mortality rate; then a daily report by region is better.

The only problem with MASIE, is that it seems to take some time to compile the information, and the daily reported data seems to match the week earlier ice maps. But the system was designed to produce accurate daily maps and information; it seems that they are short of the resources to keep it updated more quickly, and fix the glitches in their reporting system.

Stevemosher.wordpress.com

Paul

look back on this thread 6 days ago or check my posting and you can see the places where I talk about MASIE being an entirely different system. Not a time shifted shifted version of some other measure, not 10 days late or 6 days late, but entirely different. with a human square in the middle of the loop.

You seem to think that MASIE must be time delayed. The documentation, if you read further, explains that it is a daily product.
The difference in reporting would be to be an accuracy issue for the other services and not a time delay for MASIE

Also, MASIE includes more than satillite data it includes ship observations and other real time input. In terms of resolution, and diversity of sources it would appear to be the best.

You know that micro wave underestimates concentration and isnt especially accurate at the margin of the ice since the algorithms depend upon the difference in brightness temp between open ocean and ice. That underestimation can be corrected with visible band. If A micro wave product ( IJIS) shows more loss that a visisible + microwave product ( MASIE) the likely explanation is the microwave product is under estimating the concentration ( especially at the edge) and the product that has a micro wave + visible is doing a better job of estimating concentration at the edge.

Lucia (The Blackboard)

Mosher--
I for one would need to see multiple years worth of MASIE vs. IJIS data before I would consider Paul K's theory of the lag to have been tested. I'm not sure how many years-- but likely 5.

If I saw one year worth of data suggesting a lag, I might start paying some attention to this.

For now, my general sense is: If method A shows a greater loss than B over the past week or month, it's plausible that method B will show the larger loss in the upcoming week or month.

This is because all methods have some errors and over time, if the errors are not biased, the errors will tend to correct themselves. Unless I see years worth of comparison properly processed, I will interpret behavior that looks like one method is "converging" toward the other method to tell us precisely nothing about Paul's theory of the lag. As far as I can see, if a convergence occurs when the labor day weekend is past, this will neither contradict nor confirm Paul's theory.

Espen

When the cloud cover over North and North East Greenland gets out of the way you will not recognize the area from only a week ago.

Regards Espen

Paul Klemencic

Mosher and Lucia, Because MASIE provides regional data on ice extent, its pretty easy to see which day's data is being posted. I am not just looking at the total extent. The losses are in the Beaufort and Canadian region one day, then in the Laptev the next, etc...

Its not hard to quickly see that the data is date stamped six days late on the spreadsheet. The data fit the Bremen ice map from six days ago like a glove.

Paul Klemencic

For example, here is the regional map for the Chukchi. See the big drop marked on the map on August 28?

Well the maps show the big pullback in ice in the Chukchi occurred on the 22nd, six days earlier.

Same with the Beaufort map, for the same day.

I have been following the maps and comparing with MASIE now since August 20, and at first couldn't understand the difference, until I realized the dates on the spreadsheet are off by six days.

Paul Klemencic

Its also really easy to spot problems with the data, if you keep in mind the areas in the different quadrangles.

The 15 degree quadrangles between 75N and 80N contain 203k sq km.

The 15 degree quadrangles between 80N and 85N contain 122k sq km.

So look at the Chukchi region, and see that on August 21, the day before the flash melt event, in order to have 400k in that region both quadrangles between 150N and 180N must be almost filled with ice. But they weren't, not even close. There actually was only 248k there, and a day later, only 180k. You can read these off regional extents off the MASIE spreadsheet.

Losing 68k in one region overnight is a big deal...

Espen

Hi Paul

Are you sure about those figures since I have roughly calculated the 75-80 = +/- 185.000km2 and the 80 - 85 = +/- 92.000km2
Regards Espen

Lucia (The Blackboard)

Paul--

I commented on your theory involving JAXA. which should have been apparent because I used the word "JAXA". A reply that does not discuss JAXA is hardly relevant. I think you should note that Ned also used the word "JAXA" when discussing your "time lag" theory in his comment "Posted by: Ned Ward | September 01, 2011 at 21:14"

So your long discussion of a time lag between two series neither of which is JAXA would appear to be rather irrelevant to the issue of the time lag involving JAXA.

I'll admit these discussions of your time lag theories are confusing because have more than one theory about more than one time lag. One theory seems to involve JAXA relative to others; one MASIE. ( Are there any other groups lagging? It's hard to keep track.)

But no matter how "easy" you think it is to show that one group lags another, in my opinion testing any of your various theories that "A" lags "B" requires someone to either

(1) post the time series for both A and B for over multiple years so that others can see the time series you are working with and then show one series lags another by a specified number of days over a span of multiple years or

(2) read the description of how the data are assembled provided by the reporting agency and demonstrates that method will have some lag. (So, for example JAXA is a two day average, so, in that sense it must have a short lag-- but two day averaging doesn't cause a week lag.)

As far as I am aware, no one -- including you-- has tested your various theories about lags doing (1) or (2). If you are correct and it is easy to do this, since the matter seems important to you, I think you should go ahead, do it and then present the time series and show us there is a robust lag-- or point to the comment where you showed it.

Otherwise, while your speculations may pan out, as far as I can tell, they are utterly untested.

Paul Klemencic

Espen: I could be wrong, but I double checked the calcs twice. I did use approximations using areas in a circle.

Lets use the spherical calculations first, then correct for the ellipsoid shape of the Earth. Circumference = pi D, and with 36 ten-degree arc in 360 degrees, we can check the length between 80N and 90N. Diameter of the earth is 12756.32, so a ten-degree change in latitude is 1113 km for sphere. But since the earth is ellipsoid, the actual is 1116 km near the North Pole.

The first 5 degrees between 90N and 85N then has area= pi r^2 with r=1116/2 = 558 which means the area above 85N is 978k sq km, if we approximate the area using a circle of that diameter. In actuality the area it will be somewhat higher.

Likewise the area above 80N is 3913k sq km.
The difference is 2935k sq km divided by 24 15-degree quadrangles = 122k sq km in the quads at this latitude. Checks.

The area above 75N is calculated using a radius of 1670 km giving 8762 sq km, and subtracting the area above 80N of 3913k sq km, leaves 4849k sq km divided by 24 gives 202k sq km in the quads at this latitude. Checks.

Of course, I really should integrate to determine the spherical surface to get a really accurate result, but I think these numbers are close.

Please note the area above 80N is 3913k sq km, and adjusting for intruding land areas, and adding lower latitude ice, this is roughly consistent with ice pack area.

Christoffer Ladstein

Congratulations, Neven & All; this blog now got 50 Followers!

Odds for doubling that number once a year?!

If so, we certainly got a huge group to be respected and nodded approvingly to in... 15 years.... (1.6 mill. followers, HA!, would be something would'nt it!)

Paul Klemencic

Lucia, I was replying to Mosher and you in comments made just before my comment. Your comments were dated September 2 at 16:27 and 17:55 wherein both of you were discussing MASIE, and problems I raised regarding MASIE data being dated incorrectly. Mosher replied to me in the comment, and primarily discussed MASIE.

Regarding IJIS, early on it appeared that IJIS is also slower to respond to changes in the ice pack. I did think there was a delay. In order to analyze that, I tried to construct a set of numbers that would result in the supposedly two day averages reported by IJIS. Here is the relevant comment, along with the conclusions I could draw.

Clearly the JAXA series of extent numbers reported cannot be simple averages of the two days of extent measurements.

I repeated the same procedure for July 29 that reported a drop of 23.3k and July 30 that reported a drop of 23.0k, and got the same results.

JAXA-IJIS is using some kind of algorithm to dampen large extent loss days, and 'augment' low extent loss days. Use of either an underlying trend, or longer term average in the algorithm, or some sort accumulated over/under account seems likely. I like the last idea; sort of a "slush fund" for ice extent reporting.

In any case, I was unable to construct a rational set of 1d extent numbers, that would calculate to the set of rolling 2d averages reported by JAXA-IJIS. I really don't know what those reported numbers are, and how to relate them to observations of the ice pack.

Neven

Odds for doubling that number once a year?!

Things would really have to go nuts in the Arctic for that to happen. :-)

Alan Clark

I have been lurking here for some time, but I have just come across this:

http://www.arctic.io/2011/9/microwave-versus-radar-satellite-images/

I don't know how much bearing this has on the discussion.

Bob Wallace

Unless someone can find a reason why the 'death spiral' projection is flawed things are going nuts in the Arctic.

The last couple of months data points have caused the calculated curves to become steeper and the same is going to happen with this August's data. The bottom is falling out when one looks at volume.

(Might we get the death spiral chart included on the graphs page?)

There are two major inputs during the melt season, the heat input and the amount of ice available to be melted. All this other stuff, currents and wind patterns, just makes the melt rate variable over time. We're seeing volume being reduced year after year and we're seeing no reason that should reverse aside from small year to year temporary recoveries.

And at the same time we're seeing atmospheric temperatures increasing.

I'd say we're pretty far down the nuts path....

Lucia (The Blackboard)

Paul--
The comment you reposted -- and which I had previously read--is precisely the type of discussion where you describe a sort of fiddling with numbers that does not show there is any sort of lag, smoothing, damping, or averaging beyond what JAXA already tells us they do. Not only does what you did not show it, the amount of data you describe looking at is inadequate to even beginning to test the notion.

As far as I can tell nothing about what you describe yourself as having done supports your claim that "JAXA-IJIS is using some kind of algorithm to dampen large extent loss days, and 'augment' low extent loss days. Use of either an underlying trend, or longer term average in the algorithm, or some sort accumulated over/under account seems likely. I like the last idea; sort of a "slush fund" for ice extent reporting".

If you want to test your theories about some sort of systematic lag between method A and method B, or some sort of, averaging, damping or any other complicated thing you speculate is going on, you are going to have to long time series of data -- multi-year--tests for lags over a long time. Otherwise, you haven't tested for these things.

If you can find a comment where you describe doing either (1) or (2) in my comment above, please link it because I haven't seen it.

Stevemosher.wordpress.com

Thanks Alan, i'll add that to my reading list. There are two ways to look at data. look at the results and try to intuit or reverse engineer the process from that, or start with the source data. Personally, i'm a source data kinda guy. The other thing that was bothering me was the method of determining ice from brightness temperature. That relies on being able to determine a brightness temp for the ocean. Fine and dandy if there isnt a bunch of little chucks of ice.. a different animal if there is. Plus the susceptibility to storms that microwave suffers from.

Stevemosher.wordpress.com

Paul.

You and I have a different standard of proof and different objectives. Personally, Im not satisfied if the regional maps fit like a glove. I am satisfied when I can re construct the data from SOURCE. That is, take the source data, apply the source algorithms and confirm why it fits like a glove. Assume it does fit like a glove. Assume that as a given. you have a hypothesis that this is related to delays. good hypothesis. probably correct. How do I rule out other explanations? It basically comes down to a reverse engineering perspective which is fine and dandy and a first principles re built from scratch approach. I prefer the latter because I like doing it. So even if the delay hypothesis satisfies you, I'm looking to do different things. namely, build tools that allow people to re build things from source. two entirely different problems. So, go in peace. No argument here just a different problem and different approach.

Stevemosher.wordpress.com

Mosher--
I for one would need to see multiple years worth of MASIE vs. IJIS data before I would consider Paul K's theory of the lag to have been tested. I'm not sure how many years-- but likely 5"

##
ha Lucia, you know my standard. I wanna see the code. There were a bunch of times I thought I had the CRU algorithms nailed through reverse engineering.. except for tiny tiny little bits in the early century that were off. looked like a rounding error. That was my theory. Till i wrote them and they explained the very subtle difference between my reverse engineering and their actual algorithm..( they dont actually publish this step to my knowledge ).

Some of the folks who work on this stuff are 'R' buddies. Just a matter of time, maybe by AGU, there is going to be a special session on R. I'm waiting on one guy to come back from the arctic and will see if I can interest him in the project. I need a bigger machine. Satillite raw data is a PITA.

Lucia (The Blackboard)

Mosher--
Showing the code would qualify as method (2) above (September 02, 2011 at 21:04). I need method (1) or (2) to believe there is some sort of lag.

William Crump

Neven:

Thanks for letting me post my pet "Arctic Basin" position on your web site and sorry if I got going a little to far with my response.

The photos from the ship are impressive evidence of the change in the ice cap, just wish we had better thickness data for first year ice in the Arctic Basin.

FrankD:

Cryosphere Today area chart for Arctic Basin shows increasing amounts of ice in the arctic Basin.

http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/recent365.anom.region.1.html

We are well above 2007 level on this graph.

The MASIE extent chart show the 2007 plunge continuing while 2011 may be leveling off after touching 2.97 million km2.

ftp://sidads.colorado.edu/DATASETS/NOAA/G02186/plots/r11_Central_Arctic_ts.png

Curious, MASIE seems to lag Cryosphere Today - think I saw something in this blog about that.

Let's see, 2011 is .16 million km2 below 2010 and .10 million above 2007. 2011 still has 2.98 million left to go. At that rate it would be some time before the Central arctic Basin is "ice free". I still think it will go faster than these three data points indicate (which are not a sufficient data set to reach any valid conclusion, but it will not go as fast as the ice volume trend lines.

The Arctic will not be "ice free" until the Arctic Basin is "ice free".

If you can point me to a good data set that you like for both area and extent for the Arctic Basin, I will try to draw some quadratic trend lines.

Oh well, we will see what next week brings.

Paul Klemencic

Lucia and Mosher, you are both ignoring my finding. I cannot find a set of measurements that generate the rolling 2-day averages generated by IJIS, without the measurements "blowing up" on either side of the test period. In short, it is mathematically and physically impossible that the IJIS reported data are simply 2-day averages.

You can confirm this easily, if you tried. Pick three consecutive days of JAXA data, then run a model with all possible measured extents for the middle day, and use it, along with published two-day averages, to calculate the extents on either side of the three day period. You will find the published JAXA data result in a repeating pattern of very high melt days one day, with a very low or negative day the next.

The test is fast and easy to run. Simply set up a spreadsheet with the daily JAXA - IJIS reported numbers on it. Setup a column to calculate the daily extent measurement from the initial guess and the reported JAXA extent 2-day averages. Then change the input to try to find an initial guess that will work, without getting the nonsense one day extents that emerge.

You won't be able to find an initial guess that works. This test shows the JAXA can't be two day averages; its physically impossible. Ignoring this, is a convenient way to pretend that the data is simply an accurate measurement of 2-day averages, when I have shown that the data are not.

And we know from observations that the JAXA extent reports don't match changes in the ice maps very well.

JAXA- IJIS is not very useful for following and tracking summer ice melt trends and the ice systems, because we don't know what the reported numbers mean. If you know what the reported numbers mean, please tell us.

Lucia (The Blackboard)

PaulK

Lucia and Mosher, you are both ignoring my finding. I cannot find a set of measurements that generate the rolling 2-day averages generated by IJIS, without the measurements "blowing up" on either side of the test period. In short, it is mathematically and physically impossible that the IJIS reported data are simply 2-day averages.

I am not ignoring this. I am just not commenting on it.
You can confirm this easily, if you tried.

Sure. So?
The test is fast and easy to run.

It's not a test. It's a fiddle.

Paul Klemencic

Lucia, You say that you must have the source data. Great, please show us the source data you have for JAXA.

Hmmm, can you dance fast enough to that fiddle?

Paul Klemencic

At least I tried to calculate the original data for JAXA; I didn't take it completely without testing its veracity.

Here is what I know about the JAXA data set.
1. The reported data is not two day averaged measurements.
2. The data is inconsistent with day to day changes observed on the ice maps.
3. The data is inconsistent with the Bremen ice extent reports.
4. The data is inconsistent with the NSIDC ice extent report; the five day average legacy system report.
5. The data doesn't show enough low melt or positive days in the late summer to be consistent with the accuracy of the measurement system.
6. We don't know what period of time each daily reported number covers.
7. The system hasn't been around as long as the NSIDC legacy system.
8. The measurement system doesn't use as small a grid size as the MASIE system.

Jon Torrance

Will Crump,

When you say "We are well above 2007 level on this graph.", does that mean you have an archived copy of a Cryosphere Today arctic basin area graph covering the 2007 melt season? Or is that freely available somewhere on the CT web site and I just haven't noticed? Or out there in the blogosphere somewhere, in which case a link would be nice. Cheers.

Lucia (The Blackboard)

Paul

You say that you must have the source data.

No I didn't.

Lucia (The Blackboard)

Paul--
I realize you've convinced yourself of all of those things.

7. The system hasn't been around as long as the NSIDC legacy system. 8. The measurement system doesn't use as small a grid size as the MASIE system.
These are hardly news.
Anu

@Paul Klemencic | September 03, 2011 at 02:24
You can confirm this easily, if you tried. Pick three consecutive days of JAXA data, then run a model with all possible measured extents for the middle day, and use it, along with published two-day averages, to calculate the extents on either side of the three day period. You will find the published JAXA data result in a repeating pattern of very high melt days one day, with a very low or negative day the next.
==============================================

I think you're confused about the JAXA claim of "However, we adopt the average of two days to achieve rapid data release."

Here is some actual IARC-JAXA data for sea ice extent:
08,28,2011,4964063
08,29,2011,4896563
08,30,2011,4796875

They claim each released datapoint is an average of two days measured sea ice extent.
Here's possible measured extents that would support 2 day averaging:
e27n = 4977109.5
e28n = 4951016.5
e29n = 4842109.5
e30n = 4751640.5

e27n is "extent, for the 27th, noon"
Average this extent with e28n - 2 days measured extents, averaged to give 4964063.
Average e28n and e29n - you get 4896563.
Average e29n and e30n - you get 4796875.

I don't see any problems with "very high melt days one day, with a very low or negative day the next".

You can think of this as measurements every noon, and the 2-day average of two noontime extents is the virtual value for the midnight between those noons - so, JAXA reports the virtual, average midnight values. JAXA also mentions 3 to 4 "preliminary extents" per day, so half a km^2 is not impossible if they are averaging 3 or 4 measurements to derive an official "noon measurement" that they then average with the next days official "noon measurement".

It seems pretty straightforward.
Perhaps I am missing some subtlety - it's rather late here...

Dan P.

Paul, I think some of the confusion over your claims is a difference in perspective. Your claim (#1) is a strong one, as is your claim that MASIE data are delayed; these are both claims that the proprietors of the data products are misrepresenting how they produce them.

Where source data is absent, most of us are inclined to give the benefit of the doubt and trust the official descriptions. It's not that I can't be convinced that these data products are not exactly what they've been represented to be. But it would require a much more systematic demonstration than what you are providing so far.

I followed your prescription for guessing the daily JAXA data and I didn't find it particularly remarkable. It is true that you wind up with a lot more century breaks up or down no matter how you fuss. It's intriguing and I'm glad you pushed me to do it. But without some careful statistical tests (and I've been thinking how best to test it), I don't see anything alarming.

Of course we'd expect the data to be noisier. I'm not convinced it's more than what you'd expect, and even if so, it's perfectly possible there are day-to-day correlations or anticorrelations in the source data's noise, without anything fishy being introduced in the averaging.

I really appreciate all these data issues you've been bringing up, but honestly I have learned far more from your fantastic frequent regional updates on ice conditions. From a purely selfish point of view I'd love to see more of these!

Dan P.

I should clarify that I was doing exactly what Anu demonstrated, which is how I understood what you (Paul) were doing as well.

Seke Rob

Re: Jon Torrance | September 03, 2011 at 05:38

He's maybe using the side-by-side image selector at CT [follow this link], three quarter down on the left. It's not very high resolution, general impression provider, but on outline it's OK.

The last set available Aug 31, comparing same day with 2007 shows a fat fringe of red/yellow/green which 2007 does not have. Bottom melt will maybe drive for a good chunk to disappear. Water temps and weather, not so much insolation will have to do that and that is no certainty, the weather.

JAXA, only a minute decline... < 20K KmSq. Hard to think it will not slip into second spot for lowest, but it's really in the statistical realm the way it's going... but lowests for year have been recorded later in September.

Artful Dodger

Hey IJIS Math Majors:

Here's a hint:

http://dilbert.com/strips/comic/2010-02-07/

Phil263

Extent graph shows a big uptick on both U Bremen and DMI (30%). Expect to see extent gain on JAXA for Sept 3. 2008 extent minmimum looks so close, yet so far!

Artful Dodger

No Phil, the just released data from Uni-Bremen and DMI is for the same time period. I know you're ahead of your time (being Down Under), but Sep 3rd isn't over yet by UTC/Greenwich time!

Phil263

Yes I know but why the uptick then? We had two small extent losses reported by Jaxa for Sept 1 and Sept 2. small losses yes but not gains, so the uptick has to show up some time?

Espen

For those interested Patrick Lockerby just released "Arctic Ice September 2011"

http://www.science20.com/chatter_box/arctic_ice_september_2011-82197

Regards Espen

Phil263

SIA back below the 3 million mark on Day 244 (Sept 1). A drop of 78k!

Lord Soth

The ice will melt or not melt, despite all the divergent opinions on small blips in the various graph.

I am also getting rather impatient for 2011 to finally take second spot away from 2008, but I am not climbing the walls (very high).

We are now entering the phase of extremely slow melt, and where sensor noize and wind patterns are the major factor in determining the next day extent reduction.

If the wind patterns over the next week favor compression, we will get a new minimun, if they don't we get a second place finish for SIE.

A new minimun of SIA is still likely, however we should be put out of our misery on this within a week, one way or another.

Sea ice volume is in the bag for #1 spot, and this obvious fact should be confirmed by PIOMAS in the next update.

Unfortunately, we will probably have to suffer thru three more weeks of daily joy and depression, until we get the final result on SIE.

Why has SIE turned into a daily obsession for many of us?. Next year, I am going to limit my SEA ice fix to one a week or maybe once a month, if I can summon the disipline.

Yes I am a recovering Iceaholic.

RunInCircles

Paul
Forgive is this is a stupid suggestion but have you tried a running average similar to an IIR filter? This would create your slush fund.

Artful Dodger

Hi Anu,

Re: your 07:32 Sep 3, 2011

Interesting work, mate but your 1-day SIE values violate grid size constraints...

Try these numbers instead for 1-day IJIS SIE:

Date: 1dayIJIS: Delta-1:
Aug27 4,977,188 - 25,938
Aug28 4,950,938 - 26,250
Aug29 4,842,188 -108,750
Aug30 4,751,563 - 90,625

Cheers,
Lodger

Artful Dodger

Hi Phil,

Thanks for your update on Sea Ice Area.

As you may have noticed, SIA as reported by Cryosphere Today seems to take some wild swings. The underlying cause is that a passive microwave sensor such as the one on AMSR-E can not see through thick cloud.

Then, it comes down to how the Investigators choose to handle gaps in the daily data. In automated systems, this comes down to the algorithm programmed to interpret the data.

If they reuse older data for the gaps, and the ice moves, the daily area is artificially inflated. There are many permutations of ice/cloud/sensor interaction that just makes the data noisy...

How do we handle this type of data? How reliable is it?

Well, daily data is a crap shoot without an insiders knowledge of the algorithms and access to the data. The best we can do is use multi-day averages, and accept the lag in reporting time.

Personally, I follow the 10-day moving average for CT SIA, which means I have pretty good confidence that I know the state of Sea Ice Area as of August 28.

I calculate the 10-day average decrease in SIA, and compare it to the average from Climatology (the other number CT posts with each day's record).

Currently, the 10-day is sitting at -16,724 km^2 / day. Climatology is about -27,000 / day. So yup, currently sea ice area loss is a little slow.

Based on the std dev for the 10-day, I'd say about +/- 6K of the current signal is noise... anywhere from -10K/day to -25K/day could be the reality right now, but we're not able to resolve it.

So chin up, pip-pip, it'll be a while before the ice melts, and you've got a farm to breath new life.

Cheers,
Lodger

Andrew Xnn

Researchers are exploring more comprehensive models for melting Arctic sea ice.

http://web.mit.edu/newsoffice/2011/arctic-ice-melt-0810.html

"After comparing IPCC models with actual data, Rampal and his collaborators concluded that the forecasts were significantly off: Arctic sea ice is thinning, on average, four times faster than the models say, and it’s drifting twice as quickly."


"Traditionally, in winter, most of the Arctic Ocean was covered with a thick sheet of ice. But today’s winter ice cover is thinner, meaning it breaks up more easily under the influence of winds and currents. It eventually looks like an “ensemble of floes,” Rampal says, instead of one large mass. In summer, natural melting due to warmer temperatures opens the door to even more breakup. (Scientists refer to these patches of floes as “pancake ice,” because the small circular pieces look like — yes — pancakes on a griddle.)

During both seasons, ice in this state is prone to escaping from the Arctic basin, most commonly through the Fram Strait, a wide swath of ocean between Greenland and the Norwegian archipelago of Svalbard. The smaller the floes, the more likely they are to be lost through the Fram Strait, where they melt on contact with warmer waters to the south.

So, several factors are connected in a positive feedback loop: Thinner ice breaks more easily; smaller chunks of ice drift more quickly; and drifting ice is more prone to export and melting at lower latitude. But Rampal also cites examples of negative feedback loops, which may counteract some of the ice loss. For example, large cracks in winter’s ice cover help create new ice, since the extremely cold air in contact with the liquid ocean promotes refreezing, which leads to a sheet with greater surface area than before."

Artful Dodger

Hi Lord Soth,

Why We Fight:

http://www.youtube.com/watch?v=vrwT2K3SwGk

Hang in there Soldier. Your relief is moving forward, and your Unit will be rotated to the rear for some much needed R&R.

Enjoy your time, and see you when you report for duty next year!

Cheers,
Lodger

Seke Rob

Don't know whom/what to have more concerns about... the WImacs or the (Ant)Arctics and the heatball effect, for it is in the long run going to destabilize society as the Aint True and the True are getting firmer and firmer into confrontation. We need more evidence, the question being how much more evidence before it knocks the most willfully ignorant from having any audience left and is thrown off the lemming cliff.

CT SIA averages through day 244 [See Chart]... even those supposed 3 recovery years did not get inside 1 sigma (yellow bars). The mentioned -78K decline increased the 2011 mean lead again over 2007. Who'd thunk that?

Ned Ward

Paul writes: Here is what I know about the JAXA data set.
1. The reported data is not two day averaged measurements.

No, no, NO. You do not "know" that. You're guessing that, because you've done some messing around with numbers and can't think of any other explanation for the results of your messing around.

You're assuming that the two-day averaging is applied at the level of the total sea ice extent.

What if the two-day averaging is applied at the level of the individual grid cell?

Let x'(i)(j) represent the single-day estimated concentration of sea ice in pixel i on date j. They might average x'(i)(j-1) and x'(i)(j) to obtain the two-day averaged x(i)(j). Then, if x(i)(j) is > 0.15, they call pixel i "sea ice" and include pixel i in the total extent for day j.

If you look just at the daily totals (which is all we see in plot.csv) there will be completely different numbers for a two-day average computed this way, versus a two-day average of two single-day totals.

I'm not (repeat, not) asserting that this is what they do. It makes sense (much more sense to me than the method you assumed). One particular hint is that, if you take the daily totals in plot.csv, and divide them by the grid cell area, it always works out to a round number of grid cells. If they were averaging two days' totals, you would expect half of the averaged totals to end in ".5"

But if I really cared and wanted to know for sure, I would either

(A) Call up someone at IJIS or JAXA and ask them; or

(B) Track down the documentation for this product and see what it says.

The one thing I would not do is post sheer speculation about somebody else's product while labeling it as something that I "know".

Janne Tuukkanen

Off topic, I know.

The editor-in-chief of journal Remote Sensing steps down because of poor peer-review process of a paper by Roy Spencer.

This shows level of responsibility and respect to scientific process, which is rarely seen among deniers, or so-called-sceptics. When have we seen public apology from misinformers about lies and half truths they're spreading?

In the debate there is still room, however tiny for so-called-skeptics in the world of science. We can discuss about the level of climate sensitivity, or negative feedback effects -- in the forums of science.

But in public debate, we should ask responsibility from editorials-in-chief of newspapers and news agencies, who by flawed fact-checking processes are giving publicity -- and gredibility -- to charlatans. We should ask journalists to do their job, and call a lie a lie.

So I myself decided, that whenever I see in my own country's news an outright lie or misinformation about the climate change, I will write, not to some public forums, but directly and privately to the editor-in-chief of that news organization, as well reasoned comment -- with sources, as I can. News organizations should be responsible about the reporting of truth, which they're so proud to boast.

I hope some of you will come along.

Paul Klemencic

For some reason, I had difficulty commenting using numbers pasted from my spreadsheet. Hopefully this comes through...

Anu, if you extend the calculations back through August, you will calculate these extent numbers for IJIS using their "2-day averages" and the four days of daily measured extent data you suggest for August 27 to August 30. Then calculate the daily measurement needed to generate the daily change in ice extent that IJIS is measuring.

Read these resulting daily losses from the bottom up, to see how the measured extents needed to produce the results for end of August.

August 13 278903
August 14 -127027
August 15 199215
August 16 -120465
August 17 237027
August 18 -76403
August 19 151717
August 20 -77343
August 21 1948433
August 22 10783
August 21 93903
August 22 -20465
August 23 13283
August 24 -20465
August 25 78591
August 26 13283

See the pattern?

Paul Klemencic

I didn't write that last comment very clearly; the point is to extend the calculation back in time from the period where you used the initial extent input (and only one day is needed), then look at the daily change in the measured extent required to produce the "2-day averages" produced by IJIS.

L. Hamilton

"I am also getting rather impatient for 2011 to finally take second spot away from 2008, but I am not climbing the walls (very high)."

Lord Soth, FWIW the 1-day minimum this year is already well below that of 2008 per UB SIE, and just slightly below 2008 per CT SIA. Also, NSIDC SIE 5-day graph has moved below the September 2008 monthly mean.

Paul Klemencic

Ned, before I go to JAXA-IJIS, I would like at least one other person to confirm that this calculation I'm using isn't wrong. But it is so simple to do, why I am getting ostracized for asking the question? My calculations show that JAXA can't be using 2-day averaging.

Anu and Dan P are testing the calculation. Give us just a bit of breathing space to check this out.

Your comment on 2-day averaging cell by cell is interesting, and could be part of the answer for the "JAXA-IJIS lag".
But the way they talk about averaging seems to compare it with 5-day averaging used by NSIDC.

Is NSIDC using averaging cell by cell?

I am simply trying to understand why the reported daily extent change from JAXA-IJIS doesn't match daily Bremen map observations most of the time.

And I would like to know why their extents are 200k+ higher than Bremen or NSIDC extents most of August. The method I was trying to apply this August needed that measurement.

Paul Klemencic

Anu, I keypunched in an extra 3 for the August 21 data point; the correct measured loss in extent calculated for August 21 is 194843 to get your extents by the end of August.

Reason for my delay: I believe that the board here won't accept any posts that have pasted in any data containing date formatted entries.

The comments just disappear never to be seen again. I lost three, so ended up manually typing the dates and numbers in.

Paul Klemencic

For Anu and Dan P... I need to go back to typing class: Here we go again:

August 13 is 278903
August 14 is -127027
August 15 is 199215
August 16 is -120465
August 17 is 237027
August 18 is -76403
August 19 is 151717
August 20 is -77343
August 21 is 194843
August 22 is 10783
August 23 is 93903
August 24 is -20465
August 25 is 78591
August 26 is 13283

Ok, double checked, these are the numbers from my sheet, that are needed to get the end of August number.

crandles

Ned,

Thank you for the averaging by cell possibility it would appear that could explain it.

Paul, you ask why you are being "ostracized for asking the question" yet your response to Ned explanation is to say it is interesting. To me this seems to compound the problem.

I tried and gave up on trying to find the one day numbers a couple of years ago and wondered if I was doing something wrong.

I suggest there is a big difference between asking a question which is entirely appropriate and claiming you "know" IJIS-JAXA aren't telling the truth. So wouldn't it have been more appropriate to apologise for calling IJIS-JAXA liars?

Paul Klemencic

crandles: I didn't use the word "liars", nor would I ever apply this word to people, especially professionals, until I had thoroughly checked out the facts. And even then I would avoid it, unless the person was truly using dishonesty for some kind of personal gain or political reasons. None of those situations apply here.

Look, its really clear that IJIS-JAXA is not using simple two-day averaging. And I began this inquiry, when back in early and mid August, day after day, we couldn't come close to guessing what the report would show, after examining the Bremen map.

My basic hypothesis for predicting melt, was to look at regional ice, assess whether it was likely to last the season, and estimate the final extent minimum by adjusting the current reported extent. In order to improve this method, it seemed logical to check the Bremen map using blink comparison day by day, and see where the ice is moving and melting out. Through careful observation, and comparing with reported extents, then perhaps I could observe how the ice was behaving and melting, including melt rates. (This is why I was interested in estimating the changing heat transfer rates as the nature of the ice pack changed.)

I really didn't come to this site, just to do trend projection, and follow the trend. I wanted to increase my understanding of the ice system, and the ice melt process. And I have learned a lot.

The biggest "aha" moment came, when the guys here showed me the MASIE data, listing regional extents day-by-day. What a terrific product, even with the glitches. The MASIE guys need funding and support to keep that product alive. I hope they get it.

The Bremen data seems really interesting, and matches the extent estimates I was making, and is pretty close to the NSIDC reports.

The IJIS-JAXA daily reports don't match observations, and their extent numbers don't match the others. Their data is the "odd man out". That doesn't mean its wrong... It actually could be more accurately representing the extent of the ice than the other reporting systems. Their data just wasn't useful to me, with what I was looking at and trying to learn about.

Espen

Paul:

Continue with what you believe in regardless of what others may think, that's my advice. Mostly because there are still many loose ends in the Arctic Ocean!

regards Espen

Stevemosher.wordpress.com

Paul

"Look, its really clear that IJIS-JAXA is not using simple two-day averaging. "

The part you are not getting is that the data that is being averaged changes. You get a daily update. Let say today was 43. Well tommorrow you get another another update that says:
Today is 44, AND oh by the way, yesterday was actually 45.


Espen

Foxe Basin:

I know I've asked this before but have never got an answer.
Is there anyone who can explain to me how the ice in Foxe Basin can survive this far into September?
Regards espen

Andrew Xnn

There is a lot of land fast ice in the Foxe Basin.

Espen

Andrew:
I know, but why?

Xandra

Espen wrote September 03, 2011 at 18:31

”Paul:
Continue with what you believe in regardless of what others may think, that's my advice.”

That’s a good advice and I agree with that.

Moreover, me thinks it has become too much unnecessary discussions at this blog about who is right and who is wrong... totally irrelevant and uninteresting I think.

Ned Ward

Paul writes: Look, its really clear that IJIS-JAXA is not using simple two-day averaging.

I just gave you a perfectly plausible explanation for how they could be doing two-day averaging. It also happens to match the way I would do it, if I were designing the system.

It would be nice if you would just stop making these very strongly worded assertions that in reality are nothing but speculation on your part.

Neven

New update is out...

SIE 2011 update 19: the fat lady is humming

Ian Allen

Espen,
The Foxe basin is so extremely cold in winter and spring, is far from ocean and doesn't have strong currents or the giant tides which occur SE of there. If and when the ice finally melts up top the sea will warm and the Foxe and the rest of the CA will become at least a bit milder, and then the ice will not last, I reckon. I don't know if there is much subsea permafrost there, but over at Vize island / Usharov island I suspect there is, that tongue of ice lasted ages.

Andrew Xnn

I'm no expert on this, but suspect land fast ice builds up in the winter in parts of the Foxe Basin due a combination of tidal currents and strong winds that pile up the ice up much thicker than in most other areas of the arctic.

Also, notice that there is currently very little ice in the Foxe basin. It has been mostly gone for over a week.

Espen

Foxe Basin:

I was just checking Google earth, and there seems to be glacier on Ile Southampton, that could the reason for ice in the area?

Regards espen

Andrew Xnn

espen; This July 2003 view of Southampton island shows no glaciers.

Foxe Basin is to the North (top) of photo.

Espen

Andrew: Yes I can see that???
Regards espen

Espen

Volgoneft-131:

What is a Russian tanker doing north of 84??

http://www.sailwx.info/shiptrack/shipposition.phtml?call=UFTA

Regards Espen

Twemoran

After Espen had drawn my attention to Foxe Basin I too was perplexed by it's lingering ice cover. Seemed as though an unusual wispy stretch of ice just would not melt.

I have't been following the area recently (distracted possibly by North East Greenland) but will check it out. Ian's idea that possible undersea melting of permafrost is responsible is interesting.

Could this thermal signature be a way to identify areas outgassing large amounts of methane?

Michael Fliss

Not sure, Espen but it is also listed as located on the Volga River close to the Rybinsk Reservoir (58 03 n 38 50 E) and probably where it should be considering the type of craft it is and its home base of Astrakhan close to the Caspian Sea.

http://www.marinetraffic.com/ais/default.aspx?zoom=9&oldmmsi=273439610&olddate=8/28/2011%2011:00:00%20AM

Paul Klemencic

I am putting this comment here, to keep it from cluttering up the new thread that Neven opened.

Ned, you suggested this possibility earlier:
What if the two-day averaging is applied at the level of the individual grid cell?

Let x'(i)(j) represent the single-day estimated concentration of sea ice in pixel i on date j. They might average x'(i)(j-1) and x'(i)(j) to obtain the two-day averaged x(i)(j). Then, if x(i)(j) is > 0.15, they call pixel i "sea ice" and include pixel i in the total extent for day j.

If you look just at the daily totals (which is all we see in plot.csv) there will be completely different numbers for a two-day average computed this way, versus a two-day average of two single-day totals.

My comment to Ned earlier:
Your comment on 2-day averaging cell by cell is interesting, and could be part of the answer for the "JAXA-IJIS lag".

Ned, I thought further about this, and constructed a mental model test of this possibility, and unfortunately.... the idea although clever, still fails to explain the problem with the IJIS reported "average extent" data. It simply allocates the change in extent (extent loss) to one of the days or the other.

Whatever pixels are counted on day 1, are not only used in day 2, but also in day 1 to compare to day 0 pixels. Likewise, the pixels on day 2 are used both on day 2 and on day 3.

So unless the pixel count on day 2 is changed from the calculation on day 2 to the calculation on day 3, the result is simply to put extent loss in one day or the other. This would "blow up" the same as simple 2-day averaging.

Even applying a correction to the previous pixel count from day 2, before using it to calculate day 3 report, doesn't explain the difference. There simply isn't enough ice on the fringe (at 15%) that would change routinely in one day, to make that idea plausible. Carrying a small correction one day back or forward, doesn't impact the outcome.

Steve Mosher: Your suggestion that JAXA could be applying a correction to the previous day's data, suffers from the same problem. It simply allocates change in extent (extent loss) between the two days.

However, these methods suggested by Ned Ward and Steve Mosher could explain the difference, if the corrections were being made to data prior to day 1 (say over the previous week), or some of the questionable data on either day 1 or day 2 is essentially "held back" for several days, until confirmed by measurements on day 3, day 4 etc.

But this would be a very different reporting system, because this pool of questionable results would be counted as ice, until confirmed. In essence, this is one way to create the "slush fund" account of slush that might be ice, so we count it as ice. Then several days later, the additions to the slush fund are removed and begin padding the daily report losses as the slush definitively disappears, and the slush is moved into the melted category.

If the slush fund was large enough (200k to 300k sq km) this would explain the IJIS reported data. And if this is true, the day to day extent losses wouldn't match observations from the Bremen maps. And it would explain why JAXA is 200k-300k higher than similar extents for the same day from Bremen.

But also if this is true, JAXA-IJIS should explain their product isn't intended to be compared to maps to see one day extent changes, and that their extent reports could be higher by 200-300k than the actual extent. JAXA has never claimed their reports could be used this way, yet many people do use the JAXA-IJIS reports this way.

Someone recently used the JAXA data to squelch the idea that large extents could be lost overnight in a regional area of the ice pack. I don't think this person realized they were using the reported data incorrectly, and that JAXA won't see or report large single day melts, if the methodology suggested by Steve Mosher further extend over longer time periods is correct.

Interestingly, MASIE is intended to be able to report one day extents that can be compared to the ice pack overall, and specifically regionally. This is part of the MASIE mission, and a goal that they really haven't fully achieved yet.

Lucia (The Blackboard)

Paul

Someone recently used the JAXA data to squelch the idea that large extents could be lost overnight in a regional area of the ice pack.

Who? Let me break that down into two questions:
1) Who has said large extent can't be lost overnight and
2) Who has the power to squelch an idea.

The magnitude of recent loss rates has been discussed at my blog and my position has been that I have no reason to believe it is not possible and that as far as I can tell based on published extent loss rates there is no reason to believe that large extent loss rates of the sort we saw in August have not happened in the past and they have happened in August.

(Discussion here: http://rankexploits.com/musings/2011/wednesday-nh-ice-update-recent-extent-losses/ )

In terms of overall loss rate, what we saw in August 2008 appear in historic GSCF and JAXA records.

On the issue of who has power to squelch an idea-- I find it difficult to believe there is someone out there who can squelch this idea or even the contrary idea.

crandles

Re "Whatever pixels are counted on day 1, are not only used in day 2, but also in day 1 to compare to day 0 pixels. Likewise, the pixels on day 2 are used both on day 2 and on day 3."

Huh???

It is the percentages that are averaged so cell 1 can have 40%, 0%, 20%, 25% on days 1 to 4 respectively. When averaging day 1 and 2 there is an average of 20% so this cell counts when averaging day 2 and 3 it doesn't count and when averaging day 3 and 4 it does count.

You cannot possibly reconstruct the distribution of percentages from the information available.

This method can properly be described as averaging over two days and if IJIS JAXA say it uses averaging over two days then it almost certainly does.

This method of averaging over two days does not allow you to backtrack and work out one day information. So you cannot claim "This would "blow up" the same as simple 2-day averaging."

michael sweet

Paul,
It seems to me that you do not know how IJIS averages their data and you have wasted a lot of posts speculating. An old friend of mine said "you can spend a week working to save an hour in the library." Why don't you try to find the paper where IJIS describes their technique, which will describe their process in enough detail that you will be able to replicate their data? You are wasting your time asserting that published scientists have not correctly described how they measured their data.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

Your Information

(Name is required. Email address will not be displayed with the comment.)