UPDATE: Dr. Walt Meier of NSIDC writes in with some information, seethe end of the article.
I’m getting weary of answering this question in comments, so here it is front page. Note the little bump right about June 1st.
Rick W asks:
Can anyone explain the upward bump in the sea ice extent that seems to occur each June? Apologies if previously covered.
Answer:
This is a seasonal adjustment to compensate for meltwater on top of the ice, which would ordinarily be viewed as “open water”. Right about now, the Arctic sea ice gets melt pools forming on the surface. If these are not compensated for, sea ice extent will read artificially low.
That being said, I wonder why we don’t see the same adjustment at NSIDC:
I don’t know the answer, but it could be in the difference between SSMI and AMSR-E satellite sensors (NSIDC uses SSMI, JAXA uses AMSR-E).
We also don’t see an adjustment at Cryosphere Today, and they also use SSMI:
Nor does NANSEN:
Click for larger images
If anyone knows why JAXA does the adjustment but the others do not, I’m all ears. My theory is that it is sensor related, but we should find out for sure. I’m swamped today, so I’ll leave this puzzle for WUWT readers to solve.
UPDATE
Dr. Walt Meir writes in with this:
Since you mentioned it on your blog, I can fill in at least some info:
You are correct. When the melt season kicks in the surface water changes
the contrast between ice and water. To more accurately measure the
area/extent, you should adjust coefficients to account for this.
This is done for SSM/I. However, because the SSM/I algorithm is
different from the AMSR-E algorithm (and other differences between the
sensors) the adjustment is different. In SSM/I, the adjustment is
smoother and thus there isn’t that “bump”.
You have to remember that AMSR-E is a research sensor and the algorithms
are still being refined. That is one reason we don’t use AMSR-E for the
long-term timeseries (though the more important reason is the
inconsistency between the two sensors and algorithms).
– Walt
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
with all the sensor issues, are any sea ice numbers considered reliable?
Maybe they just always make the algorithmic adjustment necessary to determine whether it’s a pool of water or open ocean. I personally don’t understand why you’d make an adjustment to recognize that as of June 1, then turn it off as of Dec 1, and then flip it back again. It seems to me that if there is a way of recognizing with high probability water that is pooled versus not, then you should just always adjust for that.
I could be wrong.
The real reason for the bump is the start of the coarse fishing season. All the polar bears start digging holes in the ice.
I don’t know why they do that, but the ‘lead weight’ decline of the ice extent in early June has tapered off a bit and is starting to move back ahead of 2008.
Also it seems there’s evidence that NOAA’s SST maps are somewhat biased on the warm side when compared to the bouy readings at Unisys, compare these
http://www.osdpd.noaa.gov/PSB/EPS/SST/data/anomnight.6.11.2009.gif
http://weather.unisys.com/surface/sst_anom.html
NOAA’s maps differ and show the PDO area and North Atlantic somewhat warmer than Unisys. There’s a quite a few areas where the data matches, but a few areas where NOAA’s chart shows light positive anomalies where Unisys shows light negative anomalies.
I no longer trust ANY official data. Especially when the trends are carefully and consistently adjusted to reflect the current dogma of global warming.
Southern Hemisphere winter cold records are being boken.
UK has “Unseasonal snowfall in June”
http://www.timesonline.co.uk/tol/news/weather/article6458067.ece
The explanation is given on iarc web site http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm
Here is a copy of the explanation :
“The current version of data processing produces an erroneous blip of sea-ice extent on June 1 and October 15, which is seen in the graph of sea-ice extent as a small peak on these dates. The apparent blip arises due to switching of some parameters in the processing on those dates. The parameter switching is needed because the surface of the Arctic sea ice becomes wet in summer due to the melting of ice, drastically changing the satellite-observed signatures of sea ice. We will soon improve the processing to make the graph much smoother.”
June 9 comparisn of the Arctic ice.
More ice in the Canadian Arctic this year which could make for an interesting trip through the NW Passage.
http://www.zen141854.zen.co.uk/years.jpg
Slightly off topic: the Anchorage newspaper has reported that ash from the several eruptions of Mt. Redoubt has caused local snow and ice to melt faster because when on snow and ice, the darker color absorbs more heat. It is the same effect you get from black carbon falling on Arctic sea ice.
Question: does anyone know if enough ash from Mt. Redoubt’s several eruptions got north enough to fall on Arctic sea ice, and hence, to perhaps help cause what seems to be accelerated sea ice melting in May shown in the graphics above?
If the Caitlin team had hung around up there a bit longer they could have confirmed how much was open water and how much was pooling?
It’s all the snow we have gotten in Wyoming this May and June.
Our crews from the Wyoming DOT truck it up to the Arctic and dump it there.
(But please don’t tell anyone.)
Shawn Whelan:
Where are those images from?
Thanks,
tim
PapyJako has it right, they will be trying to make it smoother as I assume the other sites do.
Of course the main point is that 2009, has now joined the rest of the years as predicted. The slow start to the melting season was nothing to post about as being a sign of anything important.
Regards
Andy
By the way, as this site posted a new comment when the NSIDC graph almost reached the 1979 to 2000 average, are you going to post a new entry showing it has now decreased to the 2007 level ?
It’s an interesting turnabout people will want to know about I am sure you will agree.
Regards
Andy
Adjustments gone wild?
Seems pretty clear to me that the melt water adjustments are resulting in an artifact if the same bump persistently appears every year at this time of year.
One can only begin imagine the effects over zealous or under zealous adjustments have in other areas of climate research.
One thing I’ve been noticing, with the exception of the June tick, and I think is kind of neat, is AMSR always seems to move first. If you see a little tick at the end of the red line in AMSR – up, down, or leveling out – give it a week or so and the others will be moving in that direction. Or is it just me that sees that?
” Shawn Whelan (10:16:28) :
June 9 comparisn of the Arctic ice.
More ice in the Canadian Arctic this year which could make for an interesting trip through the NW Passage.
http://www.zen141854.zen.co.uk/years.jpg
”
Am I seeing that chart right? Lake Michigan has a lot of broken ice still? Anyone in Indiana, Illinois, or Michigan confirm that? If so, that certainly beats the other years.
AMSR-E, used by Jaxa, is a microwave radar which is sensitive to the frequencies of H20 (water vapour, liquid water and frozen water) . There is a very large difference in the data (temperature in essence), between open ocean and sea ice. Hence, it is rather simple matter to develop sea ice extent maps and data.
The differential, however, is not as great for slushy ice and especially for slushy water-filled snow on the surface. Typically, these conditions only exist as atmospheric temperatures in the polar regions start to get to close to Zero.
So, to limit potential errors in the data, they switch algorithms at June 1 which more easily picks ups the differentials between open water in coastal areas, open water in polynias and just slushy snow-covered ice.
Its unfortunate there is difference or bump caused by this switch and perhaps there is a way to smooth the results, start the algorithm earlier or fine-tune it but this process is not developed yet. I’m not sure why they don’t use the sensitive algorithm throughout the year but it is probably less computationally intensive.
NSIDC, Nansen and the Cryosphere Today use the AHVRR sensors which operate in near-infrared and visible spectrums so, in essence, the sea ice extent just becomes a white pixel versus dark pixel counting exercise. Slushy snow-covered surface from June 1 to October 1 are not really a problem for this procedure.
The problem with AHVRR is that clouds can get in the way so one needs to average estimates over several days. AMSR-E provides more immediate results, sees through clouds, has higher resolution and is more computationally automated but suffers from being highly sensitive to water (but that is the purpose of the sensor).
We shouldn’t assume the agencies with smoother graphs are doing a better job. They might just be concealing their ‘adjustments’ with smoothing.
I’m not against researchers earning a living in a research world where deep cuts are constantly being threatened.
However, I am always (initially at least) suspicious when seasonal structure barely varies, particularly after what I found with the CO2 “data”. [For details, see Paul Vaughan (20:48:22) (May 29, 2009) at http://wattsupwiththat.com/2009/05/22/a-look-at-human-co2-emissions-vs-ocean-absorption .]
Many investigators (it seems) are blinded by the dominance of annual & diurnal cycles. This can lead to processing that precludes the possibility of properly investigating other timescales.
– –
Re: PapyJako (10:14:09)
Thank you for these important notes.
Hey, what happened to ocean.dmi.dk.? They had a good updating graph too. The whole site seems to be gone.
This “adjustment” is a specific example of a generic issue, IMHO. The “Fudge Factor”.
We have a problem that makes the sensor “guess” how much sea ice there is with a different error in summer due to water on the surface being confused for open sea. I would surmise that this was detected in the initial use of the sensor (as it was checked against facts on the ground) and a “fudge factor” was developed. But not a very elegant one. “Good enough”, but not “right”.
(I use the word “guess” because the sensor reads a property that tracks with sea ice, but is not in fact measuring “ice”; so it is a good guess, but it is not exactly measuring ice… otherwise it would not be confused in June… This is not inherently a bad thing. We measure temperatures with an IR sensor. IR tracks temperatures rather well – modulo a bit of surface emissivity effects, so IR “thermometers” are rather good at “guessing” the actual temperature.)
So somebody put into the “processing code” a Fudge Factor. It could be a simple offset ( IF [ june july august september october ] THEN reportice= sensorice * 1.15 ) or it could be a more complicated formula with an offset that grades up over time ( IF june reportice = sensorice * 1.1 ; IF july reportice = sensorice * 1.2 ; if august reportice = sensorice * 1.4 … etc)
One could even make the gradation as a smooth daily increment. It is all a matter of what code you are willing to write. What the statement PapyJako (10:14:09) quoted says is that they are going to polish their Fudge Factor:
“The current version of data processing produces an erroneous blip of sea-ice extent on June 1 and October 15, which is seen in the graph of sea-ice extent as a small peak on these dates. The apparent blip arises due to switching of some parameters in the processing on those dates.”
i.e. they have a brute force swap from something like:
reportice = sensorice * 1.0
to something like
reportice = sensorice * 1.15
And that causes a visible step function when the date rolls over.
“The parameter switching is needed because the surface of the Arctic sea ice becomes wet in summer due to the melting of ice, drastically changing the satellite-observed signatures of sea ice. We will soon improve the processing to make the graph much smoother.”
Which says, to me: We need the Fudge Factor, but didn’t do a good enough job to keep you from seeing it and asking questions, so we’re going to polish it up enough that you won’t see it.
Now that is A Good Thing, in that it will most likely make the Fudge Factor a more accurate representation of the actual ice on the ground (or sea), but it does not make the Fudge Factor go away nor does it mean that the Fudge Factor is right. It also does not mean that the Fudge Factor will properly reflect changes when, for example, it becomes abnormally cold and the puddles don’t form as fast or as large. (That is, the Fudge Factor is not based on water puddles NOW, but on an ESTIMATE from the past, and does not track changes in actual climate or weather over time…)
If I model my Fudge Factor onto a bell curve centered on August with a gradual ramp in and ramp out, instead of a square wave step function in and out, but keep the area under the curve constant; it is now smooth. But if my Fudge Factor in both cases was 1% high on total ice area under that curve, it will still be 1% high. I’ve just moved some of the error into August and out of June. And if this year is 2% water surface instead of 1%, I’m still going to Fudge Factor out only 1%.
Now, you do need this adjustment to make up for an unsolved limitation in the sensor… but NEVER confuse a Fudge Factor with reality. It is the choice of a programmer on how to ARTIFICIALLY adjust for what we THINK is going to happen. And every step of the climate data process is full of Fudge Factors. From initial data collection (as here) all the way through the climate models.
And THAT is one of the major Brokennesses of the whole AGW thesis.
AGW and Climate Models are mostly reporting their cumulative Fudge Factors, and only mildly reporting actual changes on the ground, IMHO. From what I’ve seen in the GIStemp code (and elsewhere) the Fudge Factors are far larger than the variance in the data due to reality.
And that’s why I am happy to decry AGW as broken. Not out of some political agenda or personal desire (heck, I’d make more money betting on AGW with my stock buys if it were true than I can make betting against it when it has political support. I can’t bet on it because it is wrong, but it is very risky to bet against it because the political favor it enjoys makes those bets dodgy. No good play.)
So we have an AGW movement based on changes of “anomalies” fabricated in a series of computer programs and reported as 0.0yy C when the Fudge Factors amount to about 1.5xx C and everyone gets all excited about the yy vs the xx and ignores that 1.5 C of Fudge, because it is buried in The Code as a series of hard to figure out Fudge Factors. (Some very exotic, like the GIStemp ability to re-write all past temperatures lower for a site because in the last decade a thermometer was changed…)
FWIW, GIStemp uses an “anomaly map” to interpolate hypothetical temperatures in the Arctic. This is done based on “estimates of sea ice”. So those changes proposed to the Fudge Factor to clean up the smoothness of the sea ice report will (if this is the estimate used for GIStemp – I haven’t figured out yet exactly who’s sea ice estimate is used.) directly lead to GIStemp reporting, due to lower ice in June post Fudge Factor change, “Higher June Anomaly! More Rapid Heating of ARCTIC!” when in fact, nothing changed.
Then GIStemp will use those “higher temps” to interpolate sea surface temperatures where their are none – thus raising the temperature of the entire ocean for 10 degrees of latitude from the polar edge. More Fudge, but this time ladled on with a Very Large Ladle… and levered off of someone else changing the Fudge Factor a bit.
Yes, the house of cards is that bad. A “reasonable” change to a satellite data processing program that cleans up a reasonable Fudge Factor a little bit can be leveraged into the whole of the Northern Oceans rising in temperatures in future June Anomaly Maps. Part of why I think the Anomaly Map is a broken tool for figuring out what is really going on. I’m sticking with actual data of First Frost and Last Frost, snow on the ground, and frankly, when Ski Resorts open and close for the season. Those have little to no Fudge Factor in them…
Sidebar:
Why I started a “blog” at all, that: “I have to explain this AGAIN” feeling…
So I started taking the same text that I had put together on the fly a half dozen times and added some graphics (in some cases) and polish and put it up as an “article”. Now, every time someone raises one of The Same Old Issues, I can just post a link (and save the moderators and everyone else reading The Same Old Answer in gory detail 100 times…)
Now, after the sea ice bump is fully thrashed out, the answer to the question becomes “See this link: wattsupwiththat.com/.. “. Much better.
If you find yourself typing the same reply a half dozen times, it is very easy to put up a “blog”. I was very pleasantly surprised. It is only slightly harder than typing a comment (AND you can correct your typos and spelling errors after publication 😉
Just hit the “wordpress.com” link under “meta” at the right of any posting here, or just go to: http://wordpress.com/ and click on the ‘start a blog’ link. Then put your repost in a posting, and put a link here (at least, after the 4th or 5th time you type the same thing… )
Ok I give up, which one is supposed to be accurate? Also where did Shawns image come from? How can we have so many variations of arctic ice? What is the defintion of ice for each of those organisations? How can I ask four questions in a row? Make that five….
Tonyb.
AndyW35,
Can’t argue, the arctic ice today is being reported at about the 2007 level. It’s also about mid-range for the last nine years, which would be another way of saying the level is at the recent average or that nothing really out of the ordinary can be said to be happening. It’s approaching its annual late-June increase in melt rate, so the next couple of months should be interesting.
Personally, I’d like to see it higher, but that’s just for the humor value. I’m a bit concerned with the curvature on the graph for 2009, but the historic data bounces too much to say what that means. Just keep an eye on it and there will certainly be more posts on the topic as the Summer progresses.
Mike
Read this nonsense
http://news.bbc.co.uk/earth/hi/earth_news/newsid_8094000/8094036.stm
Yes, after 11 years of cooling the BBC states that reindeer herds are in rapid decline because of global warming!!!
Just for the record, AndyW35 is a troll who also tries to stir up trouble on ClimateAudit and is not the LEAST bit interested in the truth, except for his version of course.
[Now, now, please be nice. AndyW35 is not currently being overly offensive. ~ Evan]
[Actually Evan, I deleted one of AndyW35’s posts in its entirety about 30 minutes ago. It was…not so nice ~ charles the moderator]
[Oh. Hmmm. Well, we will allow what we can while still maintaining a modicum of decorum. ~ Evan]
Andy,
Even the low extent represented in the latest chart is within 2 SD of the baseline data.
http://nsidc.org/images/arcticseaicenews/20090603_Figure2.png
How much are we to make of this?