As I mentioned yesterday, NSIDC had an oops moment, but with the help of skeptical blogs, was notified of the problem and responded timely and appropriately. They posted this update today:
Update, April 19, 2012: The nine-day trailing average climatology on the daily data graph has been changed to a five-day trailing average, to be consistent with the five-day trailing average for the daily data.
I verified their correction for the climatology was accurate with a new overlay, combining the unaffected graph NSIDC’s Dr. Julienne Stroeve sent me from their internal server storage Tuesday night with the corrected one published on the web today:
http://nsidc.org/data/seaice_index/images/daily_images/N_stddev_timeseries.png
The extent data and the climatology now match, whereas yesterday they did not. The x-axis offset is to be expected given that we are comparing graphs with the temporal data offset due to a trailing average they implemented.
This is what the same comparison looked like yesterday, for graphs made on the same day:
So, problem solved.
Unfortunately, somebody jumped to a conclusion and has already had to issue a correction.
April19: NSIDC graph still appears wrong
now reads:
Correction : April 19 – NSIDC Graph Now Lines Up With April 16
The maxim “haste makes waste” seems appropriate.
The way NSIDC’s Dr. Walt Meier and Dr. Julienne Stroeve handled this should be an example to other agencies that don’t bother to even respond to skeptics.
Aaaaah, I see that NORSEX has now been updated to today, and it is above the 1979 – 2006 average.
I think the NSIDC freaked out seeing the Arctic Sea Ice approaching average, so they figured they could make an adjustment under the table, but that failed. I guarantee you no such change would be made if sea ice was near 2007 levels, I find it more than mere coincidence how sea ice almost crosses the average 1979-2000 line, but they suddenly decide to make changes to the average, thus making the current data appear warmer. I don’t understand why they wont use the other 10 years of available data, through 2010, even though it has large swings, I don’t see why not, but wait, it would make the new data appear cooler and less alarming, so they don’t want to do that do they? 🙂
Anyone see the NORSEX ice site which, unlike the NSIDC uses a 1979-2006 average, and currently it is almost an entire standard deviation ABOVE AVERAGE?
Looks to me like this ‘peer review’ thingy works… as long as the ‘peers’ include sharp eyed citizens with no conflict of interest or hidden agenda!
We are inside the ‘1 sigma’ range for the 1979 – 2000 average for ice coverage at this time of year. Situation ‘normal’ in the Arctic ocean. How is the ice coverage in the Bering Sea doing? Is it still at a ‘max condiiton’? Is there any historical measure of daily/weekly/monthly ice coverage for that area?
Guys asking about confidence intervals – there’s a gap in satellite cover that has to be infilled using a computer model. Next question please!
They should be proud of real data- no averages please. Whatever qualities it contains should be plotted out.
Before you all get excited about Arctic ice returning to normal, bear in mind that 2 years ago it was also at a good level in April but fairly low by September. The summer melt will obviously be led from the Western side, and the NE passage might open again, but the NW passage must be more doubtful.
Rich.
Anthony, her name is Julienne…
REPLY: Yep, I know another person who spells it the way I wrote it. Fixed – A
NSIDC deserves the standing and fluttering applause they are getting from my birdfeeder. THEY PROVIDED THE RAW DATA!!!! Would that the ‘team’ would??????
I think people wouldn’t have jumped on NSIDC as hard for the initial error were it not for the continued hysterical ramblings of a certain Mark Serreze. I understand he’s clearly passionate about what he does but he does immense damage to the credibility of the organisation he works for so that in the end simple mistakes are seen as something a little more sinister.
Its just a shame that his shenanigans counter the excellent work Walt does in engaging on this site.
You can’t say by “simple visual inspection” of a chart that “they did the math wrong”.
The 1 in this case is roughly 5 standard deviations above the mean, and is pretty close to being the maximum possible value it can theoretically have, since we’ve got zero variance on 29 out of 30 samples (and since obviously everything I just stated is scale independent in a certain sense). To put it another way, you cannot observe a data point in a sample of size N=30 that is ten standard deviations away from the mean — there are sanity limits that scale with N. With only 30 samples, how can you state with any confidence at all that the probability of obtaining 1 from the central limit theorem is 0.00000001 or whatever erf(10) is? That’s nonsense.
Actually, sometimes you can. Suppose you have 30 samples. 29 of them are anything you like — say 0 (putting the mean of the distribution very close to 0). The 30th is 1. The standard deviation is then around
Given a roughly 30 year base, by visual inspection the standard deviation around the end of March is around 5 standard deviations, which means either that a bloody miracle happened (given the variability on both sides and the size of the mean, it means that the sea ice extent has been almost exactly the same for every year but 2007 in late March) or it means that there is a simple error somewhere that somebody should look for, and look hard, until they are certain that a miracle happened. And then they should try to understand the miracle. In the meantime, the graph above is moderately unbelievable, and would be so even if the data is exactly the way it has to be to make it correct.
Perhaps you might study some math and statistics before concluding that simple inspection cannot often pick up errors. Actually, it is one of the best, and simplest ways of doing so — when a curve doesn’t make sense somewhere, it is often time to go back to the data and/or algorithm used to produce the curve.
This is the case even if the curve turns out to be true and correct! “Unusual” results are often precisely the points were interesting discoveries are made.
rgb
Guys asking about confidence intervals – there’s a gap in satellite cover that has to be infilled using a computer model. Next question please!
This may well be true, but then the obvious point is that they’ve set the sd wrong in the infilling model. It needs to interpolate the sd on the adjacent sides of the end of March, not shrink it to “unbelievable”.
rgb
Well done to NSIDC!
But I hope they understand that on the back of very recent radical changes to sea level datasets (which make it look worse – in flagrant disagreement with tide gauge datasets) and to the temperature record (HadCRUT 4 again further cools the past) we are just slightly suspicious.
When all of this data is transparent, maybe we could go back to a small degree of trust. But trust is sorely lacking at the moment.
Perhaps NSIDC might, for example, let a contract to Climate Audit to audit climate data? Corporations are required to be audited. Could it be so bad to have external auditors catch these issues rather than force unpaid scientists and engineers such as Steve McIntyre to do so?
After this incident like this, I think it’s safe to say we’ve all lost a substantial amount of trust in NSIDC, seems like they’re certainly on the pro AGW side of things.
Here is my estimate of the daily sea ice extent going back to 1972 (when there actually was satellite images available and the NASA Team developed daily sea ice extent estimates from these using an algorithm which is mostly the same as that used today).
I’m matching up 4 different datasets here (including the Cryosphere Today ice area data) so there might be some problems but having worked with this for a few years now, I think it is pretty accurate.
2012 (in Red) is closing in on the 1972 to 2011 average ice extent.
And I’ve also included a line (Purple) showing the type of change that would be required to have an ice-free Arctic on a typical minimum day. Huge changes are required in both the March sea ice extent maximum and then the amount that melts each day compared to the average to get us anywhere near to an ice-free Arctic. Just so you know, 2011 followed the average melt per day almost to the “T” so having a melt rate 20% higher than average each day from March to September is probably impossible. The ice-free Arctic people are relying on “non-physical” models.
http://img818.imageshack.us/img818/9855/dailyseiapril182012.png
It was good of Dr. Meier to participate here, but if he wants to know why so many view government-sponsored climate science with cynicism, he has only to connect the dots between Hansen, Gleick, Serreze of his own organization, and today’s hateful rant by Zwick in Forbes magazine.
This has been a pleasant interlude. People are actually behaving in a more or less civil manner. Congratulations to all concerned.
We need both the people who stir us up and those who calm us down. Complacancy is a vice, but so is hysteria. We humans are not the most stable items on earth, and I actually think we are at our best when we move between two views. That’s why we have two eyes. I myself become very compacent, when I go to bed, and each morning the alarm stirs me up. I feel the better for having been complacent for a while, but to stay in bed usually means I am ill.
By the way, concerning: “Tenuk says:
April 19, 2012 at 11:37 am……
…….It is puzzling how temperature up in the Arctic is all over the place…
http://ocean.dmi.dk/arctic/meant80n.uk.php …..”
Hopefully I can tempt Joe Bastardi to add to my reflections on the swings in arctic temperatures. As I recall it was he who suggested that, when it gets windy up there, the temperature rises. When it is calm, the temperature falls.
When I thought about this, it occurred to me that higher temperatures likely create more ice. Why? Because of the wind. It is not due to the wind-chill, but rather because winds shove the ice around, and not only piles up pressure ridges, but create open water in the form of “leads.” These swiftly freeze over, because even when the wind has temperatures ten degrees above normal, we are still talking thirty degrees below zero. This open water swiftly becomes several feet thick, whereas if the same area was not exposed and was instead insulated by as much as nine feet of new ice, only an inch or two would have been added on the very bottom.
In other words, though a windy winter may show temperatures ten degrees above normal, it is constantly piling up pressure ridges and exposing leads of open water which freeze over, and far more ice will be produced.
Which brings me to the touchy subject of “ice volume.” You are likely to hear a lot of Alarmists bring up the subject of “volume,” when both “extent” and “area” fail to alarm anyone.
As far as I can tell, “volume” is not determined by counting pixels, but rather by a bouncing radar beam from a satellite. I have my doubts about whether radar can pick up pressure ridges, especially the smaller ones, which you can’t see even in the best enlargements from outer space, for a small one may only be three feet tall and ten feet across, though it extends for miles. From outer space it looks thinner than a spider-mite’s web.
However most of an iceburg is under water. If a pressure ridge sticks up three feet it sticks down twenty-seven feet. If it towers, (like the really big pressure ridges,) twenty feet, then it has roots that go down one hundred eighty feet.
These pressure ridges are a jumble of ice mashed together, amounting to a large volume in a small area. They are made up of lots of small plates, ice as thin as a few inches thick, but as much as nine feet thick, only glued together by snow and whatever else holds chunks together when it is thirty below. When it warms they fall apart and revert to a bunch of plates floating in water, and this messes with our minds. Why?
Well, let us suppose an area a mile wide got crunched together into a big pressure ridge thirty yards across. When that ridge melts, it will revert to a mass of small chunks a mile wide. That simply does not compute, to our common sense. How can an area thirty yards across abruptly become a mile wide? Especially due to MELTING!
I think that is what happened with the ice, in 2006. In the noontime-dark of mid-winter, 2006’s extent was of the lowest in recent years, however it included so many pressure ridges that, in September, 2006’s extent was of the highest, in recent years.
I am betting a nickel, (and no more,) that the same thing has occurred this past winter. As the ice melts it will turn out that the “temperature up in the Arctic is all over the place” you mentioned meant a lot of wind was piling up lots of pressure ridges, and as these pressure ridges fall aprt they will make mincemeat of our sensible calculations, based on the sensible idea ice is of the same thickness all over the place.
In any case, roughly two thirds will have melted by next September, and September’s extent will have roughly tripled by next March.
I’m willing to bt a whole dollar on that.
Walt Meier says: April 19, 2012 at 11:21 am
If folks have other suggestions, we’d be happy to hear them and take them under consideration – post a reply here.
Hello Walt
What I would really like to see from NSIDC is Global Sea Ice Extent charts. I have looked around your site and not been able to find them, i.e.:
http://nsidc.org/data/seaice_index/
http://nsidc.org/data/seaice_index/images/
Cryosphere Today offers a Global Sea Ice Area chart;
http://arctic.atmos.uiuc.edu/cryosphere/IMAGES/global.daily.ice.area.withtrend.jpg
which offers helpful perspective and generates significant public interest and traffic.
I decided to plot the Global Sea Ice Extent March Monthly Anomalies to see for myself, so I grabbed the data from here;
ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/Mar/N_03_area.txt
ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/Mar/S_03_area.txt
and plotted the Northern and Southern Sea Ice Extent March Anomalies;
http://i42.tinypic.com/2wocarb.png
http://i39.tinypic.com/2uo2gyv.png
in order to validate the data and methodology. Here are NSIDC’s originals for comparative reference:
http://nsidc.org/data/seaice_index/images/n_plot_hires.png
http://nsidc.org/data/seaice_index/images/s_plot_hires.png
I then merged the two data sets and created a chart for Global Sea Ice Extent March Anomalies:
http://i40.tinypic.com/20f21wo.png
Based on my calculations, the slopes of the trendlines are as follows:
Northern Sea Ice Extent Anomalies March 2012: Slope = -2.5% per decade
Southern Sea Ice Extent Anomalies March 2012: Slope = 3.2% per decade
Global Sea Ice Extent Anomalies March 2012: Slope = -1.4% per decade
Would it be possible to have Global Sea Ice Extent charts added to NSIDC’s site?
Thanks for your consideration
JTF
D. J. Hawkins says:You can go to NSIDC’s web site and download the data for yourself, for crying out loud. Do anything you want with it, plot it any way you please.
Nope. Not there. Besides the obvious point: if you want me to do NSIDC’s job, then please send me their paychecks.
If you tried to plot every year on the graph it would be a jumble of lines. Try it yourself, if you’re not too lazy.
Oh I see, they don’t know how to make a graph. It’s incompetence, not deliberate scientific fraud. Well, DJ, please see Bill Illis’ graph below. See, it CAN BE DONE, just not by government functionaries.
See – owe to Rich says:…we’re trying to work it all out. … 2007 is displayed because it was the biggest Arctic melt in the recent record, so worth comparing with.
Well, Rich, “we” aren’t going to work anything out without the data, with phony gray areas, and with emphasis on one extreme year. Where is the maximum year? Is that of no interest? How does the (truncated, missing) data as displayed tell “us” (or anybody) anything about long-term change?
Bill Illis says:Here is my estimate of the daily sea ice extent going back to 1972 (when there actually was satellite images available and the NASA Team developed daily sea ice extent estimates from these using an algorithm which is mostly the same as that used today).
I’m matching up 4 different datasets here (including the Cryosphere Today ice area data)…
http://img818.imageshack.us/img818/9855/dailyseiapril182012.png
There you go. It CAN be done, and it HAS been done, not by NSIDC but by Bill Illis, Skeptic Extraodinaire.
Dear Walt,
Please send Bill your paycheck.
rgbatduke
I’m not sure I understand your issue regarding the SD’s. I see the baseline is 1979-2000, not the entire record to date. 2007 isn’t even in this data set so obviously won’t “broaden” the SD boundaries. If there has been a general decline in the ice extent over the instrumental record, wouldn’t one expect that later years would eventually fall beyond the 2 sigma boundary if the base period remains the same? It would be a question of “when” not “if”, depending only on the rate of decline.
“Walt Meier says: April 19, 2012 at 11:21 am
If folks have other suggestions, we’d be happy to hear them and take them under consideration – post a reply here.”
Very well. This is one:
The future data needs also to be compiled in the current method for decades to verify the values with new method. It would be important to verify the error levels between methodology changes.
AlexS says:
April 20, 2012 at 1:19 am
===========
I agree, noting the error rates and basis of observation (15% as opposed to 30%) is helpful.
It would also be great if they would pick-up some of the better features from The Cryosphere Today.
Latest Regional Sea Ice Coverage and Anomalies for 14 areas is a great approach. It shows exactly where sea ice isn’t forming. Showing the Complete Annual Satellite Record is also a great way to objectively communicate the sea ice record. Both are available on The Cryosphere Today.
Northern Hemisphere Sea Ice Area — Complete Annual Satellite Record
http://arctic.atmos.uiuc.edu/cryosphere/arctic.sea.ice.interactive.html
Latest Regional Sea Ice Coverage and Anomalies
http://arctic.atmos.uiuc.edu/cryosphere/
Now all we need is for NCDC to correct their spurious Mannian padded “9 pt binomial filter” : http://www.ncdc.noaa.gov/temp-and-precip/time-series/index.php?parameter=tmp&month=3&year=2012&filter=1&state=110&div=0
Like Met office Hadley , they will probably only feel the need to correct this misrepresentation when it produces a deceptive down swing.
I emailed them 3 days ago, no sign of even an acknowledgment so far.
This is a classic case of padding of the filter window to get the filtered result to go beyond the end of the filter. It’s pretty clear here that they are repeating the final March temperature over fully half the window , thus producing an unprecedented rise that does not even exist.
To Rhys Jaggar, 19 April, 9.32 am: I have been monitoring this graph for years. The Arctic ice briefly moved into the 2 std region for a few days some years ago but the recent trending is without precedent in recent years. We may be seeing a step change just like the one that lowered ice levels and caused all the GW hysteria. Could we now all draw this new situation to the attention of journalists on magazines like Time who still write about the Arctic melting?
Dr. Meier,
Thanks for the opportunity to comment about the NSIDC site.
The addition of the Ice Lights section is great and most of the articles are reasonably balanced. I have one constructive criticism that extends to all sites and the IPCC reports. Observations made about aspects of the climate system fail to properly reflect the climate system as a whole. The top-line approach fails to properly reflect the current state of scientific understanding related to climate forecasting.
Arctic Sea Ice or to be more accurate Northern Hemisphere Sea Ice Area and Extent is influenced by a number of interrelated forces; Salinity, Oceanic Oscillations, Atmospheric Oscillations and Wind Direction, Milankovitch cycles, Solar activity during a portion of the year, etc.
In my opinion, the NSIDC Arctic sea ice 101 page needs to be rewritten to properly reflect Sea Ice changes in relation to the current scientific understanding of the climate system with an emphasis on changes in relation to Natural cycles. A tall request, I’ve never seen it done properly on any site, but it would be a remarkable tool for education.
Thanks for considering the request.
Icelights: Your Burning Questions About Ice & Climate
Arctic sea ice 101
Excerpts:
Arctic sea ice is declining at an increasing rate in all months of the year, with a stronger decline in summer months. Researchers who study climate and sea ice expect that at some point, the Arctic Ocean will lose its ice cover completely in late summer. A variety of evidence suggests that Arctic sea ice is declining because of climate warming resulting from increased greenhouse gases in the atmosphere. Because greenhouse gases persist and are expected to increase, scientists see no reversal to the downward trend in ice extent.
Despite year-to-year variations, satellite data show a decline of more than 11 percent per decade in September ice extent, since the satellite record started in 1979. Before 1979, the data are less comprehensive, but shipping records and other evidence show that the ice extent has been in a continued state of decline for at least the last hundred years. Climate models predicted that Arctic sea ice loss would accompany warming temperatures in the Arctic. But the ice loss has happened faster than any models predicted, and researchers now expect that the Arctic Ocean will be ice-free in summer well before the end of the century.