During the the last week, NSIDC graphs of arctic sea ice extent have been dropping so steeply that many have called them into question. Finally NSIDC ended the daily updates and have left the last “good” image of May 21st in place in the web folder, but have placed an “out of order” sign on the website:
As we first pointed out to NSIDC back on 2/18/09 (even though it “wasn’t worth blogging about”) the sensor has been on the fritz for quite awhile, calling the whole arctic sea ice series into question. From their most recent announcement, it looks like that it is now “DOA”:
Here’s what they say now.From NSIDC’s web site:
Update: May 26 2009 The daily image update has been temporarily suspended because of large areas of missing data in the past week. NSIDC currently gets its data from the SSM/I sensor on the DMSP F13 satellite, which is nearing the end of its operational life and experiencing intermittent problems.
NSIDC has been working on a transition to a newer sensor on the F17 satellite for several months. At this time, we have more than a year of data from F17, which we are using to intercalibrate with F13 data. The F17 data are not yet available for near-real-time updates. We will resume posting daily updates as soon as possible, either from F13, if the present problem is resolved, or from F17, when the transition is complete.
It doesn’t look promising to get any usable data for the last 6 months or more, since it clearly has been corrupted by the sensor issues.
Meanwhile the AMSR-E on the Aqua satellite chugs right along on JAXA:
From NANSEN, here is a map showing differences between AMSR and SSMI. There are some huge chunks missing.
See the source image page here
(h/t to Fred Nieuwenhuis for the link)
Personally I think it was folly for NSIDC to try to use different channels on the DMSP F13 satellite to nurse the dataset along, as we’ve seen it is not just the single channel on SSM/I sensor that has had problems.
Transitioning to the DMSP F17 satellite “may” be a plan, but the AQUA satellite and teh AMSR-E package seems to be quite reliable and with a number of years of life ahead. It is also used by many other agencies to reliably gauge sea-ice.
IMHO, NSIDC is doing themselves no favors by sticking with the DMSP SSM/I satellite platform package. The science world has moved on with AQUA’s AMSR-E, and it is time for NSIDC to move on as well.
Otherwise, they are going to be “has beens” using older technology. Get with the program guys. You need good supporting data so incoming director Mark Serreze can give us his fabulous forecasts and media soundbites that don’t seem to come true.
Pamela Gray (18:50:53) : “That blip was explained before. It is a satellite adjustment for the time of year in it’s orbit, or something like that. Can’t remember exactly but it is a necessary adjustment and nothing unseemly.”
The blip in June is due to a change in the algorithm used to calculate summer ice vs. winter ice. the 2009 line will blip up next month also.
Can someone please take the battery out of their annoyingly loud fire alarm mouths? Oh noes! 2009 will see the entire earth’s sea ice melt! We’re all going to be chugging salt water on the coast! For God sakes man! Run for the hills!
ROTFL
This page explains some of the problems they have faced between the F13 and F15 satellite sensors.
http://nsidc.org/arcticseaicenews/2009/022609.html#satellites
Basically, they are switching back and forth between two data sources which have unrecoverable problems (in the sense that the NSIDC has always had problems with matching up data between two different sources to start with.)
Now I remember. Water pools on top of still frozen ice can look like open water. Thanks for the reminder.
Pamela Gray (18:14:58) :
I think you’re seeing my link clipped. If I break into two pieces:
http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh
?fm=05&fd=25&fy=2008&sm=05&sd=25&sy=2009
the fm, fd, and fy parameters refer to the left (first) image, and
the sm, sd, and sy parameters refer to the right (second) image.
Again, the year ago vs. today comparison is available at
http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=05&fd=25&fy=2008&sm=05&sd=25&sy=2009
Or, the today vs. a year ago comparison is at
http://igloo.atmos.uiuc.edu/cgi-bin/test/print.sh?fm=05&fd=25&fy=2009&sm=05&sd=25&sy=2008
Forget about all these other sources and faulty algorythms and stick with NCEP, although it can be “noisey” at times:
http://iabp.apl.washington.edu/maps_daily_ncepice.html
Glenn, I would double check your site. It also looks contaminated. Ice doesn’t melt in square, triangle, or rectangle shapes. But thanks for the link. It has other information that I am interested in.
This is from the NCEP site.
“A satellite we use in constructing the sea ice analyses is overage and now providing poor data. This has corrupted the analysis. We are working on getting better data, but this may not be a speedy process. When we have confidence in a newer process, it will be announced here. In the mean time, though figures will continue to be produced, they cannot be relied upon as formerly. 25 February 2009”
I hope they keep trying to use different sensors than everyone else is using. If there is data available from several sources, it’s better to have it all being interpreted. And we know that one sensor will fail before the other, but we won’t know which until it happens. (Yes, we know estimated lifetimes but it’s not history yet.)
There may have been a gradual resistive increase prior to overt failure. So even earlier readings may have undererported, well prior to the steep drop off.
I’m not a one for conspiracy theories or for believing in deliberate manipulation. Of course these things can happen but I presume against them until there is very clear proof.
What I do believe is that every measuring device produces results within a margin of error. Sometimes the margin of error is wide, sometimes it is narrow, it all depends on the quality of the particular device and how efficiently it is working.
With satellite measurements of surface temperature it is possible to assess how accurate the measurement is by comparing it to a different measurement taken at the same time and at the same place. But, even then, the measurement is only as accurate as the most accurate of the measurements taken. That might sound tautologous but it isn’t. A satellite system might be absolutely accurate while a ground measurement taken with a blob of mercury in a glass tube might be hopelessly wrong for any number of reasons. Which is being tested and which is the benchmark against which the other is tested? Equally, the satellite system might be flawed fundamentally yet coincide with a measurement taken on the ground because the ground measurement is equally inaccurate. Nonetheless, standards can be applied to physical measuring equipment to give us a reasonable degree of satisfaction that they provide the base measurement (plus or minus however much of a degree) against which the accuracy of satellite data can be assessed.
Where is the physical control measure for sea ice? In order to have any chance of understanding these things, I have to look on satellite assessments of sea ice extent as photographs. One day the photograph shows a certain amount of what appears to be solid water and another day it shows a different amount. I can understand that. I can also understand that photographs taken from the same place a few days or weeks or even hours apart can appear to show different things even when nothing has changed.
Faults with cameras can give misleading pictures long before anyone realises there is a fault. I suppose my question is (leaving aside conspiracy, fraud or any other jiggery-pokery): how accurate are the satellite measures when everything is working well and how can anyone know how accurate they are?
I speak as a man who hasn’t taken a photograph since 2000.
Mc: I hammered away at this for some time probably much to the annoyance of many. It was noticed (with one glaringly obvious exception) that ALL the ice adjustments were always “downwards”. This might have made some (in particular myself) a bit over paranoid so at least we can now have a record of the changes, to keep people on the toes as they say… There is a sort of record of changes at Mike Maralenas site here:
http://mikelm.blogspot.com/2007/09/left-image-was-downloaded-from.html
My understanding that at this a stage the only reliable one would seem to be DMI
http://ocean.dmi.dk/arctic/icecover.uk.php
which seems to show current = 2005 extent
I don’t know which satellite they are using
Nice Miss Gray said (19:58:21) :
“Ice doesn’t melt in square, triangle, or rectangle shapes.”
It does in my glass.
Ric Werme: From those CT comparisons it looks like 2009 has a much GREATER extent ice and definitely snow than 2008
Ric Werme (19:35:32) :
The image is clearly inconsistent with other arctic sea ice images posted at NSIDC (before being taken down). The Hudson bay is almost all ice right now. Something is totally honked up with their map.
Mr. Bigot,
Just wanted to let you know how much I enjoy your comments. I always get a chuckle out of them, even the serious ones. Glad to see you hitting the boards more often.
Mike
>Pamela Gray (18:50:53) : “That blip was explained before. It is a satellite >adjustment for the time of year in it’s orbit, or something like that. Can’t >remember exactly but it is a necessary adjustment and nothing unseemly.”
>
>The blip in June is due to a change in the algorithm used to calculate summer >ice vs. winter ice. the 2009 line will blip up next month also.
From someone who works in the GOES weather satellite operations…I have a few questions.
1) if this is a ‘winter ice to summer ice’ adjustment, where is the corresponding down-blip at the ‘summer ice to winter ice’ adjustment?
2) at this adjustment in June, which is the correct data – before the adjustment or after the adjustment?
3) presuming the data is more ‘correct’ after the adjustemnt, at what point before the adjustment does the data go out of calibration?
I don’t care if they have to adjust the instruments or not (we do spring/fall patch temp adjustments on the GOES imager but this only affects the dynamic range extremes and very little of the full image is affected). However, unless the graph is showing raw data, the graph should not show any ‘adjustment’ to the instrument. To me, this causes me to question the accuracy of the data that the instrument is providing. I would think it would be better to do a gradual adjustment over a few weeks to transition so to prevent data ‘blips’ which show no reality to what is actually happening.
Just my thoughts…
Jeff
VG (20:26:32) :
My understanding that at this a stage the only reliable one would seem to be DMI
http://ocean.dmi.dk/arctic/icecover.uk.php
which seems to show current = 2005 extent
I don’t know which satellite they are using
They’re using SSMI like NSIDC and NORSEX.
Phil.
Which website do you prefer for sea ice updates?
Thanks,
Mike
They are in transition to a much more AGW friendly management, so throwing out some misleading data for Internet Consumption should not hurt anyone now should it?
At least the timing was not catastrophic, like Al Gore waving NSIDC extent numbers in Congressmen’s faces calling American business criminals, oops been there done that with Revkins opportunity drive by piece.
For a world gone climate crazy nobody seems to be putting much effort into getting the observed data right. We are talking about a radical reshaping of civilization based on this but nobody seems to want to put the time and effort into actually provide unbiased data for comparison to the research to see if the theory is still valid.
Scientists cannot advance the theory or withdraw it without accurate observations, right now it is a dog show of who can provide the worst data quality.
JeffK (21:05:48) :
From someone who works in the GOES weather satellite operations…I have a few questions.
1) if this is a ‘winter ice to summer ice’ adjustment, where is the corresponding down-blip at the ’summer ice to winter ice’ adjustment?
As I pointed out above when I explained the source of the ‘blip’, that occurs on Oct 15th.
2) at this adjustment in June, which is the correct data – before the adjustment or after the adjustment?
Both, sufficiently earlier when here is no surface water and later when it is mostly wet.
3) presuming the data is more ‘correct’ after the adjustemnt, at what point before the adjustment does the data go out of calibration?
It’s obviously not an instantaneous transition, the actual dates are a compromise to minimize the total error.
I don’t care if they have to adjust the instruments or not (we do spring/fall patch temp adjustments on the GOES imager but this only affects the dynamic range extremes and very little of the full image is affected). However, unless the graph is showing raw data, the graph should not show any ‘adjustment’ to the instrument. To me, this causes me to question the accuracy of the data that the instrument is providing. I would think it would be better to do a gradual adjustment over a few weeks to transition so to prevent data ‘blips’ which show no reality to what is actually happening.
That would require accurate knowledge of the state of the ice surface which could be difficult, however JAXA is working on a smoother transition which they hope to have in place this year.
Thanks Phil. People like you help make this blog worth the time.
It was noticed (with one glaringly obvious exception) that ALL the ice adjustments were always “downwards”.
Not to worry.
Surface temperature adjustments are always “upwards”.
So it all evens out in the end.
{insert smiley face indicating mordant sarcasm}
a jones (17:18:58) :
And by the way I think it is the Ar-C-tic not the artic: which in the UK is a type of lorry.
The NSIDC global warming juggernaut continues to roll downhill.
According to the picture of Artic sea ice extend, the Catlin expedition should have been swimming when they were rescued.