Well tonight I get the great pleasure of redoing a post in the face of an inconvenient truth. To add insult to injury the error in my ways was pointed out by none other than my favorite liberal – Tamino who pointed out this section of the NSIDC data center. How I missed it in the hours I spent reading the site is beyond me. He did owe me one, at least we know he’s reading.
Note that unlike ice extent, the Arctic values for ice area do not include the area near the pole not imaged by the sensor (the “pole hole”). This area is 1.19 million square kilometers for SMMR (from the beginning of the series through June 1987) and 0.31 million square kilometers for SSM/I (from July 1987 to present). Therefore, there is a discontinuity in the “area” data values in this file at the June/July 1987 boundary.
This means that the sea ice area prior to 1987 was actually 0.88 million kilometers higher than the dataset indicated (if we assume 100% filled). Why it isn’t corrected in the data is due to the fact that the actual fill level is less than 100%. This results in an exaggerated downslope. Anyway here’s the new graphs (with a somewhat exaggerated downslope).
Here’s an image from the IUIC cryosphere page first which shows global trend.
Here’s the link check it out yourself. This isn’t that different from my original post but there is some down slope.
———————–Results from my exaggerated correction are below. ——————————–
Raw data looks like this after adding 880000 km^2 to the raw data before 1987.
A much more substantial downslope, value of – 37520 km^2/year or 1,050,000 km^2 over the 29 ish years. For the bootstrap algorithm the graph is below.
Bootstrap raw data above shows a reduced downslope of -26522 Km^2 per year or 740,000km^2 reduction.
The anomalies should be used instead for slope calculation because they clean up end point bias.
The slope from the nasateam plot was -40845 km^2/yr and the slope from the preferred bootstrap algorithm was -32037 km^2/year.
It was great that Watts up picked up my previous post followed by the disappointment of my huge mistake. Anyway this slope isn’t very strong but it is real, the arctic ice loss is greater than the gain in the southern hemisphere after the corrections are added in. It will be interesting to see how this changes when the increased area of 2008 data is added back into the curve.
23 thoughts on “Sea Ice Decreases Despite the Air Vent”
It’s good to see that you’re open to self-correction. Myself, I’m of the trust-but-verify school. Looking at the ice area concentration at
for November of 1978 I calculate raw ice coverage of 20.51 million km^2 of ice. That is raw data, but I don’t know what algorithm is used to come up with the values. If I add in 0.88 for the SMMR differential that gives a global value in November of 1978 of 21.39 million km^2.
Judging by your initial, uncorrected plot at Watts Up With That the first data point looks awfully close to 21.39. What is your source of data? The values you show in your corrected plots look way too high.
Not exactly catastrophic is it though?
Using linear extrapolation (not a very good idea but the AGW guys do this sort of thing all the time) I estimate that the Arctic ice will be gone in about 624 years… (20e6 / 32037 = 624)
Bah, posted too soon. That should have read “all ice” not “Arctic ice”. And of course the linear extrapolation is rather silly when Antarctic ice is increasing.
Earl, Andy and Chris,
The downslope of this graph is too high. I haven’t figured how to add the area back in so I just made it worse case. I tried to say it in the post but maybe it’s not clear.
BTW this does open the door to a whole group of problems as to how the rest of the world handles the transition. I need to find an image from 1978 to look at the size of the hole in the data back then and somehow figure out how much % coverage actually exists at different times of year. I’m not sure how big a project that will be but it doesn’t sound small.
I don’t know why they don’t do it. The have the gridded data before and after the change so they should know the behavior of the ice concentration in that part of the hole. It’s not an unreasonable assumption that the behavior of the still unmeasured part will be very similar. The more interesting question is how they do the anomaly. Do they not correct for the shift there either? It shouldn’t be a difficult problem to calculate separate averages for the old and new data and then calculate a constant factor to make a smooth transition. It’s much easier for extent, it’s hard to imagine less than 15% ice concentration anywhere that close to the pole.
Just normalize the comparison by hindering the new data to only reflect the area that was surveyed in the old data. Do you capisce?
Hang on: floating Arctic ice melting, on-land Antarctic ice growing. Floating ice doesn’t affect sea levels (owing to the Principle of Archimedes), land ice does. Shouldn’t this mean sea levels will be falling? After all, more of the SH ice is on land than the NH ice is (Antarctica is bigger than Greenland is it not?)
Now the warmists claim that GW makes Antarctica get colder anyway (ie. it’s not just noise). If that’s the case, then sea levels aren’t gonna be rising 20ft any time soon are they Mr Gore?
When will the AGW crowd learn?
By the way, this is a great blog. Keep up the good work Jeff! (And it’s nice to see someone who owns up when they’re wrong – if only Tamino and co. were so honest!)
Jeff, I am concerned about truth in the AGW debate – not politics, not right vs. left, not denialism, socialist conspiracies, big oil, etc.. I started to become more active in learning and reviewing the issues after years of passive acceptance that AGW was real and that the “science was settled”. I try my best to sift through the ad homs and pointless finger pointing that goes on in the blogsphere and focus on the science being discussed and debated. I believe (or perhaps hope) there are countless others who do the same. I am worried with what I have seen. I am I am not a scientist but I have enough understanding of science to be able to recognize the tendancy towards dissonance, advocacy and politics disguised as science on prominant pro AGW sites. There seems to be a lack of willingness to recognize legitimate scientific questions and scrutiny of AGW science.
Much of the progress in science is like throwing mud on the wall – sometimes it sticks and sometimes it doesn’t. We all make mistakes and sometimes we are humbled by them. Ego’s aside, in the big picture of progress – failures are as important as successes.
You have demonstrated integrity by how you have chosen to handle your mistake and it is appreciated by myself and likely many others who hope to learn. Keep throwing mud at the wall Jeff.
According to my understanding, yes sea levels should be falling *IF* ice was the only consideration. However, they have been *rising* since at least 1860 (as far back as measurements go), and this can likely be attributed to the oceans warming due to use leaving the Little Ice Age.
(Of course the AGWers don’t like to mention the Little Ice Age, and want all warming to be man-made.)
I’m trying to find how you did your calc. I wan’t to continue this analysis and I expect it will reduce the slope somewhat due to the type of correction I made.
My source is NSIDC, the links are on my previous post.
“Just normalize the comparison by hindering the new data to only reflect the area that was surveyed in the old data. Do you capisce?”
I thought about that, I need to find the gridded form of the data and reproduce the original data first, then I can take the next step. It doesn’t look too easy from the availability of the data but if I get my hands on it I will try.
I am considering making the assumption that summer ice levels would change and that change is the true offset for summer ice while the winter ice levels are probably the complete area of the missing data. – What do you think?
I could verify it through the annual amplitude.
What I did was a spot check. Looked at NH sea ice area given at ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/Nov/N_11_area.txt for 11/78. Amount is 8.95 million km^2.
Look at SH sea ice area given at ftp://sidads.colorado.edu/DATASETS/NOAA/G02135/Nov/S_11_area.txt for 11/78. Amount is 11.56 million km^2.
Add the two together and the raw global sea ice was 20.51 million km^2. That does not include the amount of ice at the north pole that was not detected due to lack of coverage.
I’m wondering if the time series you are using have an adjustment made for this already. I’ve gone to NSIDC a few times and I don’t see anywhere any mention of the polar hole except in the text files I linked above. None of the readme.txt files I found for various bootstrap and nasateam data directories address this. That doesn’t mean I’m necessarily right, but I think it merits a bit more investigation before reaching a firm conclusion that you need to adjust the area time series.
I don’t think the time series has the adjustment in it, but I have been wrong before. I added a graph to the top of this post which shows pretty clearly that my correction to the original is an exaggeration and my original post wasn’t that far off. I should have put it in my reply at Watts Up.
I’ve read through those same files a bunch of times. You would think they would make a strong comment right at the data, since the data is presented in a continuous time series. They could have given it in two separate series.
Also, there is almost no jump in the data at the point where the satellites switch over. I will take a closer look at it tonight, at this point I’m not really sure what the correct answer is.
I thought I should clarify a bit in case my prior description of the spot check was insufficient. Looking at the data I linked to above, Nov 1978 is the first month that ice area data is available. I am assuming that corresponds with the first data point in your graphs. My addition yields 20.51 MM km^2 but the graphs in your original post have a first data point at around 21.5 MM km^2 for nasateam and around 23.5 for bootstrap. If those are for November of 1978 then why the discrepancy with the 20.51?
I can’t see the right half of any of your graphs. Is there any way you can provide a link so that I can see complete graphs?
Keep up the good work!
Don’t beat yourself up about damage to credibility. It’s more credible for someone to correct and own up to a mistake – despite the messenger – than to stick your fingers in your ears and pretend you’re above a mistake.
The analysis you do clearly takes a lot of time and is appreciated.
If you have Firefox, go to the View menu, then Page Style, then click on No Style. You can now see the complete graphs, although the rest of the page may look a bit ugly.
Alternatively, just right click on each graph, and choose View Image or Show Picture. (Although this should work on any browser, not just Firefox, it doesn’t seem to work in IE7 for some reason.)
I hate to be a party pooper, but I am the real jleesnyder and I am not in this conversation!
Somehow, and apparently, harmlessly, my name, mail, and website are displayed under the “Leave a Reply” heading, and each entry (9, 11, 12, 14, etc) in the conversation marked Jeff Id. Obviously, I would like this resolved in a clear and responsible manner.
Please email me at the email address, using the title “Bogus Id” or so such, so that I know who” knocking…
Jeff. 12/17/08 4:42PM
disregard—i think i figured it out…
Jeff, keep up the good work. The ability to find errors and move forward is the hallmark of good science. This often uncovers errors that would otherwise go undetected.
Now, if only the AGW crowd would follow this example. First, they should step up and start admitting many of their obvious errors (Mann et al being one example). Only when that starts to happen will I start taking anything they say seriously.
Chris H Says:
December 16, 2008 at 12:56 pm
“Using linear extrapolation (not a very good idea but the AGW guys do this sort of thing all the time) I estimate that the Arctic ice will be gone in about 624 years… (20e6 / 32037 = 624)”
Your estimate of the lifetime of Arctic sea ice is irrelevant to the important effect of the sea ice on the world’s climate, reduction in albedo. If the sea ice is gone in the winter, since the sun doesn’t shine in much of the polar regions, it makes no difference to the albedo.
The important thing is the summer ice, when the sun is available to add heat to the oceans, if they are not covered by ice in summer. Looking at the North and South Polar regions together obscures the fact that sea ice could be totally melted in summer, in the Arctic. Currently, the sea ice in the Antarctic disappears in summer.
The fact that the polar oceans are cold enough to freeze over in the winter, even if there is a temperature rise, does not make up for the loss of summer ice.
Jeff should be Congratulated for having uncovered the glitsch in the data, but it is not going to affect the projections of global warming due to melting ice. The difference between leaving out, or putting in, the observation hole over the north pole, and the glitsch due to switching of satellites, doesn’t affect the conclusion that Arctic summer ice is declining, and will reduce the earth’s reflectivity of solar radiation.