Data Comparisons

GISS October Anomaly: Warmer than September.

42 Comments

The GISS Land/Ocean anomaly has been posted and is available here. October’s reported anomaly is 0.78 C, which is warmer than the 0.49 C reported for September during October and also warmer than 0.50 C reported for September in November.

It’s always fun to see what happened to recent months’ temperatures. The November update is shown on top, highlighted in gray; the table currently in Google’s cache is shown below:

Figure 1: GISS October Warmer than September.
Figure 1: GISS October Warmer than September. (Click for larger.)

As you can see, during October,

  1. September got 0.01C warmer.
  2. August and July got 0.11C and 0.01 C cooler respectively.
  3. June got 0.03C warmer.
  4. The temperatures for months Jan-May managed to remain at their previous values.

Obviously, I’ll be updating my various records. With some luck, the temperatures for June-Sept will stabilize by year’s end. 🙂

Update

Due to the unfortunate mishap at GISSTemp, I have proposed a brownie bet regarding the October GISSTemp anomaly. Visit Guess the future October GISSTemp anomaly.

42 thoughts on “GISS October Anomaly: Warmer than September.”

  2. Like I said GISS temp is crap. Two satellite records show the same result of roughly no change from Sept to Oct. I suspect the issue is the same thing that happened in March 08 where surface stations missed cooling elsewhere.

    Just don’t take my word for it. Anthony thinks the same way as well:
    From Watts up with that
    REPLY: GISS isn’t considered reliable anymore. It’s an outlier, and does not take microsite biases and UHI properly into account. HadCRUT “appears” better, but they won’t disclose their methods. Hansen would either until last year when the Y2K error that came from site surveys was exposed by McIntyre and there was a loud call for the code. Still can’t get that train wreck of Fortran programming to compile and run properly. I don’t even bother posting reviews on GISS monthly anymore because Hansen is so politically biased, he’s lost track of the science, such as his defense of vandals in England. GISS/Hansen just isn’t credible anymore. Sat data is the key to understanding, not the corrupted surface record. – Anthony

  4. BobB–
    You have to admit that comparing GISS’s updates to the one in cache is great fun! Who knows what GISS will report for November in December? 🙂

  5. Well, there is a long history in economic data reporting of revising the previous period report when the latest report is issued. IIRC for example, the latest US unemployment figures adjusted the previous two months downward as well. Of course, we know why they do that. They don’t have all the data when they issue the (preliminary) report. With GISS, we don’t know much of anything.

  6. DeWitt–
    I think with GISS, the changes are caused both by arrival of missing data from the previous month and by their algorithm that does something to replace missing data.

    The method may be fine, but it’s impossible to know whether this October report will remain the warmest October they have ever reported.

    It was mostly warm around here (except that day it snowed.)

  8. The other question is: Where’s Waldo? Did the NH not cool as fast as usual? Did the SH warm faster? The Tropics? Or was it some local phenomenon or a general warming relative to the long term average?

  14. I live bang in the middle of Central England, and am only too aware of what a cold month October was here overall. In particular, the lying snow we had at the the end of the month was a once in a lifetime event.
    Anyhow, this is confirmed by the CET record. Oct 08 had a mean CET of 9.7C, which gave an anomaly of -0.9C as against the official 1961-1990 baseline (10.6C)
    http://hadobs.metoffice.com/hadcet/cet_info_mean.html
    and by my calculations, -0.8C as against a 1951-1980 baseline (10.5C)
    http://hadobs.metoffice.com/hadcet/data/download.html

    But GISS, whether with a smoothing radius of 250km or 1200km, has central England at least 1C above average for October 2008 (compared with its 1951-1980 baseline).

    There may be a good explanation for this (?), but on the face of it I’m starting to understand why some people are so anti the GISS temperature record. If something appears blatantly wrong, it’s very hard to take it seriously.

  15. GISS does seem to tell us you were toasty warm in England. It will be interesting to see HadCruts records when they come out.

  16. I didn’t want to complicate my last post by including reference to any possible controversy in the CET record, so I’ll do it in a separate post here. Insofar as the CET record may exaggerate recent temperatures, this merely gives my last post extra weight since the negative anomaly for Oct 08 should be even greater.

    You can see from the following that the mean CET figures for the last decade have been effectively 1C above the 30-year baselines prior to 1980.
    ( http://hadobs.metoffice.com/hadcet/graphs/HadCET_graph_ylybars_uptodate.gif )
    But this cannot be seen in the max temps for the individual stations, and only in the min temps for one of them:
    http://www.anenglishmanscastle.com/archives/004482.html
    It’s all very complicated, as different stations have been used over time, adjustments made etc. etc. But there may be more to this issue than meets the eye.
    Anyway, sorry can’t say any more because been called away.

  17. It appears from Fred’s pdf that this is indeed the same problem as March 2008. South Africa and Canada are not fully represented

  18. Apparently CET is not the only location but all of the UK. Here’s the first line from the MetOffice UK October Summary: “Maximum, minimum and mean daily temperatures were all below average across the UK. Most areas, provisionally had their coldest October since 2003, but Northern Ireland had its coldest since 1993 with mean temperatures around 1 deg C below average.” http://www.metoffice.gov.uk/climate/uk/index.html

  19. Thanks Fred, I was going to post something similar, but you beat me to it. A few further details/points:
    Oct 2003 was also exceptionally cold, and only one of three Octobers since 1981 which had a lower CET than Oct 2008 – the other two being the post-Pinatubo Octobers of 1992 and 1993. So “coldest October since 2003” doesn’t really do this October justice.
    From the following it can be seen that most of the country had an anomaly last month between -0.5C and -1.0C, and there were several areas besides Northern Ireland which had anomalies of -1.0C or colder i.e. parts of Scotland, NW England, eastern England

    http://www.metoffice.gov.uk/climate/uk/2008/october/maps/TMean_Anomaly%20No%20Stations.jpg
    It’s clear from the detailed monthly assessment that “some areas had their coldest October day for over 50 years” during the cold snap at the end of the month
    http://www.metoffice.gov.uk/climate/uk/2008/october.html
    And here’s a couple of links re: how London and Surrey had their earliest lying snow in 74 years and 121 years respectively:
    http://www.thisislondon.co.uk/standard/article-23579420-details/Arctic+blast+brings+London+earliest+snow+for+70+years/article.do
    http://www.thisissurreytoday.co.uk/news/Surrey-enjoys-October-snowfall-121-years/article-435799-detail/article.html
    Here in Birmingham no one has produced a definitive statistic as far as I can tell, but the best evidence suggests it’s a minimum 74 year event like in London, and if the snow showers recorded in Oct 1934 didn’t lie (it’s not clear from the only historical source i’ve seen), then it looks like it could be a 100+ year event as in Surrey.

  20. Lucia,
    I just remembered our “conversation” from last month’s GISTemp release. Don’t you find it amusing that the anomaly value for August was initially released as 0.39, changed to 0.50 the next month(as seen above) and now is re-adjusted back to 0.39 in this month’s release?

  21. Fred– Yep! August’s temperature rose during September then fell during October. That’s what’s fun about GISS. 🙂

  23. Actually the mistake is not GISS this time. It’s with NOAA/GHCN see http://wattsupwiththat.com/2008/11/10/giss-releases-october-2008-data/

    Executive summary…
    – Russia had *HUGE* positive anomalies
    – somebody checked a few Russian station textfiles. In each one they checked, the October temperature was *IDENTICAL* to September.

    It looks like Russian September data was copied over to October data at NOAA/GHCN, which GISS uses as their “raw data”.

  25. Here’s hoping the missing Canadian and Australian data doesn’t compensate for the Russian correction. 🙂 One wonders if the August anomaly debacle (0.39 to 0.50 to 0.39) is related to the Russian data issue. Things that make you go “hmmmmm”.

  27. Comment#6479
    Lucia: that GISS image you show really tells you everything about GISS and “global” warming…. as they say hoisted by his own petard… LOL. People will eventually notice that it ain’t warming up… whatever is done to manipulate the data..and there will be of course legal consequences at the end….

  29. There is a problem with the October GISS temps. At many Russian stations, the October temperature (not anomaly, but the actual temperature) is reported to be exactly the same as September. I saw this I believe on Climate Audit and maybe Watts Up With That where they checked this out in more detail. That explains why the anomalies in some parts of Asia are something like 10C!

  30. GISS is absolutely to blame for not checking their source. How can anomalies of up to 13.7C not cause a red flag to go up? How bad is the quality control with the NOAA/GISS that they could overlook such an obvious mistake?

  31. Jared–
    I’m surprised GISS didn’t have scripts in place to throw flags. Sure, sometimes one month will match the previous one, in the tropics. But the stations don’t move around. It would be perfectly easy to have written a script that threw flags readable by humans, who could then double check.

    If such flags existed, they would surely have noticed the many stations in Russia with duplicate temperatures. (And yes, they should have noticed this even if the problem was introduced upstream of their processing.)

  35. Lucia I did a quick check on one of the russian stations. 9 mentioned by Mcintrye) over the past 100 years October has averaged 17C colder than Sept. 1 sig was about 3C. For over 100 years straight october has never been close to sept records. Its a simple matter to screen for this kind of “anomaly”

  37. From a post at Watt’s website, was the October 2007 Siberian anomaly similar to this year? The Siberian heatwave October 2007 was concurrent with the historic Arctic sea ice comeback.

  38. steven–
    Yes. It obviously would be very simple. Gavin’s statement was this:

    These other cases seem to be mostly legitimate tropical stations where there isn’t much of a seasonal cycle. That makes it a little tricky to automatically scan for this problem, but putting in a check for the total number or percentage is probably sensible going forward.

    Well… erhmm… so how does the fact that you sometimes get temperature matches in the tropics make writing flags “tricky”? Years and years of data exists. One write a script to create statistics for individual stations or groups of stations. The stations in Russia which experiences seasons will have statistics that exhibit summer/winter swings. So…. you use the local statistics. You set the flags based on local data.

    NOT HARD. Will it require a programmer a man month? Sure. But this isn’t “tricky”! And GISS has programmers assigned to calculate GISSTemp, so what’s the big deal?

    I don’t know why Captain Obvious hasn’t dropped by GISS and suggested that if they automate their data processing, and post data before humans can check it, they ought to create automated error flags. Sheesh.

  40. Shoot! I don’t want the job.

    But I think it would be wise for someone to cut funding for the GISSTemp effort, and someone else to fund elsewhere. It could be done at a DOE lab — possibly inside the DOE-ARM program. GISS temp is observationally based, ARM does data. I’m sure the various national labs involved in ARM could find someone assign to the 0.25 FTE task (this is the level Gavin identifies for the project).

    NASA could subcontract work, though it might be better if the task just went to DOE entirely. As one of the prime uses of the data is to test NASA GISS models, getting this task outside GISS would eliminate the appearance of conflict of interest. I know it’s only the appearance of conflict of interest. Still, eliminating the appearance would be a plus.

    And according to Gavin, it’s such a small task with such a small budget… so why not shift it?

  41. I now know part of the reason why GISS temperatures change retroactively:

    “John Philip Says:
    11 November 2008 at 4:01 PM
    Gavin,

    A dignified and measured response.

    Just to be clear – is 908 the number of stations in a single corrupt data file, which is one of several such files? Or is 908 the total number of GHCN stations used in GISTEMP? Seems surprisingly low, if the latter.

    JP

    [Response: The rate at which stations report varies. This month 908 stations were reported by Nov 10 for October. The number of stations that will report eventually is about 2000. Of those 908 stations, 90 had this oddity – which is a significantly higher percentage than one would expect. – gavin]”

  42. Alan–
    So, 10% of the reporting stations had this oddity. This is more than you would expect especially in Russia in the fall!

    Why there is no QA algorithm to detect that t(i) = t(i-1), and flag it is beyond me. If they need an exception for tropical stations, fine. But temperatures swing in the arctic.

Comments are closed.