A note about before and after at NSIDC

Some readers may have noticed the sharp uptick at NSIDC for the Southern Hemisphere sea ice extent, as highlighted in WUWT Sea Ice News #13:

Here is how it looks today at NSIDC:

Source: http://nsidc.org/data/seaice_index/images/daily_images/S_timeseries.png

So that there aren’t any speculations about the sudden disappearance of data, I’ve asked Walt Meier of NSIDC about it and posted the response:

Hi Walt,

Do you have any idea why the southern sea ice extent uptick disappeared?

Before, July 10

http://climateinsiders.files.wordpress.com/2010/07/s_timeseries1.png

After, July 11

http://nsidc.org/data/seaice_index/images/daily_images/S_timeseries.png

No mention of it anywhere

http://nsidc.org/arcticseaicenews/

Seems rather odd to make such a correction and not notify end users of such  a change. This lack of notice then leads to speculative phrases, like “death  spirals”. 😉

Best Regards,

Anthony Watts

He answered within minutes. Walt is very good about being responsive on issues related to NSIDC. For that he should be commended.

Hi Anthony,

It was an error in the source data we use in the sea ice algorithm we  run. We have an automated QC that will take it out, but it requires data  from the following day. We can QC manually too, but since it happened on  a Sunday no one was around to address it. It did get corrected  automatically when processing occurred this morning.

We generally don’t post a notice for isolated incidents like this since  it is just a part of routinely dealing with near-real-time data. We do a  general discussion on our FAQ:

http://nsidc.org/arcticseaicenews/faq.html#quality_control

Thanks.

walt

I may have more from Walt in the coming day or two. Stay tuned.

Advertisements

28 thoughts on “A note about before and after at NSIDC

  1. Oh thank heavens! For a day there I thought it was worse than I thought James Hansen thought it was!

  2. Why can’t everyone measuring climate be so forthright and diligent? Sad really. Walt may need to slow down. He’s giving the rest of the climate guys a bad name.

  3. The NSIDC smoothing algorithm often seems to produce retroactive adjustments. Some are up; some are down. It looks fishy at first glance but it’s legit. After years of watching this I have become quite confident that it is a normal, neutral smoothing algorithm. The NSIDC plots are less jagged than some of the others but it is fair and obviously compatible. I would trust the NSIDC data with my life.
    However, all data less than 4 days old should be viewed as preliminary. I fully expect the current flat line for Arctic sea ice to get bent downward in the next day or two. The actual slope is small but not zero. See, for example:
    http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm

  4. This is a repeat – we have had the same answer on the arctic ice with respect to anomalies and the subsequent corrections.

  5. Anthony, next time you converse with Dr. Meier you might suggest that his graph show the last few days of data — those subject to QC modifications — in a different color. A simple footnote would explain the appearance and make it unnecessary to consult the FAQ. Information density would be improved with little added complexity.

  6. Using the “ruler to the screen” method on on the enlarged graph, I show the uptick to be 3 days of data. (July = 3.6875″ Uptick = 0.375″)

  7. May I take advantage of this rather pleasant moment on WUWT to ask all climate scientists, would you please stop using the word “anomaly?” The word has a very long history and a very important meaning apart from anything climate scientists say about anomalies. The word “anomaly” means “an element of experience that has proved recalcitrant to all of our attempts to explain it by appeal to our theories.” In its ordinary meaning, an anomaly is a scientist’s nightmare, undeniable experience that defeats all explanation. No wonder climate scientists have a gloomy outlook on the future. Every bit of data they encounter is a disastrous defeat for their theories. It is as if Wall Street reporters began each show by saying “Stay tuned for today’s Wall Street Crash report.” “In today’s crash, the DJIA dropped 7 points and a crash twice that size is expected for tomorrow.” To make matters worse, “anomaly” is the only word for data reports that climate scientists use. Couldn’t some data be merely frightening or worrsiome rather than an anomaly? Finally, no doubt to drive the nail home, the gargantuan word “anomaly” is used to describe what are mere changes or variations in the data record. Couldn’t changes or variations be called changes or variations? Everyone else in the world uses the words “change” and “variation.” Could it be that climate scientists think of themselves as Dr. Doom?

  8. This “real time data” issue has come up before. As Dr. Walt explains it, and the numbers make sense, this is the way to do the job – although I like the idea Gary has @ 5:17.
    Note that the recent temperature thread had reports of bogus max-temp numbers that never get corrected. The explanations of “why not?” are as bogus as the numbers.

  9. Maybe it could be the right time to ask Dr Meier what he thinks about using PIPS to compare year-to-year ice thickness and calculate volumes with that – as opposed to PIOMAS. There’s nothing like an expert view.
    REPLY: done already- A

  10. Kudos to Dr. Meier for his responsiveness and hard work, but given the sensitivity and controversy surrounding arctic sea ice perhaps NSIDS shouldn’t post for public consumption near-real- time data that hasn’t at least had the automated quality control applied. What’s the rush? Is life and death involved somewhere? Shipping navigation issues? Pole bound, ice crawling climate publicity stunts impeded?

  11. I dont suppose ther is any chance that Dr Walt Meier can be tempted to do a guest post on WUWT regarding his job, what his organistaion does and how he thinks he can better interact with the greater population of interested climate science observers?

  12. Theo,
    From dictionary.com:
    a·nom·a·ly   [uh-nom-uh-lee]
    –noun, plural -lies.
    1. a deviation from the common rule, type, arrangement, or form.

    The climate usage of “anomaly” to mean “difference from the mean” conforms quite well with this definition. Not likely to change any time soon – that would be anomalous… (3. an odd, peculiar, or strange condition, situation, quality, etc.)
    kb

  13. Yeah, Dr. Meier has a history of responding politely and with helpful information. I’d be happy to read whatever he has to say.

  14. Frederick Michael says:
    July 12, 2010 at 4:55 pm
    I would trust the NSIDC data with my life.

    Now that is what I call FAITH.
    Tom in Texas says:
    July 12, 2010 at 6:00 pm
    Using the “ruler to the screen” method on the enlarged graph, I show the uptick to be 3 days of data.

    Now that is closer to what I see and it needs a proper explanation.
    Cassandra King says:
    July 12, 2010 at 9:43 pm
    I don’t suppose there is any chance that Dr Walt Meier can be tempted to do a guest post on WUWT regarding his job, what his organisation does and how he thinks he can better interact with the greater population of interested climate science observers?

    I would be very interested to know HOW they “determine” an area of ocean with at least 15% sea ice. The whole concept seems fraught with difficulties, especially based upon satellite images.
    First up
    What is their definition of an “area”?
    Is it a physical measure of land area, such as a square kilometre or Acre or Hectare or what?
    Is it some measure of area based upon the image resolution, such as a pixel grid?
    This is to try to get some handle on the granularity of their determinations.
    Second Up
    How do they process the satellite images so that their “areas” are of equal size?
    The Antarctic is big, the globe is round and a satellite is only ever directly above one point on the ground at any moment of time.
    Third Up
    How do they factor in the height of the sea ice in their calculations?
    The angle of observation could be very important when accessing whether a large, tall iceberg occupies 15% of a given area.
    Fourth Up
    How do you determine what is actually ice?
    How do you differentiate between land ice and sea ice?
    How accurate is this determination?
    What is the error range?
    Fifth Up
    What are you quality control and smoothing algorithms actually doing?
    Can we see a raw data time series so that we can compare the raw data with your final processed data?
    Bottom Line
    What is the overall accuracy of the area of ocean with at least 15% sea ice?
    Conclusions
    Given the data processing challenges mentioned above an overall error factor of plus or minus 15% would not seem unreasonable… which leaves me wondering if it would not be more truthful to publish data regarding areas of ocean with between 0% and 30% sea ice. I don’t think this would be popular as it would start exposing the myth of satellite science…
    I remember the story about the naked Emperor… the guy with no clothes on… well the more I read WUWT the more I need convincing that any scientist is actually wearing a white coat 🙂

  15. Maybe, Dr. Meyer could also comment on the recent changes of the arctic sea ice extend which appears to run into a zero melting gradient:
    http://nsidc.org/data/seaice_index/images/daily_images/N_timeseries.png
    In contrast http://www.ijis.iarc.uaf.edu/en/home/seaice_extent.htm
    exhibits a finite gradient, which appears to be more reseonable.
    Finally, a look on
    http://nsidc.org/data/seaice_index/images/daily_images/N_daily_extent_hires.png
    shows sea ice in the northern Baltic sea and also in the Sachalin straight, which has gone since May. For comparison of the Baltic sea data , see
    http://www.bsh.de/de/Meeresdaten/Beobachtungen/Meeresoberflaechentemperatur/SST_d.jsp#0

  16. Well done Walt Meier. I respect his approach and his willingness to address the questions from us “skeptics” in a professional manner.

  17. Some thoughts on “normal” versus “anomaly”.
    What does it mean to be “normal” in a statistical context? Think of the Bell Shaped Curve. That curve can be described with a “mean” (the calculated central average value) and the standard deviation ( a statistical measure of the spread in the data). Once we know the mean and standard deviation, we can compare sub-sets of data against the ‘normal’ distribution, and we know that the 99.7% of the data will fall in the range of the mean +/- 3 standard deviations. That is to say, for 1000 values from that population, only 3 would be expected (either high or low) outside that +/- 3 standard deviation range.
    There are statistical tests to determine that the ‘shape’ of the curve is the same above and below the mean (is the data normally distributed or skewed high/low). There are also tests to determine whether a comparison population of data are statistically the same or statistically different from that ‘normal’ population.
    Then comes our consideration of ‘anomaly’. If a small bit of data is say 4 or 5 standard deviations away from the mean, that is outside the normal distribution and therefore ‘anomalous’. When that is a ‘blip’ and the subsequent data falls back within ‘normal’ we often look for a ‘special cause’; perhaps an instrument error or a bad data file or some non-normal occurrence (e.g. a volcano spews ash or warms up and melts some ice).
    The climate folk have taken the word ‘anomaly’ – which should be understood to mean WAY OUTSIDE of Normal – and they are using it to describe tiny variations which are barely measurable.
    The late Dr Deming, world acclaimed for his statistical and quality consultations, warned management that to make decisions without understanding the statistics was the height of folly, and that over reacting to small perceived changes typically made things worse.
    Then recall the post Climategate interview with Dr Phil Jones, when asked about temperature trends for the past several years, he replied that they were ‘not statistically significant’. Hence .. no anomaly.

  18. suggest that his graph show the last few days of data — those subject to QC modifications — in a different color

    Rather than that, how about simply showing those days’ data points as points in the same color, rather than connected (smoothed) lines. It would appear as a dotted line, still visually distinct from the dashed line for the prior year’s graph.

  19. Malaga View says:
    July 13, 2010 at 1:20 am
    Now that is what I call FAITH.

    Consider the things you trust with your life: tires, brakes, bridges, etc. Is all trust faith? The NSIDC did not get my respect automatically, they earned it.

Comments are closed.