cause two out of three ain’t bad
Julienne Stroeve has responded to my query about yesterday afternoon’s three day shift between climatology and current year.
Steve, as Walt mentioned earlier, NSIDC was in the process of changing the way we do the 5-day average, switching from a 5-day centered mean to a new method that takes the average of the previous five days. As you know, we average in order to remove artificial ice from weather effects. Weather effects can result in noisy data that is not a real sea ice signal. The underlying data going into the averaging are unchanged.
Apparently, the new averaging was implemented yesterday. Note the ice has not been lowered, it has been shifted back two days. The climatology is a 9-day running mean rather than a 5-day, so the climatology will be shifted by more days.
If you have any further questions, feel free to contact me.
As Richard Muller might say, we don’t do that in science.
By switching to a five day trailing mean for current, and a nine day trailing mean for climatology, they created an artificial relative shift of two days. The center point of the current data is -2.5 days, and the center point of the climatology is now -4.5 days.
This accounts for two days error out of the three. The third day of error is because someone lobbed off the April 16 data.
The timing of this error looks very suspicious, because the data would have nearly touched the mean today – which hasn’t happened for for many years.
Here is their article: http://nsidc.org/arcticseaicenews/2012/04/daily-graph-changes/
Yes, I am taking to Julienne about this. Someone down there screwed up big time, and they need to sort it out ASAP.
Nolo contendere, that really is dumb.
They could also filter the data through a set of Bayesian priors. (The IPCC could instruct them on this, if they’re not up to speed.) One such would be that the expected ice extent at this time of year is anywhere from 6 to 16 M km**2 with equal probability. That would take care of this pesky problem of the curve approaching the reference mean once and for all.
Off-topic, but check out Obama’s Maldives gaffe:
http://harndenblog.dailymail.co.uk/2012/04/obamas-real-maldives-gaffe-abandoning-britain-on-the-falklands.html
Obama thinks he could see the Maldives from Argentina if he had a house there. I think he only knows about the Maldives because his science advisor Holdren tells him those islands are sinking due to AGW.
Looks like the Secret Service (scandal) bailed him out here, not just for the geography goof, but the much more substantive foreign policy stance weakness.
He needs no scandal to bail him out, the MSM would never report anything that would make the chosen one look bad. Come on, wake up and smell the Kool aid!
For some reason, Dr. Stroeve’s explanation of the “new”system reminds me of
when my daughter was 3 years old and would stack the deck for a game of “Candyland”.
It was cute when she did that. This, not so much.
Steve is correct that the switching to a trailing mean resulted in a mis-match in the timing between the climatology and the daily extent since different averaging periods were used (this date mis-match wasn’t a problem in the centered means). We’re working to fix the problem. However, Steve should note that switching to a 5-day trailing climatology mean actually puts the current extent farther away from the climatology.
Julienne,
There is no legitimate adjustment which will shift the current wrt to the climatology. If you are going to use different averaging periods, then you have to use a centered average for both. If you are going to use a trailing mean, then you have to use the same averaging period for both.
Neither legitimate approach will cause a relative shift. If you are using a methodology which produces a relative shift, it is incorrect.
Did you not see the above comment where I said you were correct about the mis-match and that we’re fixing it?
Julienne, thanks for responding. Are you guys and gals unaware of the timing of these changes and how this appears? There are several blogs and thousands of readers around the world watching data and graphs get adjusted before our very eyes.
This business with NSIDC is just the latest in a series of events that have shook any trust the observers may have had in any of this “science”. And, before these recent events fade from memory, it is likely millions more will have lost trust.
While I’m sure the cynic may say many of us don’t lend any trust to any of the various institutions, (some of us do, some don’t) it is the others who will be influenced by these events whom I’m speaking about.
The problem which may occur is that people are fickle and prone to overreactions. While I’m convinced that we need some considerable changes in the way this nation conducts science, I’m equally sure we don’t want to throw the baby out with the bath water.
I would encourage the people in our science institutions to become a bit more diligent and vigilant when it comes to handling our data and graphics.
The margin of allowable error decreases with the level of certitude in proclamations.
My very best wishes,
James
I don’t trust the institution that is doing this. I will not trust it in the future either.
Don’t touch the data – plot only the ‘raw’ data – uncorrected.
The retouching should be done be a team *different* from the one that primarily accesses the data.
Agreed. Trust is much easier kept than regained.
Julienne – Kudos to you for responding.
Typically all climate websites are on one or the other sides of the valley shouting (unheard) at each other.
Anyone confident with their knowledge and observations is not afraid to be challenged.
Agree.
Would also suggest scientists/websites proactively transparently explain data and statistical adjustments rather than posting changes without explanation. It leads to distrust.
John, yes the change should have been on our web site before it was implemented, that was a breakdown in communication unfortunately.
This isn’t a communications issue, it is a technical issue. If the change were mathematically correct, chances are I never would have noticed. The reason I noticed was because it is not correct.
Is it necessary to badger the woman?
While they’re sorting this mess out, they should also take the opportunity to re-calculate the averages so they reflect a 30-year period, the way everybody else does it.
Or is that too much to ask?
It is on the list of things to do in the coming months. Next time we’ll make sure the in-house communication is better so that everyone associated with ASINA knows exactly when changes have been made and that the writers have time to post the change on the web site.
Thanks, Julienne.
I’ve done a comparison of averages for sea ice area between 1979-2000, and 1979-2008, and the differences range from 1% in March to 7% in September. It will be interesting to see how your figures compare.
I think everyone in this field feels like they’re in a race and gotta get to the finish line and get their news release that proves their point out first.
Whether it’s ready or not for prime time.
And the result is chaos – awash in information and thirsty for knowledge.
They’re working to fix the problem. Hence the current graph is no good.
I’m guessing that the next version will be no good too.
you are sick
I agree that if they are using trailing data then the current and reference averages should use the same number of days.
But if they don’t correct this, you should bear in mind that it will bite them back in September (when they will probably suddenly correct it!), because moving the reference curve to the right will mean that just after minimum it will “bump”towards the current data, if that is below as it has been the last few years.
Rich.
I tend to agree with this. I’m so jaded against the scientific institutions nowadays that I think it likely that all future adjustments will be in the direction of more warming and less ice. I dare them to prove me wrong.
As you know, we average in order to remove artificial ice from weather effects. Weather effects can result in noisy data that is not a real sea ice signal.
=============
and this will affect past ‘averages’ how?
and does it mean that the new way of doing averages will show less noise, and as a result less ice
……..and the old uncorrected averages will be more noisy and show more ice
Mostly good balanced debate here – let’s not pile on – we WANT open and fair and moderate exchange of information, maybe even come to a true consensus – although fortunes will be lost then, and I won’t have anything to do in my spare time…
This is the sort of thing Rahmstorf & co. are famous for.
http://rankexploits.com/musings/2009/source-of-fishy-odor-confirmed-rahmstorf-did-change-smoothing/
http://rankexploits.com/musings/wp-content/uploads/2009/06/jeans6fnvqa.gif
Met O did the same thing.
http://climateaudit.org/2008/03/30/like-a-dog-on-a-bone/
Funny how they also find these “errors” that put them in a bad light.
I don’t believe one thing these people say anymore. They are dealing from a loaded deck. She’s got the Jack…..
I can’t believe all of the adjustments going on lately. Must be because of Rio+20…
These people remind me of little children – you have to keep an eye on them all the time or they’re sure to get into mischief.