21 thoughts on ““Someone called Ryan O””

  2. Take one million pc reconstructions using Mannian or any other method and you get one million pieces of junk and five million peer reviewed articles saying why it proves who killed Kennedy, the Roswell alien crash landing, H.G. Wells as a non fiction writer, the moon landings were fake and that God is dead.

  4. Andrew_KY–
    Did Gore specifically allude to Nazis? Or is he just tap-dancing around Goodwin’s law by referring to WWII and Churchill?

  5. Well he (Gore) once called his critics “digital brownshirts” so he doesn’t seem to feel the need to “dance around” referencing Nazis.

  6. Thanks for the link Lucia, I finally added your blog to the blogroll out of guilt for my own laziness for housekeeping duties. I’ve been going through the CA link to here for months. Ryan did some really great work here.

    I see Larry T is the ultimate skeptic and really can’t blame him. The thing I like to use as a sanity check is the simplified area weighted reconstructions. If you take a look at this link below, you’ll see the match is quite good even in the individual trend HF information.

    http://noconsensus.wordpress.com/2009/07/07/area-weighted-tpca-check/

    I’m pretty sure Ryan has to a reasonable degree, fixed the reconstruction.

  7. Lucia,

    “Winston Churchill aroused this nation in heroic fashion to save civilisation in World War II” is the money quote, apparently…

    …which doesn’t include the word ‘Nazi’ but of course requires knowledge of the existence of Nazis, knowledge of what Nazis did, and the suggestion that the Climate Denier Meanie Versions Of The Nazis Are Coming To Get You, to give the quote any meaning.

    Andrew

  8. Andrew_KY–
    I agree that going to England and discussing WWII is getting very, very close to showing Godwin’s law in action. It is true that when I read what Gore said, the Nazi image pops into my head. ( In the US, WWII references may refer to o the war in the Pacific, but not in England. )

    The reason I asked is I just like to see just how direct the Nazi references get on both sides. Al Gore seems to be snuggling right up to the line but avoiding direct Nazi references.

  9. Lucia,

    Which do you think is the greater offense:

    1) Comparing political opponents to ‘Nazis’ and using the word ‘Nazis’ to do it…

    or

    2) Comparing political opponents to ‘Nazis’ but using references instead of being straightforward

    The purpose and intended outcome of 1) and 2) are the same.

    Andrew

  10. Andrew_KY–
    They are both equally bad. I just like to monitor what, precisely, is being done. I suspect Gore has speech writers removing the overt Nazi references and substituting indirect references. Either way, it’s worth noting.

  11. Jeff–I think it’s great that RyanO has been able to post his material and get feedback. It looks like he’s at a point where he is satisfied with what he found, and can publish.

  12. Jeff, I am a mathematician by trade and have been both a database administration and a data administration. I have aIso have worked on several projects at NASA as a contractor and worked with people in places like JPL, Max Plank Institute in Germany. I am a very strong proponent of having good data and what i see is bad data bent, folded and mutilated into something that is totally meaningless bunch of chicken scratchings. My approach would not be combining sets of observations into one long temperature record through a questionable principal component analysis but in eliminating bad bad completely, do a few mathematically justified corrections and get a set of vectors showing the slope of the temperature change. Then the change rates vectors could be combined to give a least squares approximation of the global rate of change. But then again that may be taking me back to my trajectory analysis and orbit determination roots as an applied mathematician

  13. Larry T,

    The point of Steig’s reconstruction is to calculate not only a final trend but also a spatial distribution of trend. What we have is a low quality high resolution satellite dataset and a higher quality low resolution surface station dataset. Data is missing throughout, surface station location is typically on the ocean edge and a large number of minor problems exist in the data that are nearly impossible to fix. The question becomes, what is the best we can do with this mess?

    With the different lengths of missing data, my own opinion is that an area weighted reconstruction which employs nearest station infilling for missing data is about the best we can do for trend. If we simply combine slopes as you suggest the weighting of each station would become hopelessly scrambled due to missing data and proximity of other stations.

    The fact that Ryan matches this simplified calculation and it has such nice verification statistics gives me confidence that the method he’s improved on is working well and is close to the best temperature trend distribution possible with this dataset. Take a closer look at the code he provided, you may be convinced. Dr. Steig’s version was a mash in the end, but this isn’t along the same lines as the original.

  14. Is anyone not able to reach http://hadobs.metoffice.com/ ? I wanted to download some data, but I haven’t been able to reach the site all day. All I get is this:

    Web Proxy

    The requested URL is unavailable at this time. The following error was reported:
    Failed to connect to server

    For further information contact:
    admin@metoffice.com

  17. I also can’t seem to access that data. So it’s not your and lucia’s connection.

  19. Thanks much for the post, Lucia! 🙂 The highlight is much appreciated.

    For Larry T: To a certain (large) extent, I agree with you. Regardless of the verification statistics in my reconstruction, it is speculative at best. It is very difficult – actually, impossible – to recreate missing data. This type of reconstruction effort, regardless of who does it, is subject to a bunch of assumptions that cannot be proven to be true. If you had enough information to prove them . . . well, then you wouldn’t need to reconstruct in the first place.

    This effort is not supposed to be taken as a “this IS the 50-year history of Antarctica”. It is meant to illustrate what happens to the reconstruction if it is done in a mathematically valid fashion.

    On another note, there are several ways to do this, depending on what assumptions you make, all equally valid under the condition that the assumptions are true (which is unprovable). Most of these methods yield a very similar result to the above. Not surprisingly, most of these share a similar set of assumptions. A few do not yield a result similar to the above. Again, not surprisingly, they have marked differences in the assumptions.

    Therefore, if the result is highly dependent on the underlying assumptions – assumptions that cannot be shown to be true – then that result should be regarded as suspect.

    So your suspicion is not unwarranted.

  20. Lucia,

    As far as publication goes, I think there are 2 major issues left to analyze. One involves getting the 0200/1400 cloud masked data sets from Comiso to evaluate how much effect the cloud masking could have had on the spatial eigenvectors for the AVHRR EOFs. I have not had luck with this yet, and I do not intend to hold up drafting a paper. But it would one open issue that I would prefer to include in the paper.

    The second issue is a must, in my opinion. As our infilling algorithm turns out to be almost identical to the DINEOF algorithm, I have been in communication with Dr. Beckers to get his opinion. There are important differences between the two algorithms that reflect the different intended use. I am not the right person to audit my own logic that I used to write our algorithm because it would be very difficult for me to look at this problem in an entirely different light. I’m not that good at shifting my own paradigms. I sincerely believe that we need a knowledgeable third party to evaluate what we have done – in detail – before we attempt to submit anything.

    So I believe we are getting very close to being able to draft something that could be publishable. 😉

  21. I am honored to get a answer from Ryan O as in past posts that i have gotten some from “Saint Lucia”. I guess my main point is that you start with validation and verification of the data before you try to do analysis. Missing data, errors and outliers need to be dealt with. While some systematic errors can be fixed with aid of computer algorithms the majority of data problems can only be handled with good old hands on research. I think we are losing the baby in the bathwater as we normalize, fill missing data, compute moving averages, compute anomalies, group data into monthly, seasonal or yearly averages.
    My position on global climate change, I was for it before I was against it. Based on solar activity, I predicted a warming trend back when everyone was predicting another ice age. Now, I believe that we are heading towards a minimum again based on solar activity. If the effects of global warming may cause some problems (and many more good things), the coming minimum, depending on its depth, could cause massive loses of life as previous minimums have done.

Comments are closed.