tag:blogger.com,1999:blog-7729093380675162051.post291447148989276277..comments2024-03-28T13:56:47.604+11:00Comments on moyhu: My limited emulation of a Marcott et al reconstructionNick Stokeshttp://www.blogger.com/profile/06377413236983002873noreply@blogger.comBlogger27125tag:blogger.com,1999:blog-7729093380675162051.post-27118688662251889182013-03-19T10:34:54.730+11:002013-03-19T10:34:54.730+11:00This is the same crap as the EIKE paper Zorita smu...This is the same crap as the EIKE paper <a href="http://rabett.blogspot.com/2013/02/rotten-to-core.html" rel="nofollow">Zorita smuggled into Climate of the Past</a>. The really old instrumental series have a well known warming bias for early years because the thermometers were not sheltered. BEST appears to be able to correct for this (and a recent effort by Boehm handles it for a few central european stations). Links at RR.EliRabetthttps://www.blogger.com/profile/07957002964638398767noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-88341754071515697012013-03-18T12:08:35.368+11:002013-03-18T12:08:35.368+11:00No, the excitement from Borenstein et al was based...No, the excitement from Borenstein et al was based on Fig 3 in the paper which compares the distribution of proxy temperatures with two decades of instrumental; 1900-9 and 2000-9. Here's SB saying that:<br /><br /> <i>"The decade of 1900 to 1910 was one of the coolest in the past 11,300 years — cooler than 95 percent of the other years, the marine fossil data suggest. Yet 100 years later, the decade of 2000 to 2010 was one of the warmest, said study lead author Shaun Marcott of Oregon State University. Global thermometer records only go back to 1880, and those show the last decade was the hottest for this more recent time period."</i><br /><br /> I quoted above where Marcott said in the SB article that thermometer, not proxy, was the way to decide recent temp.<br /> Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-82807477762696034842013-03-18T11:15:32.069+11:002013-03-18T11:15:32.069+11:00Not the blade, but the handle. If you go to googl...Not the blade, but the handle. If you go to google images and look for holocene temperature, you see about the same behavior, and the blade, we know, is supported by instrumental records.<br /><br />EliRabetthttps://www.blogger.com/profile/07957002964638398767noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-52455692859981260082013-03-18T09:04:35.939+11:002013-03-18T09:04:35.939+11:00The people getting excited about the spike are:
S...The people getting excited about the spike are:<br /><br />Seth Borenstein,<br />"Recent heat spike unlike anything in 11,000 years"<br /><br />NBC news,<br />"Warming fastest since dawn of civilization, study shows"<br /><br />Andy Revkin etc etc<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-24120455285185898602013-03-18T03:20:20.831+11:002013-03-18T03:20:20.831+11:00OK, I wasn't sure if you were talking about th...OK, I wasn't sure if you were talking about the spike at the end (which is more like 1°C right?) or something else. I admit I don't spend a lot of energy following WUWT.<br /><br />Though oo be fair, Marcott seems (or maybe now "seemed") pretty excited about it too. Carrickhttps://www.blogger.com/profile/03476050886656768837noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-15455277361926470282013-03-17T18:58:02.320+11:002013-03-17T18:58:02.320+11:00Carrick,
There are plenty of people getting excite...Carrick,<br />There are plenty of people getting excited about this spike - just see WUWT or CA. And then there was hide the decline etc.<br /><br />The fact is that temperature didn't spike and didn't decline - we know that. Maybe the aberrant temps tell us something about the methods, but nothing about the climate.<br /><br />As to the spike, it may be the method, but it's in my reduced version too. I think Steve Mc may be right that it's a sort of end effect when dates get moved, which is OK mid-range but at the ends there is nowhere to go. I'm trying to track the numerics here.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-15804111877096907842013-03-17T17:34:58.634+11:002013-03-17T17:34:58.634+11:00I don't really understand what you referring t...I don't really understand what you referring to as to which people are getting excited about which 20C aberrations. Perhaps you could link me so I would be sure I understood your comment better?<br /><br />With the update to your post, it is looking more and more likely that the 20th century spike represents a glitch in their method (it's too high of frequency to be signal related).<br />Carrickhttps://www.blogger.com/profile/03476050886656768837noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-75122109993741714602013-03-17T16:55:45.346+11:002013-03-17T16:55:45.346+11:00Carrick,
Indeed, his recon can't show that.
I...Carrick,<br />Indeed, his recon can't show that.<br /><br />I was interested in this quote from your link:<br /><i>"The same fossil-based data suggest a similar level of warming occurring in just one generation: from the 1920s to the 1940s. Actual thermometer records don’t show the rise from the 1920s to the 1940s was quite that big and Marcott said for such recent time periods it is better to use actual thermometer readings than his proxies."</i><br /><br />That's a point that I've been making for a while. People get excited about 20C aberrations in proxies, but they shouldn't. It doesn't tell us anything about real temps; thermometers do that. It may tell us about inadequacies in the proxies, but in this case it just says there were too few of them.<br />Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-58044211114120808782013-03-17T16:24:05.580+11:002013-03-17T16:24:05.580+11:00I should have added for people who haven't bee...I should have added for people who haven't been following that particular thread across the blogosphere, that the frequency resolution of the reconstruction prevents you from making a statement of this sort. I find it mildly ironic that Marcott used the word "spectrum" in a assertion that is shown untenable by looking at the <i>spectrum</i> of his reconstruction.Carrickhttps://www.blogger.com/profile/03476050886656768837noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-76400703252787989482013-03-17T16:21:15.362+11:002013-03-17T16:21:15.362+11:00Nick thanks for updating your code.
On a related ...Nick thanks for updating your code.<br /><br />On a related topic, it probably would have been better if Shaun Marcott had avoided <a href="http://talkingpointsmemo.com/news/recent-heat-spike-like-nothing-in-11000-years.php" rel="nofollow">making this statement:</a><br /><br /><i>"In 100 years, we’ve gone from the cold end of the spectrum to the warm end of the spectrum,” Marcott said. “We’ve never seen something this rapid. Even in the ice age the global temperature never changed this quickly."</i><br /><br />The statement may or may not be right, but you certainly can't use Marcott's reconstruction to reach that conclusion.Carrickhttps://www.blogger.com/profile/03476050886656768837noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-61367145676229587002013-03-17T15:59:52.156+11:002013-03-17T15:59:52.156+11:00Thanks,
png certainly looks like a handy package. ...Thanks,<br />png certainly looks like a handy package. Lots of things I could use it for.<br /><br />One of my complaints with R is that it is so unwilling to tell me the pixel values of its axis endpoints (user to pixel mapping). I've worked out ways that basically involve rescaling from character sizes etc, but it's a pain. Otherwise I could just get the pixel values of the Marcott pic in paint and calculate the lineup.<br /><br />Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-39606323236115920272013-03-17T15:30:02.383+11:002013-03-17T15:30:02.383+11:00You say: "I made a png file with a transparen...You say: "I made a png file with a transparent background, and pasted it over the Marcott Fig 1B, which is the 5x5 recon over 11300 years. There was a lot of trial and error to get them to match. I kept both axes and labels so I could check alignment."<br /><br />Nick, for doing this sort of thing, try the readPNG program in the PNG package. YOu can then place the png image onto your screen using rasterImage. you can set the parameters so that you can plot directly onto the png image and save the annotated image.<br /><br />I've done this for some Marcott graphics and have saved the raster settings. I'll post some examples up at CA some time soon.<br /><br />One way of directly reading the Marcott xls file is to use an older R issue. I had an R.11 issue on my computer and used the older xlsReadWrite and it worked fine.Unknownhttps://www.blogger.com/profile/07966049624301704045noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-29680181847442191702013-03-17T15:14:52.114+11:002013-03-17T15:14:52.114+11:00Thanks, Willard,
I've added an update and note...Thanks, Willard,<br />I've added an update and noted it over there.<br /><br /><i>"Sometimes, I do wonder how we could have imagined reality."</i><br />At University, I learnt that reality is an optical illusion caused by alcohol deficiency.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-13903460572571903522013-03-17T14:28:23.659+11:002013-03-17T14:28:23.659+11:00STOP ALL THE PRESSES!
Thus spake Chewbacca:
>...STOP ALL THE PRESSES!<br /><br />Thus spake Chewbacca:<br /><br />> Emulating a process requires using the same process. Nick Stokes didn’t even try to do much to replicate their work. Either he didn’t replicate anything, or their method has little difference from a straight average.<br /><br />http://judithcurry.com/2013/03/16/open-thread-weekend-11/#comment-303310<br /><br />Sorry, he was not finished!<br /><br />Thus he spake again:<br /><br />> Regardless, if the latest part of the reconstruction isn’t robust, that calls into question the entire reconstruction.<br /><br />Sometimes, I do wonder how we could have imagined reality.willardhttp://neverendingaudit.tumblr.comnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-34069400645749904752013-03-17T11:49:50.523+11:002013-03-17T11:49:50.523+11:00Hello Nick. I don't have the paper yet. Unfort...Hello Nick. I don't have the paper yet. Unfortunately, the paywalls I subscribe to are mostly medical journals (I'm a biostats guy). I know a few gents who I'm sure have it. I'll reach out to them. <br /><br />BTW, great job on the R code. One of these days I need to sit down and get committed to it. My coding background is C#. What takes R 10 lines requires 100 lines in C#, LOL! <br /><br />All my best,<br /><br />HankHAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-78009936691264267072013-03-17T11:36:31.895+11:002013-03-17T11:36:31.895+11:00Paul,
I looked in more detail at Lance Wallace'...Paul,<br />I looked in more detail at Lance Wallace's spreadsheet. I think what he has done is to simply sort all the proxy obs anomalies by published time, and then make scatter plots with various degrees of smoothing by time over the aggregate. This tends to weight proxies according to their volume of numbers - ie if they report more frequent numbers, they get more points in the scatter plot.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-6696899069058152002013-03-17T11:01:57.085+11:002013-03-17T11:01:57.085+11:00Paul,
Thanks for the Wallace link. It isn't cl...Paul,<br />Thanks for the Wallace link. It isn't clear to me whether it is from a new recon or a re-plotting of M13 results. But it's more similar to M13 than mine is.<br /><br />Yes, sometimes Matlab would be handy, but I didn't have a major problem with R - in fact I didn't even notice it until I tried to make the program turnkey by reading the XL file in directly. I had been using openoffice, which only saves in the older XL formats.<br /><br />ps Something odd happened here - I wrote a response to you a while ago, but it's not here now. It may turn up somewhere.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-21349563589324189812013-03-17T10:56:07.407+11:002013-03-17T10:56:07.407+11:00Anon,
"Why would everyone else (last example ...Anon,<i><br />"Why would everyone else (last example Gergis et al) do a calibration to ensure selecion of proxies with the best temperature response ?"</i><br />because of the nature of the proxies. For tree-rings etc, the only way to assign a temperature to the width or whatever is by calibration against observed air temp.<br /><br />But Marcott et al choose proxies for which that is not required. They independently find the relation between observation and temperature. Some can be tested in lab - others by observation in different, usually marine, environments.<br /><br />And yes, every source of error tends to loss of retail. They point this out early on. But they can expect to get a better picture of the long term, partly because their proxy calibration is not affected by short-term events that might influence a calibration period and be reflected back.<br />Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-65992698111703336782013-03-17T10:36:23.681+11:002013-03-17T10:36:23.681+11:00Dating errors also disperse min/max values.
In su...Dating errors also disperse min/max values.<br /><br />In sum, for several reasons this reconstruction should be biased towards a straight line with low variability. Hard to see anything of use from this exercise or any insight in the true temperature range of the holocene.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-58813048064170472692013-03-17T10:07:49.956+11:002013-03-17T10:07:49.956+11:00Why would everyone else (last example Gergis et al...Why would everyone else (last example Gergis et al) do a calibration to ensure selecion of proxies with the best temperature response ?<br /><br />Marcott et al did admit reduced variablitiy due to low resolution. I did not notice any admission of reduced variability due to the possible use of poor quality proxy data, which also reduces min/max ranges.<br /><br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-29335802879926034652013-03-17T09:29:06.009+11:002013-03-17T09:29:06.009+11:00Hank,
Yes, I think everyone is getting fairly good...Hank,<br />Yes, I think everyone is getting fairly good agreement on the bulk of the time period. The end effect definitely seems to be virtually an artefact of the small number of proxies, and of proxies dropping out in arbitrary order.<br /><br />I don't know if you have the paper yet, but their Fig 3 does it more sensibly. They just show the distribution of proxy results, so these late spikes have no special status. Then they show the thermometer-measured difference over the century. They also show the effect of projections.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-79460596096895596942013-03-17T02:22:20.410+11:002013-03-17T02:22:20.410+11:00Nick, when I saw your results I recognized that la...Nick, when I saw your results I recognized that last 1,500 years. I overlaid my 24 proxy results (in black) on top of your results and got relatively good alignment. My weighting algorithm is different, accounting for some slight differences. In my results, I graphed only the last 1,500 years so got considerably more detail on the x axis. That uptick at the end is more a gradual recovery from the LIA starting ~450 BP. I'm not seeing that terrifying blade at the end that dwarfs the entire Holocene as Marcott depicts. There's more going on than whats in the raw proxies.<br /><br />http://img41.imageshack.us/img41/7807/recons2.gif<br /><br />HankHAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-77158441565367830132013-03-16T22:55:02.130+11:002013-03-16T22:55:02.130+11:00Thanks Nick. Your result looks similar to that pro...Thanks Nick. Your result looks similar to that provided by lance Wallace at WUWT,<br /><br />http://wattsupwiththat.com/2013/03/14/marcotts-hockey-stick-uptick-mystery-it-didnt-used-to-be-there/#comment-1248601<br /><br />Although he doesn't say so, I guess he just did a straight averaging rather than area-weighting as you've done here, showing that the general picture is "Robust" :) <br /><br />I have been meaning to do this myself but haven't got round to it yet.<br />You should try Matlab - it can read in the XL file really easily! <br /><br />PaulM Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-19005008663063625082013-03-16T22:39:14.894+11:002013-03-16T22:39:14.894+11:00No, the proxies don't need to be correlated wi...No, the proxies don't need to be correlated with air temp observations for calibration, and many of them do not have the required overlap. Even those that do have not enough resolution for such a check to be helpful.<br /><br />But yes, it certainly has reduced variability, as they acknowledge.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-2774265119394492502013-03-16T22:31:07.054+11:002013-03-16T22:31:07.054+11:00Is there any quality test of the proxies, if they ...Is there any quality test of the proxies, if they correlate at least some time with temperature ? If not, I would say the reconstruction will have, among other issues, reduced variability.Anonymousnoreply@blogger.com