tag:blogger.com,1999:blog-7729093380675162051.post8433022593729358419..comments2024-03-16T02:27:38.423+11:00Comments on moyhu: A necessary adjustment - Time of ObservationNick Stokeshttp://www.blogger.com/profile/06377413236983002873noreply@blogger.comBlogger6125tag:blogger.com,1999:blog-7729093380675162051.post-36698973536698696922014-11-25T13:30:51.117+11:002014-11-25T13:30:51.117+11:00Thermometers are also used to measure temperature ...<a href="http://www.thegreenbook.com/products/thermometers/" rel="nofollow">Thermometers</a> are also used to measure temperature in laboratories while critical chemical tests are being conducted. Since these thermometers do not need to be in contact with the object whose temperature is being measured, things become all the more simple. Anonymoushttps://www.blogger.com/profile/08348367560139366345noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-25331120043265402992014-10-03T09:03:18.537+10:002014-10-03T09:03:18.537+10:00Zeke is right, in GHCN the time of observation bia...Zeke is right, in GHCN the time of observation bias is indeed only applied in the USA. If you do not do so, the statistical homogenization methods (pairwise homogenization algorithm, PHA) takes care of this problem and the results are almost the same.<br /><br />This will, however, often not be the case in other countries. In other countries the time of observation is typically not determined by the observer, but by the meteorological office. The time of observation can still change sometimes, but then these changes are made to the entire network. That means that you cannot see these changes by comparing a station with its neighbours, as the PHA does. Only at the national borders would you detect and correct such time of observation changes. It is thought that these changes are random and do not influence the global mean temperature much.<br /><br />In the CRU temperature dataset, the data is homogenized by the national weather services. They know their own time of observation problems. That the global trend of CRU and GHCN is similar is an indication that the time of observation changes do not influence the global mean much. Victor Venemahttps://www.blogger.com/profile/02842816166712285801noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-82730627109639877672012-10-03T16:56:24.758+10:002012-10-03T16:56:24.758+10:00Zeke, I just remember Steven Mosher talking about ...Zeke, I just remember Steven Mosher talking about other countries (I think he mentioned Canada) that at least have TOBS issues. I won't argue with you what's i the data for GHCN!<br /><br />It shouldn't be that difficult to set an upper limit on TOBS only, which is really all that's needed. <br /><br />I won't argue it would be silly to not make a correction, but it'd be nice to know how big the correction is that was done.Carrickhttps://www.blogger.com/profile/03476050886656768837noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-29246563718834940502012-10-02T04:00:17.317+10:002012-10-02T04:00:17.317+10:00Carrick,
TOBs adjustments aren't generally do...Carrick,<br /><br />TOBs adjustments aren't generally done outside of the U.S., at least for the data included in GHCN. That said, the automated homogenization algorithms seem to do a good job of picking out and correcting discontinuities introduced by TOBs changes (both Berkeley and Menne's PHA have been shown to do this when applied to non-TOBs adjusted data), so in practice they should be accounted for in GHCN v3.<br /><br />Since there isn't a "global" TOBs-only adjustment series available, it is difficult to estimate how large the effect of TOBs adjustment is on global temperatures. Zekehttps://www.blogger.com/profile/09757819498566612533noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-41236304225513405152012-10-02T03:23:52.778+10:002012-10-02T03:23:52.778+10:00Nick: The fact is that if an adjustment is appro...Nick: <i> The fact is that if an adjustment is appropriate, then it is required. It's not optional</i><br /><br />With the proviso that you can always not make the adjustment then quote the uncertainty introduced by not making it.<br /><br />In my area, where we commonly deal with systematic effects, that's one way we quantify "how important the systematic effect is" and gives some idea "how good a job we need to do to correct it."<br /><br />My preference is to show "before and after" when practicable.<br /><br />That said, do you remember what the difference in global trend (say 1960-2010 or pick your own interval) is with and without TOBS? I would have pulled it from Vose but it wasn't immediately obvious how to do so.<br /><br />I'd guess given the limited area over which the TOBS adjustments were made, it would be less than a 5% effect on trend (possible within bounds of the systematic uncertainty introduced from irregular geographical sampling issues, remember to factor in how historically SST was calculated when you parse that).Carrickhttps://www.blogger.com/profile/03476050886656768837noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-62386439952685698352012-10-01T23:06:55.638+10:002012-10-01T23:06:55.638+10:00Very nice explanation. Thanks Nick.
Kevin CVery nice explanation. Thanks Nick.<br />Kevin CAnonymousnoreply@blogger.com