tag:blogger.com,1999:blog-7729093380675162051.post1813008250086423225..comments2024-03-28T13:56:47.604+11:00Comments on moyhu: Why is this June hotter seen with GHCN V4 than V3? - and lots of active graphics.Nick Stokeshttp://www.blogger.com/profile/06377413236983002873noreply@blogger.comBlogger14125tag:blogger.com,1999:blog-7729093380675162051.post-8309350693506815052019-08-01T18:48:52.640+10:002019-08-01T18:48:52.640+10:00Nic,
As far as I know, only GISS v3 uses the recon...Nic,<br />As far as I know, only GISS v3 uses the reconstructed Byrd series. GISS V3 gets the Antarctic data through SCAR.<br />The reconstructed Byrd is not included in GHCNv3, GHCNv4, GISSv4, or BEST.<br />Here is BEST's own Byrd "reconstruction" with gaps:<br />http://berkeleyearth.lbl.gov/stations/166906<br />Olof Rhttps://www.blogger.com/profile/18244733455655978307noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-26002897368332048972019-08-01T13:37:12.959+10:002019-08-01T13:37:12.959+10:00"I get the V4 result. The underlying Tav valu..."I get the V4 result. The underlying Tav values in V4 are different to those in V3. That is the problem!<br /><br /> Err that is how pHa works.<br /><br /> V4 has more stations than V3. When you do pairwise homogenization you have to collect<br /> all the highly correlated neighbors. A station in v4 will have more neighbors than that<br /> same station in v3.<br /><br /> Next, the SNHT process will identify those stations and their points in time when<br /> they disagree with thie neighbors. Increase the neighbors and these points will change.stevenhttps://www.blogger.com/profile/06920897530071011399noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-37112749979050650752019-08-01T08:19:40.837+10:002019-08-01T08:19:40.837+10:00Thanks, Nic
That gives me a lot to work with, I...Thanks, Nic<br />That gives me a lot to work with, I'll see what I can do.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-14160456508023521292019-08-01T07:26:38.057+10:002019-08-01T07:26:38.057+10:00Yes - I realised immediately and blogger amazingly...Yes - I realised immediately and blogger amazingly gave me a delete option !<br /><br />You can put it back ;-)Clive Besthttps://www.blogger.com/profile/10486120708699060846noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-43720131901228866382019-08-01T05:42:41.521+10:002019-08-01T05:42:41.521+10:00Thanks, Nick
I clicked on Ctrl shift K in Firefox ...Thanks, Nick<br />I clicked on Ctrl shift K in Firefox and theis non-Google error shows first:<br />‘src’ attribute of [script] element is empty. why-is-this-june-hotter-seen-with-ghcn.html:612:1<br /><br />In Chrome, the first error message in the Console on invoking Ctrl shift I is:<br />why-is-this-june-hotter-seen-with-ghcn.html:875 [Violation] Parser was blocked due to document.write([script])<br /><br />Hope this helps. Note that I'm now finding displaying the main graphic at all, in both Firefox and Chrome. It works sometimes but not others (possibly a SSL or TLS error - I'm finding that pages using encryption quite often don't load first time?), and not at all for the linked page. The error messages I get in Firefox for that web page are:<br />Synchronous XMLHttpRequest on the main thread is deprecated because of its detrimental effects to the end user's experience. For more help http://xhr.spec.whatwg.org/ ghcn0.js line 3 > eval:1:3349<br />Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at https://s3-us-west-1.amazonaws.com/www.moyhu.org/data/ghcn/new/anom2019_6.js. (Reason: CORS request did not succeed).<br />‘src’ attribute of [script] element is empty.<br />Nic<br /><br /><br /><br />Nic Lewishttp://nicholaslewis.orgnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-91653566234806758822019-08-01T03:48:31.403+10:002019-08-01T03:48:31.403+10:00Thanks, Nic.
Sorry about the troubles with the ch...Thanks, Nic. <br />Sorry about the troubles with the checkboxes. I could try to diagnose that - if after a failure in Firefox you click on Ctrl Shift K, you should see a debug window, and the console tab will bring up a list of error messages. Ctrl shift I is similar in Chrome, and may make the error clearer. There will be junk from the various hitches with the Google system, but probably the top message will be the key one. It should give the line number where the fault is.<br /><br />I don't know the details about Byrd etc, but V4 does have more stations. I have another graphic <a href="https://moyhu.blogspot.com/p/blog-page_24.html" rel="nofollow">here</a> in similar style which shows the V4 stations, this time with a triangle mesh. On this version you can click on station locations for more details.<br /><br />Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-29521331594164849732019-08-01T03:29:26.340+10:002019-08-01T03:29:26.340+10:00Clive,
It didn't seem to me that you did quit...Clive, <br />It didn't seem to me that you did quite get the V3 result. It's close recently, but not so much further back. And I wonder about whether the subsets match. If more stations from the V4 subset of the 3500 report than of the V3 subset, there is still a version of the discrepancy here.<br /><br />I'm planning a test that you could probably do too. Suppose you have done a V4 mesh analysis; you have nodes Z4, anomalies A4, with mesh M4 and weights W4. And from V3 also Z3, A3 etc. Then with M4, interpolate A4 onto Z3, with result X. Then X.W3 should match the V3 average A3.W3, rather than the V4 average A4.W4. Maybe you don't explicitly use weights; then it would be the corresponding integration procedure.<br /><br />Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-30124105535941425432019-08-01T02:59:42.929+10:002019-08-01T02:59:42.929+10:00Clive,
Are you looking at it with the nodes switch...Clive,<br />Are you looking at it with the nodes switched off? The initial picture is red because of node density.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-59642386419702021452019-08-01T02:48:58.899+10:002019-08-01T02:48:58.899+10:00This comment has been removed by the author.Clive Besthttps://www.blogger.com/profile/10486120708699060846noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-21767257816697117612019-08-01T02:11:49.897+10:002019-08-01T02:11:49.897+10:00Interesting Nick,
You are right of course that wh...Interesting Nick,<br /><br />You are right of course that where there are sparse stations near the coast then the triangulation will pick up one land station and two SST points as vertices. These differences mean that spatial detail will be better in V4 than V3. However, I am more concerned about systematic adjustments made to the underlying station data.<br /><br />To check this I restricted the calculation of the monthly and annual temperature of V4 to only their versions of V3 stations. If the station data were the same then I should get back the V3 result. I don't - I get the V4 result. The underlying Tav values in V4 are different to those in V3. That is the problem!<br /><br />http://clivebest.com/blog/?p=8981<br /><br /><br />Clive Besthttps://www.blogger.com/profile/10486120708699060846noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-61782054794198099482019-07-31T22:10:53.067+10:002019-07-31T22:10:53.067+10:00 South of 60?
V4 has 95 stations.
B... South of 60?<br /><br /> V4 has 95 stations.<br /> Byrd is two different stations in V4<br /> <br /> As you note there is no continuoius record. Dont need them.<br /><br /> Not sure when we make the change over to V4, need to ask Robert.<br /><br /><br /> <br />stevenhttps://www.blogger.com/profile/06920897530071011399noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-11144754025608013822019-07-31T20:51:45.536+10:002019-07-31T20:51:45.536+10:00Excellent post, Nick. The graphics are impressive,...Excellent post, Nick. The graphics are impressive, although with LOESS switched off nothing at all appears on the map that I see: the station locations are missing, however I toggle the V3nodes, V4nodes and SST buttons. I've tried in both Firefox and Chrome. <br /><br />I'm surprised that v4 shows much higher warming in West Antarctica than v3. There are only a very limited number of stations there and their records have been readily available for many years. Moreover, there is essentially no continuous record available in the interior of W Antartica that stretches back to before 1980. Do you know if GHCN v4 has followed GISS and BEST in using the unhomogenised Bromwich et al composite reconstruction at Byrd, which basically stiches together the manned Byrd station record that covers 1957 to the early 1970s with the 1980 on automatic Byrd station record, with no offset despite their differing location, construction and local environment? That reconstruction shows a high warming trend, but its method is inconsistent with normal homogenisation approaches (and is in direct conflict with BEST's 'scalpel' method).Nic Lewishttp://nicholaslewis.orgnoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-8963493332674946572019-07-31T19:54:13.926+10:002019-07-31T19:54:13.926+10:00Thanks, Steven
I wrongly thought they took in TOBS...Thanks, Steven<br />I wrongly thought they took in TOBS-adjusted data. I'll correctNick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-88934230129839205672019-07-31T19:40:39.307+10:002019-07-31T19:40:39.307+10:00technically V4 does not do a discrete TOBS adju...technically V4 does not do a discrete TOBS adjustment. TOBS only gets done for USHCN.<br />in v4, since its global and since there is not good global metadata on TOBS, Pairwise Homogenization<br />is the only method used.<br /><br />stevenhttps://www.blogger.com/profile/06920897530071011399noreply@blogger.com