tag:blogger.com,1999:blog-7729093380675162051.post5926955031320972523..comments2024-03-28T13:56:47.604+11:00Comments on moyhu: Updating GHCN - Stations aren't "dying"Nick Stokeshttp://www.blogger.com/profile/06377413236983002873noreply@blogger.comBlogger10125tag:blogger.com,1999:blog-7729093380675162051.post-89375363617119306122010-02-16T13:03:24.775+11:002010-02-16T13:03:24.775+11:00That could be.
So that tardy China block is being...That could be.<br /><br />So that tardy China block is being (maybe) updated each month, just trailing by a few months? Bizarre. I've been intrigued by these split-country groups. Algeria is also weird, with a group that's up to date, and a group that isn't.<br /><br />I noticed a poor little station in Tibet was straggling on its own. Lhasa, I think.carrot eaternoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-8386237016766971062010-02-16T12:38:03.980+11:002010-02-16T12:38:03.980+11:00Actually, I do have a v2.mean dated on my system 9...Actually, I do have a v2.mean dated on my system 9/Dec, and a v2.max from July. I'm trying to devise a script to show automatically which files were updated over the period and by how much.<br /><br />I speculate that some of the block delays are QC, People may be so nervous now that if one station report needs investigation, a whole batch is held up pending resolution.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-20239327129641122252010-02-16T11:55:01.579+11:002010-02-16T11:55:01.579+11:00If only we can dig up a v2.mean from sometime in l...If only we can dig up a v2.mean from sometime in last December. <br /><br />From the above, it looks like about 1200 stations are reliable reporters each month. Taking a very naive count, we might expect (1252-1182) to show up in the next week or so, or 70 more stations. I'm assuming that the number lost to QC is about the same in any given month.<br /><br />The following are big chunks missing from Dec to Jan, for a total of 90:<br />North Korea, for 7.<br />Mongolia, for 34.<br />Paraguay, for 10.<br />Peru, for 12.<br />France, for 17.<br />Greece, for 10.<br /><br />I wonder if one of the above countries then isn't actually a reliable reporter, but happened to catch up on the backlog at the end of the year.carrot eaternoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-85649255379131595332010-02-16T08:45:28.639+11:002010-02-16T08:45:28.639+11:00Zeke,
I've always pictured a poor little grad ...Zeke,<br />I've always pictured a poor little grad student slaving away in the basement, looking up old written records from 1895 and digitising them, or finding keystroke errors. I think the effect of any corrections will be more noticeable in 1895 than 1995, because of the density of data.<br />__<br /><br />I took a peek at the GISS code, and if USHCN and GHCN both have data for the same station, it uses the USHCN version. Which means I gave up too quickly when trying to compare the station numbers, it must be easy to do...carrot eaternoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-28600533124475435512010-02-16T07:43:22.702+11:002010-02-16T07:43:22.702+11:00This also helps explain why various surface temp r...This also helps explain why various surface temp records are adjusted post-facto, as additional stations report. Though I'm still somewhat baffled by the occasional tiny adjustment in, say, 1895 temps that happens in GISS :-pZekehttps://www.blogger.com/profile/09757819498566612533noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-60027700543720877972010-02-16T07:34:19.702+11:002010-02-16T07:34:19.702+11:00Well, after some random checks, the US stations th...Well, after some random checks, the US stations that were dropped from GHCN are indeed in USHCN (sensible), and the US stations that remain in GHCN are not in USHCN (why not?). I didn't find any that were dropped completely in that time span, but there are probably some. I couldn't quickly whip up a comparison script because I don't see any obvious correlation between the station IDs of the two sets.<br /><br />You can't use the USHCN alone because it isn't updated monthly. So GISS needs the US GHCN stations to do live updates each month. But the USHCN comes with TOB adjustments, which is important for the US record because it introduced a systematic bias. So you want to also use the USHCN when it does come out.<br /><br />Actually, now I'm curious. Both sets are oversampled for the US, but I'd think you'd get slightly different numbers from each, if USHCN v2.0 comes TOB-adjusted and pre-homogenised, and the GHCN is not. I wonder if GISS's US numbers change a bit whenever they add on the latest USHCN numbers. <br /><br />By the way, I don't think v2.0 uses metadata for anything but the TOB adjustment.carrot eaternoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-1997829670310285942010-02-16T06:11:40.500+11:002010-02-16T06:11:40.500+11:00EM Smith attributes (I think) much of the US reduc...EM Smith attributes (I think) much of the US reduction in recent years to GHCN's slowness in accepting USHCN v2. I haven't checked that myself, but the timing does seem to make that possible. But it may also be a recognition that the US was just over-represented, and if they need to economise their efforts, that's where they should do it.<br /><br />In the same way, I think you're right that GISS will pick up the missing stations with USHCN, which is one of the data sets they use. But it probably doesn't matter - even with the reduced GHCN numbers, the US is likely the best covered part of the globe. I suspect GISS likes USHCN not for the denser cover, but the better metadata.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-50213925483606659712010-02-16T05:52:53.604+11:002010-02-16T05:52:53.604+11:00I see.
Things that caught my eye: France, Greece...I see.<br /><br />Things that caught my eye: France, Greece, North Korea, Mongolia, Paraguay and Peru haven't been added at all yet, for Jan 2010. Once they come in, that'll change the numbers a lot.<br /><br />Libya just gave up reporting after Jan 09. They must have a backlog. There are other countries with total or partial backlogs.<br /><br />It looks like a lot of US stations died between 2004 and 2006, but I think most of these stations still appear in the USHCN files and therefore still make it into GISS. I've never cross-checked the USHCN and GHCN to see how many of them made it. Any thoughts on that?carrot eaternoreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-64107337422466364452010-02-16T05:08:17.541+11:002010-02-16T05:08:17.541+11:00Yes, that was my plan in writing the automated scr...Yes, that was my plan in writing the automated script. There'll be updates.Nick Stokeshttps://www.blogger.com/profile/06377413236983002873noreply@blogger.comtag:blogger.com,1999:blog-7729093380675162051.post-30168041698317760442010-02-16T00:19:29.374+11:002010-02-16T00:19:29.374+11:00I think the better way to illustrate the point wou...I think the better way to illustrate the point would be to sit there, download v2.mean every so often, and see how quickly the Jan 2010 reports get added in.<br /><br />The entire WUWT post was just ridiculous. Over the top language throughout, with no moment to pause for common sense. EM Smith should learn that 'blog first, think later' is a sure recipe for putting your foot in your mouth. This isn't the first time.carrot eaternoreply@blogger.com