date: Wed Mar 4 15:43:31 2009 from: Phil Jones subject: Fwd: Re: Fwd: Re: Canada to: John Kennedy John, See below. Best to set all 1991-2008 to missing for 718260. I'll do that in my file. I'll then stop the numbers going in from now on. It seems as though Canada has used an earlier code for a newish station on Baffin Island. Cheers Phil Date: Wed, 4 Mar 2009 15:27:50 +0000 (GMT) From: David Lister To: Phil Jones Subject: Re: Fwd: Re: Canada Phil, I have had a look at the station series 718260. It is one that has been coming via CLIMAT - probably resuming in October 2002. It does now appear to be PANGNIRTUNG (66.15N, 65.73E and 23m) which is on Baffin Island. CRU records firmly believe it to be NITCHEQUON (53.2N, 70.9E and 536m). CRU are not the only ones to associate 718260 with NITCHEQUON (Quebec). The discontinuity in Tmean (most apparent in summer) (see attached series) does suggest that the station 71826 via CLIMAT is now PANGNIRTUNG. Given that the series is very intermittent via CLIMAT and does not match the earlier part, it seems that we should delete its modern subset. The series from Lucie Vincent do not include either location. In addition, the WMO conversions that you have built into the updating process (from CLIMAT so that stations under new WMOs get to their correct destination) does not seem to be implicated in the current problem. There is a long listing of North American stations/codes etc. at: [1]http://www.weathergraphics.com/identifiers/master-station.dat Cheers David On Wed, 4 Mar 2009, Phil Jones wrote: David, When you've a moment can you have a look at this station - 71826. Cheers Phil X-IronPort-AV: E=Sophos;i="4.38,300,1233532800"; d="scan'208";a="3186121" Subject: Re: Canada From: John Kennedy To: Phil Jones Date: Wed, 04 Mar 2009 11:06:23 +0000 X-Mailer: Evolution 2.0.2 (2.0.2-35.0.4.el4_6.1) X-Canit-CHI2: 0.00 X-Bayes-Prob: 0.0001 (Score 0, tokens from: @@RPTN, f028) X-Spam-Score: 0.00 () [Hold at 5.00] SPF(none,0) X-CanItPRO-Stream: UEA:f028 (inherits from UEA:default,base:default) X-Canit-Stats-ID: 18448918 - 899a4dc58909 (trained as not-spam) X-Antispam-Training-Forget: [2]https://canit.uea.ac.uk/b.php?i=18448918&m=899a4dc58909&c=f X-Antispam-Training-Nonspam: [3]https://canit.uea.ac.uk/b.php?i=18448918&m=899a4dc58909&c=n X-Antispam-Training-Spam: [4]https://canit.uea.ac.uk/b.php?i=18448918&m=899a4dc58909&c=s X-Scanned-By: CanIt (www . roaringpenguin . com) on 139.222.131.184 Thanks Phil, The Canadian station, which has an alarming discontinuity in its temperature series, is 718260. Apparently the station associated with the ID is now in a different place. Cheers, John On Tue, 2009-03-03 at 12:28 +0000, Phil Jones wrote: > > John, > > Attached is a file I use when updating the anders file. > Amongst other things > it has a list of 42 pairs of Canadian IDs. The left number is the > one the data has > in the file, while the second is the number Canada is currently > using for that station. > > If you were to implement something similar in your routine > updating you would > get up to an additional 42 sites with current data for Canada. The > 42 sites are in the > file, so the file size won't change. > > I never could figure out how Canada numbered its stations. It's > the one country > where it seems totally random! > > I assume you have the other attached file. For these Canadian > stations a small > temperature is added to these sites each time a new value comes in. > These numbers > were supplied by Lucie Vincent. This is because her homogeneity > work adjusted to > the method of calculating mean T in use in Canada in the 1950s. The > current method > is different. The difference this makes is only important in eastern Canada. > > All these have been taken into account by David Lister in the > anders file that went > through to the end of 2008, that I gave you last week. > > Earlier today we got an email from Australia - see below. > > So, Australia is still issuing the wrong CLIMATs as far mean T is concerned. > > In the files I gave you last week, all Australian data post-Nov94 has mean > T calculated the way it should be using (Tx+Tn)/2. Using the correct data > warms Australia as a whole by +0.15C compared to what is released. > > Cheers > Phil > > > ---------- Forwarded message ---------- > Date: Tue, 3 Mar 2009 18:14:17 +1100 > From: Blair Trewin > To: David Jones , David Lister > Subject: RE: FW: Australian temperature data [SEC=UNCLASSIFIED] > > I've finally had a chance to have a look at this - it turned out to be > more complicated than I thought because a change which I thought had > been implemented several years ago wasn't. > > Up until 1994 CLIMAT mean temperatures for Australia used (Tx+Tn)/2. In > 1994, apparently as part of a shift to generating CLIMAT messages > automatically from what was then the new database (previously they were > calculated on-station), a change was made to calculating as the mean of > all available three-hourly observations (apparently without regard to > data completeness, which made for some interesting results in a couple > of months when one station wasn't staffed overnight). > > What was supposed to happen (once we noticed this problem in 2003 or > thereabouts) was that we were going to revert to (tx+Tn)/2, for > historical consistency, and resend values from the 1994-2003 period. I > have, however, discovered that the reversion never happened. > > In a 2004 paper I found that using the mean of all three-hourly > observations rather than (Tx+Tn)/2 produced a bias of approximately > -0.15 C in mean temperatures averaged over Australia (at individual > stations the bias is quite station-specific, being a function of the > position of stations (and local sunrise/sunset times) within their time > zone. > > Blair Trewin > National Climate Centre > Bureau of Meteorology > > > > > Prof. Phil Jones > Climatic Research Unit Telephone +44 (0) 1603 592090 > School of Environmental Sciences Fax +44 (0) 1603 507784 > University of East Anglia > Norwich Email p.jones@uea.ac.uk > NR4 7TJ > UK > ---------------------------------------------------------------------------- -- John Kennedy Climate Monitoring and Research Scientist Met Office Hadley Centre FitzRoy Road Exeter EX1 3PB Tel: +44 (0)1392 885105 Fax: +44 (0)1392 885681 E-mail: john.kennedy@metoffice.gov.uk [5]http://www.metoffice.gov.uk Global climate data sets are available from [6]http://www.hadobs.org Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ---------------------------------------------------------------------------- Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ----------------------------------------------------------------------------