date: Fri Jun 18 10:28:50 2004 from: Phil Jones subject: Re: Punta Arenas MSLP question to: "Gareth Marshall" Gareth, Faraday's drift is getting a little worse. Could this be that the Ukrainians brought in a new barometer when they took over? The drift here is less clear than at Punta Arenas, but you argue it starts at about the same time in 1990. Equally well you could say it began in the late 1990s. Both recent trends look like barometer drifts, unless there is something subtle with ERA-40. I've been doing some work with Adrian Simmonds at ECMWF. We're looking at temperature from ERA-40 and the CRUTEM2v dataset. I'll attach a draft paper (as pdf) that will come out as an ECMWF report. As you know ERA-40 assimilated the station temperatures and the pressure values, whereas NCEP only took in the pressure. Also ECMWF improved the Antarctic synops by getting them from the READER website whereas NCEP used what they had. I'll probably show a couple of plots from this work at the meeting at BAS in September. It might be worth sending these two plots to Adrian (Adrian Simmons ). I believe he's travelling a bit at the moment. He's off to the CLIVAR meeting in Baltimore. I'll be seeing him at a meeting in Geneva in mid-August. Cheers Phil At 15:17 17/06/2004 +0100, you wrote: Phil Thanks for your reply. I have looked at validating ERA-40 around Antarctica as a matter of course, for any studies that I'm using it for. As, I mentioned in my email, generally things improve in 1974 and then once more in 1979 (see attached Faraday plot). You are correct in that ERA-40 values are too high early on and then too low. I hadn't bothered to check the MSLP-STP differences before contacting you but clearly the shift is less than 5 hPa so it doesn't seem to be a switch from one to the other. Also I've done a plot for Rothera and there is a similar jump when we switched over to an electrical barometer and also when the station changed height and depie us telling the WMO they still have the old height ... Having done a little investigation I'm inclined to agree with you about the 2001/11 value being surface rather than MSLP; we have a CLIMAT message for this month with an alleged surface pressure of 997.8 and MSLP of 1003.7. The ERA-40 value is 997.5. We don't have many CLIMATS/SYNOPS from Punta but the two pressure values are often the wrong way around so it wouldn't surprised me if some mistake was made in this case. You make an interesting comment about the drift in the bias possibly being down to a changing barometer. Do you think this might be the case with Faraday, which as you can see is getting worse. Cheers Gareth >>> Phil Jones 06/17/04 12:54pm >>> Gareth, Thanks for the plot. I'd thought of doing something like you're doing but haven't yet had the time. Are you doing this for all Antarctic sites (those around the coast) and also the islands south of 40S and the southern tip of S. America? As for Punta Arenas, your plot difference is about 3hPa from late-1989 onwards. I've looked at some months of the CLIMAT messages for 85934 but the MSLP-STP is between 5.7 and 5.9hPa for the months in 2002 and 2003. So, a simple switch of the two pressure averages isn't going to make things better. It might make things worse depending on which way round you've plotted the differences. You say the ERA-40 bias, so I'm guessing that before the break ERA-40 was too high, then too low afterwards. I say this in a relative sense, not meaning ERA-40 is wrong. There are 4 outliers post 1986. The two in 1992 look suspiciously high at 85934. The last month in your plot (Dec 2001) could be an STP value. There is one odd value in 1987 as well. Also in your plot, the difference with ERA-40 closes towards zero suggesting the barometer is gradually improving. Me best guess is they got a new barometer in 1989, or altered the adjustment to SL in some way. Also, a site move is possible but if done well MSLP shouldn't be affected. Have you tried the two stations at Ushuaia or Rio Gallegos? Do they show something similar? They will have more missing data making assessment more difficult. Stanley might help as well. I used Punta Arenas in this paper Jones, P.D., Salinger, M.J. and Mullan, A.B., 1999: Extratropical circulation indices in the Southern Hemisphere based on station data. Int. J. Climatol. 19, 1301-1317. to derive a N/S gradient with Stanley to try and explain southern S. American temp. trends. I also used it in the earlier paper in 1991. Cheers Phil At 11:22 17/06/2004 +0100, you wrote: >Phil > >I have attached a figure showing the bias in MSLP in the ERA-40 reanalysis >at Punta Arenas. It shows the usual high southern latitude improvements in >1974 with the advent of VTPR assimilation and again in 1979 with TOVS. >However, you can also see a clear step change in 1989-1990. I am not aware >of any significant changes happening in the reanalysis at this time so it >looks like (i) station move or (ii) the data before the step are actually >surface pressure rather than MSLP (I think the station height is 37 m). >I'm letting you know as the observations are some you sent me. > >Cheers > >Gareth > 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 ----------------------------------------------------------------------------