cc: m.manton@uea.ac.uk, david.parker@metoffice.gov.uk
date: Tue Nov 16 16:47:48 2004
from: Phil Jones <p.jones@uea.ac.uk>
subject: Re: Fwd: Re: Mean Monthly Temperature
to: "Hans Teunissen" <HTeunissen@wmo.int>, "Greg Hammer" <gregory.r.hammer@noaa.gov>, <Thomas.C.Peterson@noaa.gov>, "H Kontongomde" <HKontongomde@wmo.int>, "Richard Thigpen" <RThigpen@wmo.int>

    Dear All,
        Encouraging Malta to send its GSN data for Luqa (the airport) to Greg is fine. Perhaps
    at the next AOPC meeting we should consider how we should be responding to
    questions about how to calculate monthly averages. What is in the WMO Guide
    in 1983 may be correct, but if you look in the WWR volumes for 1951-90 (all 4) you
    will see that Malta has calculated its monthly mean pressures based on 4 synoptic
    hours and the mean temperature based on the max and min.
        So the advice to change will introduce a homogeneity into the Luqa record. The
    example given is fine, but other equally plausible possibilities could be conjured up.
    The whole question is are we after a consistent long record or the best estimate of the
    true temperature. I would argue for the former. Whether the method gives a better estimate
    of the truth is not the issue - long-term homogeneity should be paramount. I can easily
    show that anomalies (from a common base period) derived from these two methods
    (mean of 4 or 8 synoptic hours and the mean of min/max) will give virtually
   indistinguishable
    different trends. There will be absolute differences though when the change is made and
    this will screw up the homogeneity.
       This same issue came up with the Antarctic stations in the last teleconference call.
   Can
    we discuss what advice we should be giving? Can this Hans be added to our agenda for
    April? We will likely have to discuss this with other commissions. As far as I believe,
    most inhomogeneities in station temperature time series are due to two things:
      - site moves
      - changes in the way monthly mean temperatures are calculated
     in a roughly equal percentage.
     So, the advice being given for Luqa WILL introduce an inhomogeneity.  Tom's software is
    so good it will spot it but only after about 5 years !
      Perhaps we need a Statement of Guidance on calculating daily and monthly averages.
    Why am I saying this as you'll only say go ahead and write it, but at the moment we seem
    to be giving the wrong advice.  I hope GSN is not getting countries to think too deeply
   into
    their data. They need better advice if they do. Maybe by April we can see if Malta has
   sent
    anything !
    Cheers
    Phil
   At 16:03 16/11/2004, Hans Teunissen wrote:

     Dear All:

     The original corrspondence that was the basis for the exchange I just sent you (some as
     bc's), FYI.

     Hans.

     =================================================================
     Dr. Hans W. Teunissen                                Tel: +41.22.730.8086
     Global Climate Observing System (GCOS)     Fax: +41.22.730.8052
     c/o World Meteorological Organization         E-mail: [1]HTeunissen@wmo.int
     7 bis, Ave. de la Paix
     CP 2300, CH-1211 Geneva 2
     Switzerland
     =================================================================
     Date: Tue, 16 Nov 2004 13:53:56 +0100
     From: "Alan Thomas" <AThomas@wmo.int>
     To: "Hans Teunissen" <HTeunissen@wmo.int>
     Subject: Fwd: Re: Mean Monthly Temperature
     Mime-Version: 1.0
     Content-Type: text/plain; charset=US-ASCII
     Content-Transfer-Encoding: 7bit
     Content-Disposition: inline
     Dr Alan R Thomas
     Director, GCOS Secretariat
     c/o WMO
     Case postale 2300
     7bis, Avenue de la Paix
     CH-1211 Geneva 2
     41-22-730-8275
     41-22-730-8052 (fax)
     >>> Alexander Karpov 16/11/2004 1:42:10 PM >>>
     Dear Mr Porter,
     Thank you very much for your message related to the  procedure for MMT
     calculations.  Accordingly, the secretariat had approached an expert in
     this field and the principle Author of "handbook on CLIMAT and CLIMAT
     TEMP Reporting", Dr Oleg Alduchov (World Data Centre B, Russian
     Federation) to provide comments on your query.  Based on his input, we
     would like to inform you on  the following.
     1. Unfortunately, it should be noted that changes in the estimation of
     daily mean values introduced in your Service in January 1995 were not in
     compliance with the WMO GUIDE TO CLIMATOLOGICAL PRACTICES (WMO-No. 100),
     1983, second edition.  The Guide is also accessible through
     [2]http://www.wmo.ch/web/wcp/wcdmp/home.html. You may also wish to note
     an appropriate excerpt from Chapter 8 of this Guide - CLIMAT REPORTS,
     which reads as follows:
     "For the mean monthly pressure, be it station pressure or standard
     level pressure, it is advisable to use a true mean or a corrected value
     to correspond to a mean based on 24 direct observations a day. The mean
     of four equally spaced observations may suffice as a close approximation
     to the true mean."
     ..........................................
     For monthly mean temperature the same remarks apply as for computation
     of monthly mean pressure."
     It is understood that the same rule has to be applied for the
     computation of all other monthly mean values.  It should be also
     mentioned that the use of 8 equally spaced observations gives slightly
     better estimation of "true daily mean".
     2. As regards the current practice of your service - half of a sum of
     minimal and maximal temperatures - for daily mean values, it might give
     a good estimation of monthly mean value only for the periods without any
     significant changes of weather and "standard/normal" sinus-like
     changes of observed parameters.
     3. To illustrate the above, one could assume that during 21 hours we
     had a very stable temperature 25 C, followed by a significant fall of
     the temperature down to 18 C during last 3 hours.  According to  current
     practice you mentioned, the calculation of the daily mean value would
     give 21.5 C.  However, the WMO-recommended procedure gives the value of
     23.25 C.  Naturally, in this case the last value would better represent
     the daily mean temperature.  More generally, for many stations it is
     proved that anomalies for minimal and for maximal temperatures have
     different statistical properties. Therefore, the daily mean values based
     of the half sum of minimal and maximal temperatures often provide a
     systematical bias with respect to the "real daily mean temperature".
     Actually, the same is correct for many other observed meteorological
     parameters.
     4. In addition to the above, please be informed that the third edition
     of the WMO GUIDE TO CLIMATOLOGICAL PRACTICES is now in preparation and
     will comprise some new developments and changes.  However, it would be
     advisable to follow the guidelines laid down in the current version
     until the third edition will be published.
      I trust that the above information would  help you to meet stated
     requirements in compiling climatologcal reports.  Should you have any
     other specific questions, do not hesitate to contact the secretariat and
     directly Dr O.Alduchov (His address is aoa@meteo.ru).
     Yours sincerely,
     A. Karpov
     >>> <saviour.porter@maltairport.com> 12.11.2004 10:44:09 >>>
     Dear Mr. Karpov,
     1. We have just received WMO/TD No.1188 and have noted that when
     computing
     the Mean Monthly temperature, the values for 00Z 06Z 12Z and 18Z should
     be
     taken into consideration. These should be added up, and then divided by
     4
     -  this becomes the mean temperature for the day and so on until the
     end of
     the month, -  the mean temperature for the month is then computed from
     these daily values.
     2. However my predecessors, since at lease January 1995, have computed
     the
     mean daily temperature differently - namely they found the maximum
     temperature for the day, added to it the minimum temperature for the
     night
     and divided this by 2, and they called this the mean temperature for
     the
     day. The values for the Standard Deviation is also computed from these
     values. We have inherited from them this procedure - however it gives
     quite
     a different value to the procedure described in my previous paragraph
     (para.1) . One of these systems gives large errors.
     3. I would be most grateful to you if you could verify which system
     should
     be used - and if this is the system outlined in para 1 would you
     please
     tell us what should be done to rectify matters.
     Yours sincerely
     Saviour Porter
     Permanent Representative of Malta with WMO
     Chief Meteorological Officer
     Malta International Airport plc
     Tel: (+356) 2369 6527
     Fax: (+356) 2124 6694
     MIA Weather Website : [3]http://www.maltairport.com/weather

   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
   ----------------------------------------------------------------------------

