RE: Leap seconds

From: Ted Molczan (
Date: Sun Jun 30 2002 - 09:48:09 EDT

  • Next message: "RE: Leap seconds"

    Russell Eberst asked:
    > Whatever happened to 'leap seconds'? They were designed to be 
    > inserted at the end of December or June (today) to allow the 
    > Earth to catch up with a regular time standard. However, 
    > there hasn't been a leap second announced for many months.
    I had not noticed this, but now that Russell has mentioned it, a long
    time does seem to have passed since the last leap second was inserted. I
    know little about leap seconds, but it turns out that there is quite a
    bit of information on the web.
    I have learned that by international agreement, the decision whether or
    not to insert a leap second is made by the organization International
    Earth Rotation Service (IERS). Its web site provides lots of
    The relevant information is found among the four standard bulletins
    issued by the IERS, which can be accessed here:
    Bulletin C "contains announcements of the leap seconds in UTC".
    Following the URLs leads me to:
    The last Bulletin C, issued on 2002 Jan 14, announces that no leap
    second will be introduced at the end of June 2002:
    The history of UTC offsets, provided at the following URL, reports that
    the last leap second insertion was on 1998 Dec 31:
    Now, to answer Russell's question, I went back to the bulletins page,
    and found that Bulletin A "contains rapid determinations for earth
    orientation parameters". This sounded promising, so I followed the URLs
    The URL labelled, "IERS Bulletin A -- Rapid Service/Prediction of Earth
    Orientation" leads to:
    The table of COMBINED EARTH ORIENTATION PARAMETERS informs that on 2002
    Jun 27, UT1-UTC = -.228762 s. Since the purpose of the leap second is to
    maintain the difference between UT1 and UTC to within 0.9 s, it is now
    clear that a leap second will not be issued because it is not necessary.
    Judging by the table of PREDICTIONS on the same page, it appears that
    UT1-UTC will be -0.37118 s one year from now, so no leap second is
    likely to be required through at least Jun 2003.
    How is this relevant to visual satellite observation? By convention, the
    time of positional observations is reported as UTC. Experienced
    observers regularly achieve timing accuracy of 0.1 s or better. The
    analysts who produce updated orbits from these observations find that
    errors of 1 s tend to stick out like a sore thumb. Depending upon the
    circumstances, errors that large may result in the observation being
    assigned a rather low statistical weight or being rejected outright.
    To an analyst attempting to fit an orbit to a set of observations
    spanning 30 Jun or 31 Dec, the insertion of a leap second on those dates
    results in an apparent 1 second error in the UTC positional timings,
    when comparing observations made before and after the insertion.
    A practical solution is to subtract 1 s from the observations made prior
    to the insertion of the leap second. This eliminates the 1 s error, and
    results in an epoch that is accurate going forward from the date of the
    leap second insertion.
    Here is an exercise. The first elset was issued about 19 h before the
    leap second insertion at the end of 1997 Jun 30; the second one about
    three after the insertion:
    1 00271U 62010A   97181.18205596 +.00000006 +00000-0 +10000-3 0 09014
    2 00271 086.6536 054.4864 0329540 060.7273 302.6159 09.41297471210592
    1 00271U 62010A   97182.13850598 +.00000006 +00000-0 +10000-3 0 09063
    2 00271 086.6543 054.3470 0329535 059.5568 303.7463 09.41297490210680
    Compare predictions made using both elsets on 1997 Jun 30 and 1997 Jul
    Ted Molczan
    Unsubscribe from SeeSat-L by sending a message with 'unsubscribe'
    in the SUBJECT to

    This archive was generated by hypermail 2b29 : Sun Jun 30 2002 - 09:53:08 EDT