Leap SecondsDennis McCarthyU. S. Naval Observatory
Definition of SecondsRotational Second1 / 86,400 of mean solar dayEphemeris SecondFirst used in 19561/31,556,925.9747 of tropical year 1900Length of year based on 19th century astronomical observations
Atomic SecondSI second: 9,192,631,770 periods of the radiation corresponding to the transition between 2 hyperfine levels of the ground state of the Cesium 133 atom (adopted 1964)Realizes the Ephemeris SecondFrequency based on lunar observations from1954.25 to 1958.25SI second preserves the rotational second of mid-nineteenth century
Time ScalesRotationalUT1 is modern realization of historical astronomical time scales includingMean Solar TimeGreenwich Mean TimeGreenwich Civil TimeUniversal Time (without suffixes)Weltzeit
Time Scales (continued)AtomicTAI (International Atomic Time)Follow-on from A.1 (maintained at USNO with input from 9 other laboratories originally. - now only USNO)AM (at BIH with input from many laboratories)A3at BIH with input from 3 best laboratoriesbecame AT (or TA) in 1969, TAI in 1971othersAll atomic time scales were made equal to UT1 corrected for seasonal effects on 1 Jan 1958 0h 0m 0smay be considered modern realization of Ephemeris Time (offset in epoch)
Earth RotationWell documented decelerationTidalChange in figure
Historical AnswersUTC (Coordinated Universal Time)Begun in 1960 as cooperative effort of U.S. Naval Observatory and Royal Greenwich Observatory to make coordinated changes to clocksin 1965 BIH defined UTC with respect to atomic timeEpoch and frequency adjusted to match UT1 corrected for seasonal variationsCurrent UTC adopted beginning in 1972no changes in frequencyleap seconds so that |UTC-UT1| < 0.9 sUTC consistent with previous definitions of legal time
Causes for ConcernFrequency of leap seconds increasingIncreasing public annoyanceSoftware issuesUnpredictableContinuous second counts: days with 86,401 secondsTime stamping 23h 59m 60sCommunications problemscoordination of events during a leap secondGrowth of systems based on independent time scales
Things to ConsiderNavigation1 second = 1/4 mile at the equatorComputer softwareContinuous second counts? 61-second minute?CommunicationsMaintain synchronization over the leap second?Legal definitionsMean solar time? Religious observancesSunrise, noon, sunset?
OptionsStatus quoDiscontinue leap secondsUse TAIRe-define secondIncrease tolerance for |UTC-UT1|Smooth over the leap second stepPredictable periodic adjustment of UTCConventional adjustment of UTCEvery leap year? Every 10 years?Predict leap seconds based on deceleration model
Status QuoPro:No changes requiredMinimize concerns of celestial navigatorsCon:Frequency of leap seconds increasingCommunications, software problemsGrowth of systems based on independent time scales
Discontinue Leap SecondsPro:Eliminate causes for concern Con:Unlimited growth of |UTC-UT1|Legal definitions of time?
Use TAIPro:Eliminate causes for concern Con:TAI must be more accessibleLegal definitions of time?
Similar to elimination of leap seconds TAI must be more accessible
Redefine the SecondProFundamental Solution ConRequire redefinition of physical unitsTemporary solution
Smooth Over Leap Second StepProEliminates the extra secondConRequires seconds of different lengthsDate of adjustment unpredictableImplementation?
Increase Tolerance for |UTC-UT1|ProEasy to accomplish ConLarger discontinuitiesDUT1 code limitationsDate of adjustment unpredictableWhat is an acceptable limit?
Periodic Adjustment of UTCProDate of adjustment is predictable ConNumber of leap seconds remains unpredictableLarge discontinuities
Conventional Adjustment of UTCUnknown number of leap seconds at predictable intervals Known number of leap seconds at predictable intervals
Conventional Adjustment of UTC
What to do?Question needs studyURSIIAUITU-R