r N79 21462 Rotation of the and Polar Motion," Services- Bernard Guinot Bureau International de 1'Heure 61, Avenue de 1'Observatoire, Paris 75014, France

Abstract. The need of a continuous monitoring solar attraction can be modeled, but others due of the polar motion appeared at the end of the to various local causes and to the plate motions 19th century, and was at the origin of one of the are not sufficiently known to correct the ob- oldest international projects : the establishment servations.] Figure 1 shows on an auxiliary of the International Latitude Service. This ser- sphere of unit radius the directions of the in- vice still operates, but other organizations now stantaneous P, of a reference fixed determine the polar motion, using the astrometric pole P0 and_pf the zenith Z of a station. The measurements of latitude and time and also origin of the astronomical longitudes is a fixed Doppler observations of artificial satellites. On point 0 on the of Po. We can see that the the other hand, since the advent of atomic clocks astronomical latitude cp and longitude L vary with in 1955, the universal time has become a measure the coordinates of the pole x and y. The univer- of the rotation of the Earth, which is also cur- sal time UTl is simply linked to the angular rently required and which must be evaluated by a motion in space of the PO meridian around P ; it Service. The services providing polar motion and is therefore dependent on x and y. The fundamen- universal time data will be described, the preci- tal equations used in classical services are, for sion and accuracy of these data will be estimated. each station i, UTC being the worldwide time reference, Introduction x(t)cosL0j. (1) The full description of the rotation of the Earth in space is traditionally given by the mo- [-x(t)sinL y(t)cosL 0,1.Itarf P 0,1. +[UT1-UTC] t tion of the rotation axis with respect to the ± - UTC]t , (2) Earth (polar motion), and in space (luni-solar , ), and by the angular posi- where

continuous monitoring. Except for the ILS, the computations of

The astrometric methods refer to the direc- tions of the plumb-lines of the observatories. We will assume that the Earth is rigid and that these directions are fixed within the Earth. [This is not true ; some motions due to the luni- Fig. 1. Variation of astronomical latitude and Proc. of the 9th GEOP Conference, An Ititennitiomtl SvmpiMiitni tni thr Applictiiion* of GiWi'.vv IH Gemkninim -s. Ocluher2-S, 1178. Depl. of Geodetic Science Rept. No. 280, The longitude with the coordinates of the pole. Ohio State Univ.. Columbus. Ohio 43210. 13 •T.'T-

Classical Astrometry Doppler

Visual zenith telescope Visual zenith telescope Doppler receivers for Transit instrument TRANSIT satellites (international Astrolabe stations) Photographic Zenith Tube 5 instr. 75 instr. 20 iristr.

X

\ f International International Bureau interna- Defense ^Mapping Latitude Service Polar Motion Serv. tional de 1'Heure Agency (USA) ILS IPMS BIH *~ DMA y V •**• J / 9 y x, y, UT1 x, y

Fig. 2. Data flow to the services.

- the way of removing some systematic errors, noise by the pair variance (or Allan variance) - the averaging time, ? ( I2 v ( T,) = mean ofU i+l " V - the smoothing techniques on observational data a « • and evaluated results. The Doppler determinations of polar motion is an example of the satellite techniques for the This function is represented by stability curves study of Earth rotation, which are discussed by (Fig. 3), as it is customary for the characteri- Aardoom [1978] at this Conference. Therefore we zation of the stability of oscillators [Barnes will only point out that in these techniques, et al., 1971J . Thus instead of speaking loosely instead of referring the observations directly of the precision of the results, we-can speak of to the quasi-non-rotating directions of stars, their stability. one uses intermediary objects, the directions of which are computed in the non-rotating reference In general, for small values oft, aa(t) frame. Providing that the motion of the object follows a law in l/\ft, as in the case of white can be correctly modeled, UT1, x and y can be noise, but for larger values oft, O"a ( ~C) derived from the observations. In practice, a reaches a minimum which is called the flicker a spurious drift of UT1 cannot be avoided, but the floor. For still larger values oft, a (~0) coordinates of the pole can be obtained to a generally increases. In this latter domain it is large extent free from systematic errors and sometimes difficult to make the distinction drifts. Although astrometry only measures angles, we will use the meter as the unit for polar motion, assuming that the radius of the auxiliary sphere X and Y is the polar radius of the Earth. Figure 2 shows the general organization of the services in 1978.

Precision and accuracy of the results

A matter of importance is the degree of confi- dence the user may have in the results published by the Services. This problem is not easily sol- ved. In most cases, the data are given without any information on their precision. In some cases a standard deviation is given, computed from the internal consistency of the data contributing to the determination of a raw value over a given averaging time r. But even when this information is given, it is far from being sufficient : it is 10 20 SO 100 5 10 well known that the errors are not a white noise days years and that the standard deviation does not vary proportionally to 1/tfr . Fig. 3. Precision of the pole coordinates as a Let us suppose .that AQ, A^,...,^ are measured function of the averaging timer, ^(r) is the quantities obtained at instants t0, to + T,..., pair variance. AST stands for global astrometric to + nt , by averaging over intervals T. If a0, services : IPMS and BIH special solution for as- a^,..., ajj are the random errors of these quanti- trometry alone. The current BIH results includ- ties, it is possible -to characterize the random ing DMA have about the same precision as DMA. between random and systematic errors. For which requires well calibrated micrometer screws. instance, for the Earth rotation parameters, the The stations, located on the 39°8'N parallel increase of °a(^) witht can be the consequence are now : of changes in the network of observing stations, changes of programs and methods in the stations, Mizusawa, Japan, longitude 141° E and/or plate motions. Kitab, URSS, 67° E In the following, the stability curves will be Carloforte, Italy. 8° E given. One must be aware that they represent an Gaithersburg, USA, 77° W estimation. For values of E smaller than a month, Ukiah USA, 125° W we can assume that the observational noise is larger than the true noise of the observed quan- Initially, the declination errors were re- tities themselves, and the use of a high-pass moved by the so-called chain method. Two or more filter gives fairly reliable values of the ran- groups of stars are observed every night. Assum- dom errors a. But for larger values of t> , one ing that the latitude does not vary during the has to make less reasonable assumptions. The night, the difference between group results re- "three-corner-hat method" is not available presents the contribution of declination errors. because there are only two truly independent As only night observations are possible, a full series of data, the astrometric and the Doppler. year is required for a complete evaluation of The accuracy of a series of results expresses the declinations, which permits to refer all the the degree of conformity with an adopted stan- observed latitude to the same standard before dard. In our case, the standard is not well solving equations (1). Next year, the cycle is defined. Let us take the motion of the pole. In resumed and the star positions can be improved. the case of astrometry, we only define the di- That explains that no definitive results can be rection of the pole relative to the direction of published, as long as the same stars are obser- the zeniths. On account of plate motions, the ved (normally during 6 to 12 years, because reference to the zeniths is vague. The satellite after some time precession makes the list of method refers the position of the pole to the stars obsolete). Earth. Similarly, the plate motions displace the This procedure was improved in 1922 by Kimura, stations and also the plate on which the pole who gave the ILS its present form. The Kimura z moves. Nevertheless, some types of errors can be term of equation (3) recognized, such as annual terms due to wrong positions of stars, or drifts due to erroneous x(t)cosL .+y(t)sinL +z(t) = 9 -

16 Bureau International de 1'Heure (BIH) for the year a are published towards June of year a + 1, in an Annual Report. In addition, the The BIH was created in 1912 and has been BIH operates a rapid service, giving with lower located at the Paris Observatory since that date. precision the results of week w on Thursday of Its present director is B. Guinot (since 1965). week w + 1, under a contract with the Jet The initial task of the BIH was to unify time Propulsion Laboratory (USA). by publishing the time of emission (in universal The BIH computations are devised in order to time or at that time mean solar time) of radio keep the spurious annual term as constant as time signals. With the improvement of time possible. The satellite data are introduced after comparison methods and the advent of operational annual corrections, which express then, in the atomic time standards in 1955, the universal time initial BIH system. Thus the annual systematic measurements could be referred to a common very error is due to the annual errors of astrometric uniform time scale, presently designated by measurements at initial epoch, in 1968 ; for po- International Atomic Time TAI, or to its sister lar motion, it is of the same order as for IPMS time scale UTC. Thus, besides establishing TAI ( 1m peak-to-peak) ; for UT1 it can be of the or- and UTC, the traditional task of the BIH evolved der of 3 ms, peak-to-peak. The spurious drift of in giving the difference between universal time the BIH pole should be less than 3 cm/year. The and TAI or UTC - a measure of the irregularity spurious drift of UT1 is mainly due to the drifts of the Earth rotation. On the other hand the in- of the star catalog equinoxes ; a drift of 1 ms/ crease of the observational precision necessita- year appears possible. ted taking into account the polar motion (IAU The precision of the BIH results is given by recommendation, 1955), leading to the definition the stability curve of figure 3 and 4. of UT1. As the coordinates of the pole were not The BIH is involved in the EROLD project available from the ILS in due time, N. Stoyko, (Earth rotation by Lunar Distances) and in the formerly in charge of the BIH, began to establish MEDOC experiment (Polar motion by Doppler obser- his own set of coordinates from the latitude data vations of satellites). A revision of past data of an increasing number of participating observa- was recently undertaken by Feissel. tories. In 1965, in order to reduce the delays The BIH, as the IPMS, sends its current of availability of the results, Guinot began results free of charge to about 800 addresses. to solve equations (l) and (2) simultaneously. These data are reproduced in several national Thus, the overlap of functions with IPMS appears publications. to be complete ; but this was unavoidable, be- cause the scientific unions did not make a com- Defense Mapping Agency (DMA) prehensive evaluation of the situation in 1962 when creating the IPMS. The computation of the pole position based on There have been many changes in the BIH work Doppler observations of Transit satellites origi- since its creation, due to the evolution of nated at the Naval Weapons Laboratory, USA, (now techniques. The following information refers to the Naval Surface Weapons Center), as a conse- the present situation. The BIH uses all data on quence of the research of Anderle and Beuglass Earth rotation from any source, as soon as their [1970], A service was organized under the name systematic errors can be sufficiently well of Dahlgren Polar Monitoring Service, giving the modeled. Until 1972.0, only astrometric data coordinates of the pole since 1969. In April (the same as for IPMS) were used. Since 1972, it 1970, the responsibility for the computation of was possible to use the pole determinations by the orbits of the Transit satellite, and there- satellite observations from the DMA (see below). fore of the pole motion, was transferred to the The computational methods were described by Topographic Center of Defense Mapping Agency, Guinot and Feissel [1968] and with more details without changes of the computation programs. by Feissel [l972] . They consist first in estab- Many improvements took place, which are re- lishing a "system" by determination of the ini- ported, together with a comprehensive bibli- tial latitudes and longitudes, and the annual ography by Oesterwinter [l978]. The most import- systematic corrections to data. Then, equations ant one appeared in the 1972 results, where the (1) and (2), with appropriate weights, are polar motion, instead of being derived from solved for individual measurements of *?t £ and station residuals, was computed directly in the [UTOi - UTC]t. It is thus possible to adopt a least square solution together with the orbit short averaging time. parameters. Other improvements resulted from Raw five-day values of x, y, UT1-UTC have been better gravity field models, from better station available since 1967. In 1972.0, the satellite coordinates, from the increase of observing sta- data began to contribute to these raw values ; tions (about 20 presently). they presently receive about half the total The reduction techniques were described by weight for polar motion. Nevertheless, for com- Anderle [l973]. It is reminded that 48-hour time parison purposes a solution from astrometry spans are used, leading to two-day raw values of only is continued ; it is based on the same data the pole coordinates for each satellite. These as IPMS work. Smoothed five-day values are values, with their standard deviations, are made published filtered with a cut-off at about 30 available within a few days in DMA reports. Five- days ; they extend in an homogeneous series since day averages are also given weekly in the USNO 1962. Series 7 Circulars. The current results for month m are published The possibility of systematic errors due to in BIH Circular D at the beginning of month m + the model of forces was investigated by Taton 2, with provisional raw values. Improved results [ 1972J and by Bowman and Leroy [1976] . Errors 17 with periods 5 to 6 days and 11 days were found, good scientists and technicians, this metrological work is not considered as attractive ; it is not but they can be easily filtered. Longer term easily supported by national and international errors may exist, but are probably small when organizations, which prefer to grant new projects. compared to the errors due to changes in the station network and plate motions. The size of Acknowledgments seasonal effects, if any, cannot be estimated, but appears to be much smaller than for astro- The IPMS and BIH are services of the Federation met ry. of Astronomical and Geophysical Services (FAGS), The precision curve is given by figure 3. A a member of the International Council of Scienti- source of long term, noise could be the succes- fic Unions (ICSU). The allocation of FAGS to the sive refinements of the model of forces : these services, based on ICSU and UNESCO funds, effects are now very small. although essential, only represents 5 to 10% of Thus, the DMA maintains a polar motion ser- the real cost of the central bureaus. The remain- vice which is not officially recognized by ing is supported mainly by the host countries, scientific unions, but is essential, as being and also by many establishments of outside coun- more accurate, more precise and more rapid than tries, private fundations, and other international the official ones. The possibility of managing organizations. a Doppler network by scientific organizations is The observation stations, the Doppler service being tested in the MEDOC project [Guinot and are entirely supported by national sources. Nouel, 1976 ; Nouel and Gambis, 1978J. It would be much too long to give a list of all contributors, they will recognize themselves, and Other series of results accept our thanks in the name of the scientific community. As prior to 1962 no latitude series other than those of northern ILS stations contributed to References the official work on polar motion, Fedorov and his collaborators at the Kiev observatory comput- Aardoom, L., Proceed. GEOP 9 (1978), this volume. ed the polar motion from all known measurements. Anderle, R.J., Geophysical Surveys, JL, 147, 1973. They were able to get some coordinates starting Anderle, R.J., and L.K. Beuglass, Bull, geodesi- from 1846, and high precision coordinates from que, 96, 125, 1970. 1890.0 to 1969.0 Fedorov et al. , [1972J. Barnes, J.A., et al., IEEE Trans, on Instrum. and The Gosstandard of USSR computes a solution Meas., IM-20, N° 2, 105, May 1971. for UT1 based on the 21 instruments for universal Bowman, B.R., and C.F. Leroy, Proceed, int. geo- time measurements operating in USSR and neighbour- detic symp., p. 141, Las Cruces, oct. 1976. ing countries.. This solution is based on an ori- Fedorov, E.P., et al., The motion of the pole of ginal optimum estimation ; it is published in the the Earth from 1890.0 to 1969.0, Ed. : V.A. bulletins "Vcemirnoe Vremja" (Series E) of the Bulkina, published "Naukova Dumka',' Kiev, 1972. USSR Gosstandard. Feissel, M., BIH Annual Report for 1971, part E, 1972. Conclusion Guinot, B., and M. Feissel, Proceed. UAI Symp. N° 32, p. 63, D. Reidel Pub. Co., 1968. Although their organization is not optimum, Guinot, B., and F. Nouel, Proceed, int. geodetic the services have produced uninterrupted series Symp.,, p. 159, Las Cruces, Oct. 1976. of pole coordinates and UTl values, with good Markowitz, W., Bull, geod., 59, 29, 1961. homogeneity. They have proved their ability Nouel, F., and D. Gambis, Proceed. IAU Symp. 82 to issue routinely the data with the short delays (Cadiz, May 1978), in press. needed in modern research. The causes of this Oesterwinter, C., Proceed. IAU Symp. 82 (Cadiz, success are worth considering when preparing new May 1978), in press. services. Sevarli<5, B.M., Publ. Obs. astron. Belgrade, 5_, The work of the visual observers was (and is 1957. still) essential. One must be reminded that the Taton, N., Determination du mouvement du pole a same person has often to work more than six hours partir d1observations de satellites arti- in the middle of the night, at outside tem- ficiels, These, Universite de Paris VI, 1972. peratures, ignoring holidays and week-ends. Even Vicente, R.O., and S. Yumi, Publ. intern. Latitu- with new automatized instruments the importance de Obs. Mizusawa. T_, N° 1, 41. of the attendance should not be under-estimated, Yumi, S., Annual Report of the IPMS for the year in service operation. 1974. 1976. The astronomical instruments are very reliable. The Doppler receivers can be easily replaced in case of failure. Are we sure that other techniques, with sophisticated devices, will not suffer from interruptions, especially when based on a small number of stations ? The operation of a service is quite different from an experiment. It has to run continuously, at the highest level of quality, in spite of many difficulties : vacations, illnesses, pregnancies, strikes, computer failures... Although requiring 18