JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 100, NO. C12, PAGES 25,331-25,352, DECEMBER 15, 1995 The temporal and spatial characteristicsof TOPEX/POSEIDON radial orbit error J.A. Marshall Space Geodesy Branch, NASA Goddard Space Flight Center, Greenbelt, Maryland N.P. Zelensky, S.M. Klosko,D.S. Chinn,S.B. Luthcke,K.E. Rachlin and R.G. Williamson Hughes STX Corporation, Greenbelt, Maryland Abstract. Satellite orbit error has long been the bane of oceanographers who analyze altimetry data. However,radial orbit erroron TOPEX/POSEIDON (T/P) hasbeenreduced to the 3 to 4-cm root- mean-square (rms) levelover a 1 O-day repeat cycle, which represents anorder of magnitude im- provement over earlier altimetry missions such asGeosat. Consequently, oceanographers arenow able todirectly evaluate the absolute ocean topography to unprecedented accuracy levels. Whilesig- nificantly reduced, theT/P orbit error stillrequires quantification. Thisstudy examines thespatial and temporal characteristics of the T/P radial orbit error, asassessed through theanalysis of laser tracking residuals and orbit comparisons withindependently generated trajectories. Spectral analy- ses of the orbit differences between the orbits determined fromsatellite laser ranging and Doppler Orbitography and Radiopositioning Integrated by Satellite data and theindependently determined reduced dynamic Global Positioning System (GPS) ephemerides indicate that the predominant pow- er is at theonce-per-orbital revolution frequency with 2- to 3-cmpeaks. Whentheorbit differences are colinearly aligned toa fixed geographic grid and spectral analysis isperformed ateach geograph- ic grid point, a nearly 60-day period isfound with maximum amplitudes in the 2- to 4-cm range. The contribution of both conservative and nonconservative force and measurement mismodeling to this error signal areassessed. We demonstrate that the-•60-day error period seen at fixed geographic lo- cations arises from weaknesses in thedynamic ocean tidal models used in theorbitcalculations. New tidal models have been developed which significantly reduce this error. Second-generation or- bits incorporating many model improvements have been computed and demonstrate a significant re- duction in the radial orbiterrorsignals. Some orbiterrorstill exists, andmethods for further model improvements and thepossibility of achieving 1-cm radial rms orbit accuracy in T/P are discussed. 1. Introduction Satellite altimetry hasprovided the oceanographic community with a g!- hal synoptic observational data setof the ocean surface topography and,when combined withknowledge of thegeoid, the majorgeostrophic currents. In the past, insufficient knowledge of the spacecraft's radial position haslimited the application of these datasets. Seasat radialpositioning accuracy wasestimated to be 1.5m [Marsh andWilliamson, 1980; $chutz andTapley, 1980],and Geosat was between 25 and 85 cm,depending upon thegravity field used [Haines et al., 1990,1994]. Consequently, investigators devel- oped empirical strategies to remove theorbiterrorfrom the altime- tric data at the expense of losing valid long-wavelength oceanographic signal [Cheney etal., 1983]. These techniques, how- ever, have proved unnecessary and provide degradedoceano- graphicresultswhen analyzingthe TOPEX/POSEIDON (T/P) altimetry datausing the precise orbits on the mission geophysical datarecords (GDRs). TheT/P satellite flies in a nearly circular orbit at an average alti- tude of 1336km with an inclination of 66.05 ø.It hasa period of Copyright 1995 by theAmercian Geophysical Union. Paper number 95JC01845. 0148-0227/95/95JC-01845505.00 112.4 min and a ground track repeat cycle of 9.9156days after com- pleting 127orbital revolutions. The rate of change for theargument of perigee is nearzeroin order to satisfy the "frozen" repeat orbit criteria. Orbit maintenance maneuvers areperformed about every3 months to keep theorbitground track repeating to within_+ 1 km and themean orbiteccentricity near zero. The rms radial orbiterrors for T/P are between 3 and 4 cm due largely to improved gravityand nonconservative forcemodels [Nerern et al., 1994;Tapley et al., 1994a].However,even orbit errors of this small magnitude can adversely affect certain altimeter applications and need to be under- stood. Accordingly, the purpose of this paper is to assess thetempo- ral andspatial characteristics of theT/P radialorbiterror, identify possible sources for theerror, and explore strategies for reducing or eliminating theireffect. The factthat no measure of absolute orbit accuracy exists makes the process of assessing orbit qualitydifficult.Several different methods areused to inferorbitaccuracy; these include agreement withthetracking data, orbit overlap tests generated fromsubsets of thetracking data, andorbitcomparisons with independently com- puted ephemerides. No single test directly measures the complete orbit error, but asan aggregate, they canprovide a reasonable esti- mate. For past altimetric missions thealtimeter rm•ge and crossover residuals werewidely used asanindependent measure to quantify errors in the satellite's radial position. However, for thisT/P study, mesoscale sea surface variations, uncertainty in modeling ocean 25,331