Numerous MSS grids have been developed in the past by various groups using sea surface height data from a number of altimeter satellites. The mean sea surface used for the initial Topex/Poseidon GDRs was that developed by Basic and Rapp (1992). The OSUMSS92 was given on a 0.125 x 0.125 degree grid calculated from Geos-3, Seasat, and Geosat ERM altimeter data with high frequency signal estimated with the aid of a high resolution bathymetric data set. This MSS was placed in the reference frame defined by the Geosat data set. Comparisons between OSUMSS92 and Topex data indicated a root mean square difference of approximately +-17 cm (over one cycle) after translation and bias effects were taken into account (Rapp, Yi, Wang, 1994).

With the availability of the Topex/Poseidon, ERS-1, 35 day repeat cycle, and ERS-1 168 day repeat cycle, as well as Geosat data, various groups embarked on the creation of an improved sea surface. At the May 1995 meeting of the Topex/Poseidon SWT a discussion took place where several groups (University of Texas at Austin, CNES/GRGS, Ohio State) described the data and procedures being used for MSS determination. This discussion led to the identification of a number of issues (Rapp and Nerem, 1995) related to MSS definition and generation that were to be resolved through additional computation and revised MSS grids. Within one month after the SWT meeting four new grids (from UT/CSR, OSU, CNES/GRGS, GFZ/D-PAF) were available for testing and validation. Actually UT/CSR and OSU produced two grids: one in which an inverted barometer (IB) effect had been removed and another where no such correction was made.

Tests were carried out by comparing the sea surface height and along track gradients found from the MSS grids with actual data from Topex, Geosat, and ERS-1 (35 daya repeat). Mean sea surface height tracks and individual cycles were examined. Statistics on the differences were computed at UT/CSR, OSU, JPL, GSFC, and GFZ/D-PAF with the results being distributed by e-mail during June and July 1995. In addition contour plots and color images were made available to the designated evaluation group. Based on criteria (goodness of fit with sea surface heights and slopes to Topex/Poseidon data (primarily) and Geosat and ERS-1 (secondarily) and geographic grid coverage (third) described in an e-mail message of July 21, 1995, Nerem and Zlotnicki recommended that the IB corrected OSU MSS be adopted for the processing of the new GDRs. The message notes that the "UT/CSR and OSU models were practically indistinguishable in terms of their performance" and that all MSS grids represented "spectacular improvement in mean sea surfaces".

The OSUMSS95 is based on a one year mean Topex sea surface height track, a one year ERS-1 (35 day repeat), a one year Geosat ERM track and the first cycle of the 168 day repeat track of ERS-1. The values are given on a 1/16 degree grid with an IB correction made in the processing of the sea surface height data. The values are given in the mean tide system and refer to an ellipsoid whose parameters are: a=6378136.3m and f=1/298.257. The center and axis alignment of this ellipsoid corresponds to the Topex/Poseidon reference frame. The MSS grid extends from 82N to 80S. The scale of the MSS values is defined by the Topex altimeter measurements with no bias correction made. The grid values in several land regions are given as geoid undulation values computed from the merged JGM-3/OSU91A potential coefficient model. These regions were: 60N to 40N, 60E to 100E; and 60N to 40N, 240E to 260E. In addition a separate MSS calculation was made in the Caspian Sea region (60N to 35N, 45E to 60E) recognizing the level of the Caspian Sea is about 30m below the geoid. The details on the development of the OSUMSS95 are given in Yi (1995) where numerous comparisons and evaluations can be found. For example, the standard deviation of the difference between TOPEX (cycle 25) sea surface heights and along track gradients is 9.3cm and +-0.62 cm/km. Both values represent a significant improvement over the use of the OSUMSS92.

As noted in the initial tests and as pointed out by Yi (1995) and Anzenofer et al.(1996), the OSUMSS95 exhibits, in a few regions,track signature in images created from the gridded data. The patterns are primarily seen in areas of significant ocean variability (e.g. western boundary currents) where averaging of the 168 day ERS-1 data was not possible. In addition this could cause residual radial orbit error on other altimeter data through the crossover procedure used. The track pattern signature has the potential for causing cross track gradient errors. To assess the magnitude of the possible error a cross track gradient correction was calculated with the OSUMSS95 and the CSRMSS95 (that showed little track signature), using cycle 25 of Topex data in the Gulf Stream region. Cycle 25 was chosen because it is almost 1 km from the the nominal track. The standard deviation of the difference between the cross track gradient corrections implied by the 2 MSS grids was only 0.5 cm in this extreme situation.

Improved procedures have been developed at OSU that eliminate the track signature problem and retain the high frequency content of the data. The improved MSS and the OSUMSS95 were compared to sea surface height data in the Gulf Stream region (42N TO 32N, 290E TO 300E). The standard deviation between the sea surface height data and that predicted from the two sea surface grids was hardly changed: ( for Topex cycle 25:+- 21.7 cm (OSUMSS95) to 21.7 cm (NEW MSS); for ERS-1, cycle 11:+- 19.1 cm (OSUMSS95) to 18.6 cm (NEW MSS); for GEOSAT(ERM), cycle 5:+- 23.0 cm (OSUMSS95) TO 24.8 cm (NEW MSS)). Along track gradient comparisons showed little change (e.g.+-.60 to 0.62 cm/km for the Topex data). Cross track gradient changes have not been computed.

The MSS being incorporated on the new GDRs is a significant improvement over the previously used MSS grid (OSUMSS92). Improved MSS grids can be obtained in the future using longer time spans of data and with improved techniques for handling data for which averaging does not eliminate variability effects. Care must be given to the retention of high frequency signal and the reduction of high frequency noise.

References

Anzenhofer, MT Gruber and M Rentsch, Global High Resolution Mean Sea Surface Based on ERS-1 35- and 168- Day Cycles and Topex Data, In: Rapp, Cazenave, Nerem (ed.), Global Gravity Field and Its Temporal Variations, IAG Symp. 116, Springer Berlin , 1996

Basic, T and RH Rapp, Oceanwide Prediction of Gravity Anomalies and Sea Surface Heights Using Geos-3, Seasat and Geosat Altimeter Data and ETOPO5U Bathymetric Data, Rpt. 416, Dept. of Geodetic Science and Surveying, The Ohio State University, Columbus, 1992

Rapp, RH, Y Yi, and YM Wang, Mean Sea Surface and Geoid Gradient Comparisons with TOPEX Altimeter Data, J. Geophys. Res., 99(C12), 24,657- 24,667, 1994

Rapp, RH and RS Nerem, Geoid Undulation and Mean Sea Surface Recommendation, in Minutes of TOPEX/POSEIDON Science Working Team Meeting of May 1995, Jet Propulsion Laboratory, JPL D-12817, August 15, 1995

Yi, Y, Determination of Gridded Mean Sea Surface from TOPEX, ERS-1 and GEOSAT Altimeter Data, Rpt. 434, Dept. of Geodetic Science and Surveying, The Ohio State University, Columbus, 1995