Knowledge Bank

Sea surface dynamic topography (SSDT) has been determined from the separation of the Topex sea surface height and the geoid undulations implied by the JGM-2/OSU91A merged potential coefficient model. The SSDT was initially analyzed to determine annual and semiannual variations in the ocean surface, corrections to the M2, S2, K1, O1 constituents of the Cartwright/Ray tide model, and bias and bias rate terms. All correction terms were represented by spherical series to degree 8 (for most testing purposes) and to degree 15 for final runs. The most complete analysis was done with Topex cycles 4 to 58. From this solution we found that the root mean square annual signal was ±4.5 cm in the Northern Hemisphere and ±3.5 cm in the Southern Hemisphere with an overall amplitude of ±4.0 cm. Large annual signals were associated with features such as the Kurishio Current (±10 cm), Gulf Stream (±7 cm) and the Falkland Current (±6 cm). The tide correction terms were ±2.9 cm (M2), ±1.5 cm (S2), ±0.8 cm (O1), ±1.4 cm (K1). The improvement in the tide model was proven through comparisons at the 104/95 tide gauge data set provided by C. Le Provost. The bias rate value was determined at each normal point and then averaged to determine a mean rate of 4.8 mm/yr. After removing an altimeter drift rate of 2.0 mm/yr the net rate is 2.8 mm/yr. This rate could be a combination of sea level change and effects associated with the various correction terms applied to the data. The second main theme of the report relates to the representation of SSDT in spherical harmonics (SH) and orthonormal (ON) functions. Procedures were implemented to estimate either SH or ON coefficients. It was found that the estimation of SH coefficients was more stable than the ON estimation. For those applications on which spectral information of SSDT is needed a SH model can be transformed into an ON model. It was found that the lower degree SH coefficients are quite sensitive to the bias correction applied to the sea surface height data. This sensitivity does not occur in ON coefficient estimation. For the analysis with data from cycles 4 to 58 a bias term of 55 cm was applied to the data. This value leads to a zero value for the degree zero term in the ON expansion. An SH expansion to degree 24 was computed using a priori degree variance computed from 10 selected cycles of Topex data. The SH expansion was transformed to an ON expansion and both coefficient sets are given in the report. The degree 2,0 coefficient in the SH representation was -37.8 cm quite comparable with other altimeter based solutions. [Full text of abstract available in document.]