To determine the surface geostrophic currents (which are proportional to the slope of the dynamic topography) from satellite altimetry, one needs an accurate and independent estimate of the geoid. Currently, these estimates are given in terms of spherical harmonics. While the high-degree terms (short wavelengths) are either uncertain or completely unknown, the low-degree terms (long wavelengths) are reasonably accurate. Furthermore, in terms of spherical harmonics, most of the dynamic topography's power is contained in the low-degree terms. If the dynamic topography is treated as the signal and the geoid uncertainty as the noise, the signal to noise ratio is the highest for these low-degree terms. Therefore a spherical harmonic expansion of the difference between the altimeter-derived means sea surface and the geoid estimate should reveal the large-scale circulation of the ocean surface layer when one examines the low-degree terms. Methods based on this principle are proposed and partially demonstrated over the Pacific Ocean using the mean sea surface derived from the SEASAT alimeter and the Goddard Earth Model 9 earth gravity model. The preliminary results show a well-defined clockwise gyre in the North Pacific and a much less well defined counterclockwise gyre in the South Pacific. When the dynamic topography so obtained is compared to Wyrtki's (1975) dynamic topography derived from hydrographic data, the agreement is within the limit of geoid uncertainties ans satellite orbital errors. To fully determine the three-dimensional of the ocean circulation, one still needs hydrography: the three-dimensional density distribution of the ocean. A method is presented for the determination of the ocean circulation and the improvement of the geoid in the combined problem of geodesy, hydrography, and satellite altimetry.