Evaluation of GEOSAT data and application to variability of the Northeast Pacific Ocean

Abstract

A portion of the northeast Pacific ocean was chosen within which to evaluate and use altimetric data from the U.S. Navy Geodetic Satellite GEOSAT. The zero-order accuracy of the major GEOSAT geophysical data record (GDR) channels was verified, and occasional gaps in the altimeter coverage were noted. GEOSAT'S 17-day repeat orbit allowed use of collinear-track processing to create profiles of the difference between the sea surface height along a given satellite repeat, and the mean sea surface height along that repeat's groundtrack. Detrending of sea surface bias and tilt on each repeat reduced orbit and other long wavelength errors in the difference profiles.

The corrections provided on the GEOSAT GDR were examined for their effects on the difference profiles of three test arcs. It was found that only the ocean tide, electromagnetic bias, and inverted barometer corrections varied enough over the arc lengths (~4400 km) to have any noticeable effect on the difference profiles. Only the ocean tide correction was accurate enough to warrant using it to adjust the sea surface heights. The recommended processing of GEOSAT data for the area included making the ocean tide correction, three-point block averaging successive sea surface heights, and forming the mean height profiles from 18 repeat cycles (to reduce aliasing of the M2 tidal component). A set of difference proftles for one GEOSAT arc indicated that a reasonable estimate of GEOSAT's system precision was -4.5 cm (RMS). The mid wavelength range (100-500 km) of these profiles was found to be the only range in which oceanic mesoscale features could be separated from altimeter errors.

Mean alongtrack wavenumber spectra of oceanic variability for two GEOSAT arcs were compared with a SEASAT-derived regional spectrum of Fu (1983). Agreement was good, with GEOSAT showing less system noise at short wavelengths, and greater oceanic variability at long wavelengths. The GEOSAT spectra were fit well by a k-1.5 slope at wavelengths from 100 to 1000 km.

Sea surface temporal variability as a function of location was compared with the SEASAT results of Cheney et al. (1983). Qualitative agreement was excellent and quantitative differences were largely accounted for. GEOSAT picked up the variability of the major current systems of the northeast Pacific, including the Alaskan, Californian, and North Equatorial currents. Error bounds on GEOSAT-derived oceanic variability showed that the effects of uncorrected electromagnetic bias, inverted barometer, and wet troposphere were significant. Further work in the areas of error modelling, orbit determination, and geoid calculation were called for.