Orbit and tide error reduction for the first 2 years of TOPEX/POSEIDON data

Abstract

More than 2 years of TOPEX/POSEIDON data (repeat cycles 1–81) are processed with orbit and tide error reduction schemes that preserve the oceanic signal. The rms crossover difference (XD) between heights belonging to the same cycle is a good measure of the errors because the real oceanic variability manifests little of itself in times less than 10 days. This rms XD is reduced from 10.8 to 8.1 cm after the tide error correction and to 6.0 cm after the orbit error reduction and editing, a more than threefold reduction in power, even for the remarkably error‐free TOPEX/POSEIDON data. If the mesoscale variability that manages to reveal itself in less than 10 days (e.g., in the western boundary currents) and residual coastal tide errors are screened out, the global area‐weighted residual XD is only 4.3 cm, implying that for variability studies the residual error is only 3.0 cm for instantaneous point measurements of altimetric sea level away from coastal regions, far exceeding the prelaunch mission requirement. The important issue of signal preservation (which has been shown analytically elsewhere) is addressed here directly using data treated with orbit reduction as well as untreated data. It is shown that the time‐compartmentalized orbit error reduction leaves the annual heating and cooling cycle intact as anticipated by the analytical proof, whereas conventional orbit error removal procedures would wreak havoc. The error of the Cartwright and Ray (1991) tide model is estimated to be in excess of 5 cm (global area‐weighted rms including shallow seas). The crossover‐difference‐inducing part of the orbit correction (which includes any short‐term, large‐scale, along‐track variations) is estimated to be 3.3 cm (along‐track rms) for the Joint Gravity Model 2 orbit. The total orbit error is estimated to be 3.5 cm, an independent confirmation of estimates based on satellite tracking and other means.