| description abstract | Sea surface height may be measured by a satellite-borne altimeter and its along-track slope used to infer geostrophic currents. A major difficulty is that, in general, the local geoid and satellite orbit are not known to the accuracy desired. Thus, comparison is often made between repeat flights of an altimeter along fixed ground tracks in order to infer the changes in the currents. In practice, it is convenient to calculate mean height profile from many repeat passes and use this as a reference, so that individual altimetric profiles yield variations about this mean. It is thus important to derive a high-quality reliable estimate of the altimetric mean in order to minimize the errors in the inferred flows for the individual repeats. This work examines various methods for deriving the mean profile. Using simulations, it is shown that the error of an algorithm in retrieval of the original mean can be expressed as the sum of the error due to oceanographic signals and random noise (which is the same for all methods and is unavoidable) and an error level associated with that particular algorithm. This latter term, referred to as the net error of a method, is shown to increase with data loss and noise level and decrease with number of repeats as expected. However, the magnitude of this term does differ markedly between methods and changes with extent of data available. The height signatures of oceanographic features (fronts and eddies) are also included in the simulations-although they have a marked effect on the overall accuracy of the mean, they make only a minimal change to the net errors, that is. those particular to individual methods. | |
