The application of least-square collocation and variance component estimation in crossover analysis of satellite altimetry observations and altimeter calibration

Abstract

ABSTRACT In this study, the collocation method accompanied with variance component estimation is used for least square adjustment of crossover observations in order to determine the effects of radial errors on the observations of satellite altimetry. The collocation is used for time series analysis of sea surface height observations both for predicting the possible missing observations in each cycle, and for approximating the observation of each cycle at crossover points. In addition, use is made of the variance component estimation to quantify the noise variance of observations and improve the least square evaluation of radial errors. For analysis of radial errors, two different approaches are followed, in the first approach, the radial errors are assumed to behave like a series of trigonometric function, the coefficients of which are unknowns which should be determined from observations. In the second approach, the values of radial errors, for ascending and descending passes are determined. Our results show the efficiency of collocation algorithm for highly accurate time series analysis of altimetry observations and moreover, they reveal the effectiveness of variance component estimation for true noise specification of observations which can significantly improve the results of least square adjustment. The outcome of this study can be used to calibration of altimeters. The numerical results indicate that the mean range biases of Topex/Poseidon, Jason 1-2 and ENVISAT in the six single and dual crossover points using the first and the second methods are about 0, 84, 33, 204 and 0, 98, 41, 286 mm, respectively.