Journal of Oceanography, Vol. 59 (No. 6), pp. 765-781, 2003
Tsurane Kuragano* and Masafumi Kamachi
Oceanographic Research Department, Meteorological Research Institute, Tsukuba 305-0052, Japan
(Received 22 August 2002; in revised form 7 March 2003; accepted 14 March 2003)
Abstract: Sea surface height anomaly maps of realistic eddy activity were obtained by applying space-time optimum interpolation to altimeter data. Analysis error and rate of reconstructing eddy signals were investigated by taking account of: 1) dependency on orbit configurations of single and multiple altimeters; 2) dependency on space-time scales of realistic, dominant eddies; and 3) effect of space-time scales of eddy propagation. Large-scale sea surface height anomalies are subtracted from altimeter data by applying an along-track filter to allow easy handling of eddy signals. The space-time scales of the first-guess error in the optimum interpolation are statistically evaluated by fitting a space-time anisotropic Gaussian function to space-time-distributed correlation coefficients of sea surface height using the TOPEX data. The results of the optimum interpolation clarify the followings: 1) ERS has a better capability of reconstructing eddy signals than TOPEX. Comparison of maps from multi-altimeter data shows that TOPEX+ERS has a better capability than Jason-1+TOPEX in lower latitudes and vice versa in higher latitudes, though the differences are small. 2) The small space-time scale yields a low reconstruction rate in marginal seas and alongside the equator. The persistent timescale is large, and westward propagation is dominant in the subtropical and subarctic regions, where the reconstruction rates are high. 3) The optimum interpolation, taking account of eddy propagation, provides higher reconstruction rates than that taking no account of the propagation. The effect of propagation on the optimum interpolation is greater when it is applied to single-altimeter data than to multi-altimeter data.