LAUSR

Abstract The filtering procedure is usually mandatory for computing the mean dynamic topography (MDT) by combining the mean sea surface (MSS) and geoid due to the inconsistent spectral contents between them. Recently, a rigorous fusion method is proposed to solve this inconsistency problem through the least square method (Becker et al., 2012), where MDT is parameterized by the Lagrange basis functions (LBFs). In this study, the latest CNES-CLS18MDT with its associated error information is introduced for the parameterization experiments, where the LBFs consist of 4, 16, and 36 polynomial parameters (4P/16P/36P) are implemented and compared, respectively. The experiments are conducted in two research areas close to the Kuroshio and Gulf Stream. Besides, two oceanographic, two geodetic MDT models, and the average of the above models are introduced as the comparison data to assess their respective agreement with the recovered MDT based on LBFs. Results show that the standard deviation (SD) of the misfits from the results of 16P against the mean MDT is 2.6 cm (Kuroshio current) and 6.8 cm (Gulf stream), respectively. Compared with the 4P/36P, the 16P reduced the misfits in these two regions by 0.8–1.1 cm and 0.3–0.5 cm, respectively. The misfits between the results using 16P and the MDT models are ~5 cm lower than that using 4P/36P in regions with a strong variation of ocean current. In total, the 16P is more favorable in representing the characteristics of MDT and has a better versatility when parameterizing MDT under different ocean states.