A new global marine gravity model NSOAS24 derived from multi-satellite sea surface slopes
Abstract. Judging from the early release of the NSOAS22 model, there were some known issues, such as boundary connection problems in block-wise solutions and a relatively high noise level. By solving these problems, a new global marine gravity model NSOAS24 is derived based on sea surface slopes (SSS) from multi-satellite altimetry missions. Firstly, SSS and along-track deflections of vertical (DOV) are obtained by retracking, resampling, screening, differentiating, and filtering procedures on basis of altimeter waveforms and sea surface height measurements. Secondly, DOVs with a 1'x1' grid interval are further determined by the Green's function method, which applies directional gradients to constrain the surface, least-square fit to constrain noisy points, and tension constraints to smooth the field. Finally, the marine gravity anomaly is recovered from the gridded DOV according to the Laplace Equation. Among the entire processing procedures, accuracy improvements are expected for NSOAS24 model due to the following changes, e.g., supplementing recent mission observations and removing ancient mission data, optimizing the step size during the Green's function method, and special handling in near-shore areas. These optimizations effectively resolved the known issues of signal aliasing and the “hollow phenomenon” in coastal zones. Numerical verification was conducted in three experimental areas (Mariana Trench area, Mid-Atlantic Ridge area, Antarctic area, representing low, mid and high latitude zones) with DTU21, SS V32.1 and shipborne data. Taking NSOAS22 for contrast, NSOAS24 showed improvements of 1.2, 0.7, 1.0 mGal in 3 test areas by validating with SS V32.1, while declines of 0.6, 0.5, 0.3 mGal, and 0.2, 0.4, 0.3 mGal occurred in STD statistics with DTU21 and shipborne data. Finally, the NSOAS24 was assessed using two sets of shipborne data (the early NCEI dataset and the lately dataset from JAMTEC, MGDS, FOCD, and SHOM) on global scale. Generally, NSOAS24(6.33 and 4.95 mGal) showed comparable accuracy level with DTU21 (6.20 and 4.71 mGal) and SS V32.1 (6.40 and 5.53 mGal), and better accuracy than NSOAS22 (6.64 mGal and 5.64 mGal). Besides, the new model is available at https://doi.org/10.5281/zenodo.12730119 (Zhang et al., 2024).