The potential of using airborne gravity gradient tensor (GGT) for coastal quasi-geoid (QG) refinement is explored, and the contributions introduced from individual GGT components and their combinations are quantified and evaluated. High-resolution QGs, with a spatial resolution of ~0.5 km, are computed over St. George’s Bay in southwestern Newfoundland, Canada. The findings indicate that fully focused Synthetic Aperture Radar (FFSAR) and Surface Water and Ocean Topography (SWOT) altimetry data are effectively in differentiating the performance of various QGs in coastal areas. The application of the vertical gravity gradient obtains the highest quality QG when utilizing individual GGT components. The combination two or more components results in improved QGs compared to the results derived from individual components. The integration of full GGT yields the best QG, with standard deviation (SD) of misfits against Sentinel-3A FFSAR (SWOT) altimetry data being 1.13 (2.49) cm, representing reductions of 28.48–53.50% (5.32–15.02%) compared to results derived from individual GGT components. Comparisons of the QG computed by fusing full GGT with the Canadian gravimetric QG CGG2013 and high-degree global geopotential models further underscore the advantages of using GGT in QG modeling, revealing SD reductions of 55.16–65.02% (7.55–33.95%) against FFSAR (SWOT) altimetry data. These findings underscore the effectiveness of using airborne GGT in coastal QG modeling, particularly in recovering short-wavelength signals and addressing challenges in satellite altimetry over coastal environments. Additionally, this study highlights the superiority of using full GGT over individual components in QG modeling.

Bathymetry provides instrumental information for studying sedimentary processes, global climate change, and benthic morphologies. The advantages and applicabilities of different techniques for bathymetry detection vary. We propose a framework for bathymetry enhancement from multisource data based on spherical radial basis functions (SRBFs). A case study is conducted over the Paracel Islands in South China Sea (SCS), where Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) measurements, airborne gravimetric observations, echo soundings, and the reference model DTU18BAT are merged. Numerical results suggest that the fusion of ICESat-2 observations dramatically enhances the quality of the computed bathymetry model near the islands, the root-mean-squared error (RMSE) of which is reduced by 45.35%–67.95% compared to existing models when validated against the satellite-derived bathymetry (SDB) with decimeter-level accuracy. By additionally fusing the airborne gravimetric data, bathymetry is further enhanced by ~22.49%, particularly over islands with sparse ICESat-2 trajectories. Comparisons with surveyed airborne bathymetric lidar data over the northern Antelope Reef yielded results consistent with those obtained from the SDB, suggesting that SDB is possible to serve as control data in waters devoid of ground truth data. Further analysis reveals that the models constrained by echo soundings performed better than existing models in deep waters, with reductions of 17.79%–44.99% in terms of RMSE. By fusing airborne gravity data, bathymetry is improved by ~10%, highlighting the utilization of airborne gravimetry in both shallow and deep waters. The proposed SRBF approach offers an effective way to merge heterogeneous data for high-quality bathymetry determination.

The lack of marine gravimetric measurements and the presence of severely contaminated altimetry data pose multiple challenges in high-quality quasi-geoid (QG) and mean dynamic topography (MDT) determination over islands, where the application of nadir altimetry alone is inadequate. We explore the potential for regional enhancement using wide-swath data from the Surface Water and Ocean Topography (SWOT) mission. Numerical experiments over the Paracel Islands in South China Sea underscore the superiority of using SWOT data in QG/MDT computation. Comparisons with the airborne gravimetry-derived QG reveal that the Root Mean Square Errors (RMSEs) of QGs derived from the SWOT data are reduced by 38.24–60.90% compared to those derived solely from nadir altimetry. The QG profiles retrieved from the Sentinel-3A/B altimetry using the fully-focused SAR technology are effective in discriminating the quality of different QGs near islands. The RMSEs of SWOT-derived QGs constitute reductions of 6.27–17.52% compared to those computed from nadir altimetry alone. Notable improvements up to 4 cm are observed when Sentinel-3A/B tracks approached islands. The mutual comparison of the QGs computed from the SWOT gravity anomaly (GRA) and vertical gravity gradient (VGG) data suggests that the VGG-derived QG has improved quality, and the utilization of VGG recovers more small-scale signals. The SWOT-derived MDTs reduce the bubble-like errors up to several centimeters compared to those computed exclusively using nadir altimetry. Our findings highlight that utilizing SWOT data enables the acquisition of an accurate QG/MDT with an RMSE less than 1 cm compared to that derived from airborne gravimetry.

We study the role of airborne gravimetry for seamless bathymetry modeling over the Paracel Islands in northern South China Sea, and investigate the possibility of using ICESat-2 data and satellite-derived bathymetry (SDB) to evaluate bathymetry models over shallow waters. We use ICESat-2 data for training Sentinel-2 imagery and derive the SDB data with a Root Mean Squared Error (RMSE) of 0.29-0.50 m, which is lower than 10% of the maximum depths. The local bathymetry is modeled by using a modified version of S&S band-pass filter, and a partition-wise scheme is applied for determining the scaling factors. Numerical experiments verify the feasibility of using ICESat-2 and SDB data to assess bathymetry models. By utilizing the airborne gravity data, the fit between the computed bathymetry and the SDB data is significantly improved, by 18.7-58.0% over different shallow waters compared to recently released bathymetry models. The bathymetry predicted from the airborne data has also higher performance in deep water areas, which performs best in all these depth ranges from 500 to 3000 m. In comparison to the existing models, the RMSEs of the misfits between the computed bathymetry and the National Oceanic and Atmospheric Administration depths are reduced by tens to hundreds of meters in different depth ranges. Our study highlights that using airborne gravimetry for bathymetry modeling over island areas is advantageous, in both shallow and deep waters; and that ICESat-2 and SDB data can largely alleviate the lack of in-situ depths over shallow waters.

Ocean eddies, known as the weather of the ocean, represent gyrating water masses that have horizontal scales from 10 km up to at times 500 km. They transport water mass, heat, nutrition, and carbon and have been identified as hot spots of biological activity. In radar altimetry, they affect alongtrack measurements of sea level height and lead to problems in the subsequent generation of sea level maps. Monitoring eddies is therefore of interest among others to marine biologists, oceanographers, and geodesists. In this paper, using occlusion sensitivity maps (OSMs) we investigate different neural network architectures that address the task of automatic detection of ocean eddies, which is challenging due to their spatio-temporal dynamic behavior. Thus we analyze the importance of the spatial context that is needed to infer correct semantics and compare them between the different architectures. For this, we use data from satellite altimetry since it offers sea surface heights precise enough to expose the presence of eddies. For detection, we utilize a transformer neural network called Teddy which can exploit temporal and spatial information in the data. For evaluating our approach, we use gridded data sets for the area of the western part of the southern Atlantic from 2000 to 2011. Our results are evaluated primarily by employing the dice score metric and show that transformers can infer the semantics with similar performance compared to state-of-the-art CNNs but at the same time are less sensitive towards structural changes due to different modeling of the spatial information of the data.