Accurate bathymetric data is crucial for various aspects such as marine resource exploitation and marine ecological conservation. Currently, satellite-derived bathymetry (SDB) based on empirical and physical models has been widely utilized in constructing underwater terrain in shallow seas. However, the application of such SDB models is limited in remote island reef areas lacking in-situ measurement data. To overcome this issue, the manuscript proposes an unconstrained SDB optimization method without in-situ measurement data, utilizing satellite multispectral imagery (Geoeye-1) and spaceborne LiDAR data (ICESat-2). By classifying the seafloor substrate in coral reef areas into sandy and coral, based on the depth invariant index (DII), we employ an adaptive logarithmic ratio model for unconstrained SDB. The ICESat-2 LiDAR data are then used to correct the SDB results, achieving bathymetry optimization in the coral reef area of the Xisha Islands. Additionally, the proposed method is applied to Yuanzhi Island of the Xisha Islands, and the accuracy of the bathymetric results is evaluated against ALB (Airborne LiDAR Bathymetry) data. The findings demonstrate that compared to conventional methods, our method can improve the accuracy of SDB results with good adaptability. In the Yuanzhi Island area, the proposed method yields SDB results with an R2 of 0.93, an MAE (Mean Absolute Error) of 0.94, and an RMSE (Root Mean Square Error) of 1.12 m, compared to ALB data. The average error is less than 10 % of the maximum depth, essentially meeting the requirements of the International Hydrographic Organization (IHO) standards for depth measurement error when depth is <20 m. This study can offer a novel approach for enhancing bathymetric accuracy around offshore and remote islands, where gathering underwater terrain data is challenging.

中文翻译：

---

一种结合深度不变指数和自适应对数比的卫星测深方法——以西沙群岛无原位测量为例


准确的测深数据对于海洋资源开发和海洋生态保护等各个方面都至关重要。目前，基于经验和物理模型的卫星测深 （SDB） 已广泛应用于浅海水下地形的构建。然而，这种 SDB 模型在缺乏原位测量数据的偏远岛屿珊瑚礁地区的应用受到限制。为了克服这个问题，该手稿提出了一种无约束的 SDB 优化方法，没有原位测量数据，利用卫星多光谱图像 （Geoeye-1） 和星载 LiDAR 数据 （ICESat-2）。通过根据深度不变指数 （DII） 将珊瑚礁区域的海底基质分为沙质和珊瑚，我们对无约束 SDB 采用了自适应对数比模型。然后使用 ICESat-2 LiDAR 数据来校正 SDB 结果，从而在西沙群岛的珊瑚礁区域实现测深优化。此外，将所提方法应用于西沙群岛的元直岛，并根据 ALB （Airborne LiDAR Bathymetry） 数据评估了测深结果的准确性。研究结果表明，与传统方法相比，我们的方法可以提高 SDB 结果的准确性，具有良好的适应性。在远直岛地区，与 ALB 数据相比，所提出的方法产生的 SDB 结果为 R2 为 0.93，MAE （平均绝对误差） 为 0.94，RMSE （均方根误差） 为 1.12 m。平均误差小于最大深度的 10%，基本满足国际水文组织 （IHO） 标准对深度为 <20 m 时的深度测量误差的要求。 这项研究可以提供一种新的方法来提高近海和偏远岛屿的测深精度，在这些岛屿上收集水下地形数据具有挑战性。