Bathymetry over broad geographic areas using optical high-spatial-resolution satellite remote sensing without in-situ data

• Published in: International journal of applied earth observation and geoinformation 2023-05, Vol.119, p.103308, Article 103308
• Main Authors: Xu, Yan, Cao, Bin, Deng, Ruru, Cao, Bincai, Liu, Hui, Li, Jiayi
• Format: Article
• Language: English
• Subjects: Analytical approach; Bathymetry without in-situ calibration data; Empirical approach; High-spatial-resolution image; Satellite remote sensing-based bathymetry
• ISSN: 1569-8432; 1872-826X
• DOI: 10.1016/j.jag.2023.103308

•This article proposes an analytical-empirical hybrid bathymetric approach.•It can estimate depths from the high-spatial-resolution satellite multispectral images.•It enables depth estimation from satellite imagery without in-situ depth calibration data. High-spatial-resolution satellite remote sensing (RS) images facilitate mapping fine-scale bathymetry over broad geographic areas using either empirical or analytical methods. However, inferring bathymetry from such images remains challenging in practical applications. For the empirical approach, high-quality in-situ depth calibration data that are required to establish a reliable empirical bathymetric model are either unavailable or excessively expensive. For the analytical approach, high-spatial-resolution RS images without an adequatenumber of spectral bands can be problematic in deriving depths from a bio-optical radiative transfer model (RTM). This paper proposes an analytical-empirical hybrid approach for estimating shallow bathymetry over broad geographic areas using optical high-spatial- and low-spectral-resolution satellite images without in-situ depth calibration data. In the proposed approach, the calibration data group that best matches the relationship between depths and the corresponding logarithmic blue/green band ratios at the time of image acquisition is identified by comparing a radiative transfer model-generated calibration data set to image-extracted reference data. Then a band-ratio bathymetric model that best fits the bathymetric images is constructed using the best-fit calibration data group. Lastly, depths are estimated from the high-spatial-resolution satellite multispectral images using the best-fit band-ratio model. Two types of high-spatial-resolution satellite images covering seven oceanic islands in the Yongle Group within the South China Sea (SCS) were used to test the proposed approach. The derived color-coded digital depth model (DDM) visually showed the depth distribution of shallow water areas around the islands. The accuracy assessment showed that the proposed approach performed well in shallow water areas, and can attain a bathymetric accuracy similar to those reported by the traditional satellite-derived bathymetric methods. The proposed approach can be used as an alternative for estimating depth when existing empirical models are not applicable due to a lack of in-situ depth calibration data.