| 基于光谱分层的浅海水深遥感反演方法 |
| |
| 引用本文： | 楚森森,程亮,程俭,张雪东,刘晋铭.基于光谱分层的浅海水深遥感反演方法[J].海洋学报,2023,45(1):125-137. |
| |
| 作者姓名： | 楚森森 程亮 程俭 张雪东 刘晋铭 |
| |
| 作者单位： | 1.南京大学 地理与海洋科学学院，江苏 南京 210023 |
| |
| 基金项目： | 国家自然科学基金(42001401)；中国博士后科学基金(2020M671431)；中央高校基本科研业务费(0209-14380096)。 |
| |
| 摘 要： | 卫星水深反演是水深测量的一种重要手段，其中Stumpf比值算法和Lyzenga多项式算法应用广泛并诞生了大量改进算法，但这些算法没有顾及不同光谱的测深极限与适用范围，为此本文提出一种基于光谱分层的水深反演方法。首先，根据红、绿、蓝光谱对水体的穿透能力差异，提出一种基于影像本身的无参数光谱分层策略，提取红光层、绿光层、蓝光层；然后，根据不同光谱层的波段测深性能，分光谱层构建水深反演优化模型，获取浅海水深反演结果。以我国南沙海域长线礁和美属维尔京群岛巴克岛为实验区，本文方法对经典Stumpf比值算法和Lyzenga多项式算法进行改进后，水深均方根误差、平均绝对误差、平均相对误差分别降低了0.41～0.89 m、0.35～0.65 m、4%～19%，尤其在红光层，即水深较浅区域，平均相对误差降低了58%～149%，精度提升明显。因此，改进算法在提高卫星水深反演效果方面具有可行性和有效性。
|
| 关 键 词： | 多光谱 浅海 水深反演 Sentinel-2 Stumpf算法 Lyzenga算法 光谱测深 |
| 收稿时间： | 2022-05-04 |
Shallow water bathymetry using remote sensing based on spectral stratification |
| |
| Affiliation: | 1.School of Geographic and Oceanographic Sciences, Nanjing University, Nanjing 210023, China2.Jiangsu Provincial Key Laboratory of Geographic Information Science and Technology, Nanjing 210023, China3.Collaborative Innovation Center for the South Sea Studies, Nanjing 210023, China4.Key Laboratory for Land Satellite Remote Sensing Applications, Ministry of Natural Resources, Nanjing 210023, China5.Defense Engineering Institute Academy of Military Sciences, Beijing 100036, China |
| |
| Abstract: | Shallow water bathymetry using remote sensing is an important technology in marine geodesy. The Stumpf ratio and Lyzenga polynomial methods, as classic representative algorithms, are widely used and many improved algorithms have been developed. However, these methods do not take into account the bathymetry ability of different spectra. In this study, we propose a bathymetry method based on spectral stratification. Firstly, according to the difference in the penetration ability of red, green and blue spectra to the waterbody, an image-based nonparametric spectral layering strategy is proposed to extract the red, green and blue layers. Then, based on the band bathymetric performance of different spectral layers, we constructed a bathymetric inversion optimization model by spectral layers and further obtained the shallow water bathymetry. The Changxian Reef in the Nansha sea area and Buck Island Reef in U.S. Virgin Islands are selected as test cases to validate the proposed method. Compared with the classical Stumpf ratio and Lyzenga polynomial methods, the root-mean-square error (RMSE), mean absolute error (MAE) and mean relative error (MRE) of the proposed method have decreased by 0.41?0.89 m, 0.35?0.65 m, and 4%?19% respectively. Particularly, the accuracy of the red light layer (i.e., the shallow water depth) is significantly improved, and the MRE is reduced by 58%?149%. Our findings suggest that the proposed bathymetry method based on spectral stratification is feasible and effective in improving the accuracy of the shallow water depth estimation and has good potential for future applications. |
| |
| Keywords: | |
|
| 点击此处可从《海洋学报》浏览原始摘要信息 |
|
点击此处可从《海洋学报》下载全文 |
|
