关键词: GF5-AHSI数据, 水深遥感反演, 最优标度回归模型, BP神经网络模型

Abstract: Hyperspectral remote sensing water depth inversion which has many advantages such as convenience, quickness, and economy is a supplement to traditional measurement methods of water depth, and is worth studying. The research area of this study is located in Hengsha, Shanghai, which is a typical shallow water area of tidal flats. The research data was obtained from GF5-AHSI hyperspectral remote sensing and measurement of water depth at the same time. The modeling parameters were extracted through data transformation and correlation analysis. The single-band ratio model, multivariate linear regression model, optimal scale regression model and BP neural network model were used for water depth inversion in this area. By comparing and verifying the accuracy of the four models, it is found that the optimal scale regression model is better than the other three models, with R2 reaching 0.972 and RMSE of 0.47 m, which is suitable for inversion of shallow water depth in Hengsha.

Key words: GF5-AHSI data, water depth inversion, optimal scale regression model, BP neural network model