Study on the Inversion of Dissolved Oxygen in Water Based on HJ-1A Hyperspectral Data

The content of dissolved oxygen (DO) is a necessary condition for the survival of various marine organisms and an important water quality parameter for monitoring water pollution. Hyperspectral data is gradually applied to the inversion of water quality parameters due to its advantages such as high spectral resolution and wide spectral range. In this study, HJ-1A hyperspectral images were used to invert the dissolved oxygen concentration in the waters near Hong Kong. Firstly, the 2nd-level image product taken by the HJ-1A satellite's HSI sensor was converted into the appropriate format, and then atmospheric correction using FLAASH and geometric calibration were performed. Next, a correlation analysis was conducted on the measured dissolved oxygen concentration and spectral reflectance to select sensitive bands. Finally, the bands corresponding to reflection peaks and absorption troughs were selected for band combination, and a semi-empirical linear regression method was used for modeling. The model determination coefficient R2 was 0.639, indicating a relatively high accuracy of inversion.