[Estimation of water clarity in offshore marine areas based on modified semi-analysis spectra model]

The main objectives of the research described in the present paper are to develop a semi-analysis model of water clarity for case 2 waters without inputting the absorption and scattering coefficient, which are not easy to be obtained for offshore marine areas so far. Based on the Zsd (Secchi depth)inversion theory, a simple semi-analysis spectra model was established for offshore seawater clarity by analyzing the relationship between vertical diffuse attenuation coefficient K(d) (490) and the beam attenuation coefficient c(490) with remote sensing reflectance. This semi-analysis spectra model needed two band reflectance ratios on- ly, while tidal correction was produced for this model to improve the precision of the retrieving results. The semi-analysis spectra model was applied to ASD hyperspectral reflectance data measured in the Pearl River Estuary Ecological Zone (October 21, 23, 2012, November 2, 2012; N=20) and the Xuwen Coral Reef Protection Zone (January 13, 14, 2013, N=25) which covered different water body of tidal times and different pollution sources. The results indicated that the changing tendency of predicted values was consistent with the synchronous measurement values after comparing them. However, water clarity calculated by the ASD hyperspectral reflectance measured in spring tidal time, generated 0. 4 m deviation compared with in-situ water clarity, while water clarity calculated by the ASD hyperspectral reflectance measured in neap tidal time is close to the in-situ water clarity. So the tidal correction coefficient of 0.4 was further applied for the model. After modification, the coefficient of determination between the inversed and measured water clarity was 0. 663, the average absolute error was 0. 14 m and the average relative error was 19.5%. Research demonstrated that this semi-analysis inversion algorithm just needs two band reflectance ratio to complete the inversion of water clarity, which is simple and works relatively well for lower clarity (less than 2 meters) waters compared to He' (2004) and Doron' (2011) algorithms.