基于众源水面照片和透明度数据的水体吸收和后向散射系数反演方法研究

吸收和后向散射系数是十分重要的水体固有光学量，一般采用昂贵的大型专业水体光学仪器在水下实测，难以大范围开展。利用水色卫星遥感可以反演水体吸收和后向散射系数，但受到遥感空间分辨率、时间分辨率、云覆盖、大气校正以及水体浑浊程度等因素限制。吸收和后向散射系数与水体颜色指数和透明度之间存在紧密的物理联系。水体颜色指数和透明度测量简便，可通过众源获取的新型方式采集。因此，本项目将研究基于众源水面照片和透明度数据的水体吸收和后向散射系数反演方法，为吸收和后向散射系数提供新的数据来源。首先，研究基于手机水面照片的水体颜色指数计算方法，分析不确定性，并进行相机参数校正。然后，基于水体辐射传输机理，分析水体颜色指数和透明度与固有光学量之间的物理联系，基于这两个简单参量构建吸收和后向散射系数定量反演模型。最后，在不同浑浊程度地表水体检验基于众源数据的模型反演效果，分析不确定性，并研究降低不确定性的改进方法。

The absorption and backscattering coefficients are important water inherent optical properties. Professional hydrological optical instruments have been employed in the field to obtain in situ values of absorption and backscattering coefficients, but the large volume and expensive price of the instruments make it hard to carry out the in situ measurements on a large scale. Water color satellites can be used to retrieve the water absorption and backscattering coefficients, but the use is subjected to a set of conditions, such as satellite spatial resolution, temporal resolution, cloud cover, atmospheric correction and water turbidity. The absorption and backscattering coefficients have direct physical connections to water color index and clarity, which can be easily measured and obtained by the new crowd sourcing way. Hence, in this project, we will research on the retrieval method of absorption and backscattering coefficients from water color index and water clarity, which will create a new data source to the derivation of absorption and backscattering coefficients. Firstly, this project will research on the calculation method of water color index from the smartphone pictures, then analyze and reduce the uncertainties brought by the cameras and observation conditions. Then, the mechanical relations between the two water optical parameters and water inherent optical properties will be studied, and models will be constructed to retrieve water absorption and backscattering coefficients from water color and water clarity. In the final, the constructed models will be tested and validated based on crowd sourcing water color and clarity data, which are collected from surface water bodies on a large scale. To improve the performance and feasibility of the models, the model parameters will be tuned to water bodies with different turbidity.