Abstract: Inspired by combined modeling idea, we constructed a CM-BP (Combined Model-Back Propagation) for multispectral remote-sensing retrieval of river water turbidity by integrating the advantages of multiple remote-sensing retrieval models based on analyzing spectral characteristics of water bodies and selecting sensitive bands. With the in-situ observation data in 2012-2013 in typical reaches of the middle and lower Hanjiang River as a case study, we tested and assessed the applicability of the proposed CM-BP with domestic satellite data of high temporal and spatial resolutions as multispectral remote-sensing data sources. Furthermore, we compared the proposed CM-BP with traditional band-combined models in terms of retrieval precision, and meanwhile dissected the temporal and spatial distribution characteristics of the turbidity of Hanjiang River based on retrieval results. Results demonstrated that CM-BP boasts higher retrieval accuracy, applicability and stability than band-combined models. From the perspectives of spectral, temporal and spatial resolutions, multispectral remote-sensing data of Chinese domestic satellites like environment and disaster monitoring and forecasting small-satellite constellation, GF-1, and ZY-3 can be considered as optimal data sources for water quality retrieval of river. Remote-sensing data from Chinese domestic satellites meets the demand in high-accuracy, real-time and large-scale of remote-sensing retrieval of river water quality, thus providing a data basis for environmental monitoring studies of river.

Key words: turbidity of river water, Chinese satellite, multispectral image, remote sensing, combined retrieval model, water environmental monitoring of river