Journal of Yangtze River Scientific Research Institute ›› 2020, Vol. 37 ›› Issue (11): 141-148.DOI: 10.11988/ckyyb.20191003

• INFORMATION TECHNOLOGY APPLICATION • Previous Articles     Next Articles

Remote Sensing Retrieval of Total Suspended Solids Concentration for Typical Reach of Hanjiang River Using Hyperspectral Data

XIAO Xiao1, XU Jian1,2, ZHAO Deng-zhong1, CHENG Xue-jun1, LI Guo-zhong1, ZHAO Bao-cheng1, XU Jian1   

  1. 1. Spatial Information Technology Application Department, Yangtze River Scientific Research Institute,Wuhan 430010, China;
    2. School of Water Resources and Hydropower Engineering, Wuhan University,Wuhan 430079, China
  • Received:2019-08-16 Revised:2019-12-16 Online:2020-11-01 Published:2020-12-02

Abstract: A hyperspectral model of retrieving suspended solids concentration in typical reaches of the middle and lower Hanjiang River is constructed based on effective information variable selection and neural network algorithm according to the water quality and hyperspectral data measured in 2012-2013. The performance and estimation effectiveness of the model are analyzed and assessed, and the distribution characteristics of suspended solids concentration in the waters of the study area are discussed. Results demonstrate that the hyperspectral retrieval model based on variable importance in projection index and neural network advantage has excellent retrieval accuracy, stability and adaptability; in contrast, for the single-band model and ratio model based on simple correlation analyses, the selection of modeling samples has a huge influence on the accuracy of the model, leading to poor retrieval accuracy, stability and adaptability. The concentration of suspended solids in the typical reaches of the middle and lower Hanjiang River varies within 18.8-187.0 mg/L. Seasonal differences are quite obvious: the concentrations of suspended solids in spring and summer are lower than those in autumn.

Key words: suspended solids in water, hyperspectrum, concentration of total suspended solids, remote sensing retrieval, variable importance in projection index, BP neural network, middle and lower Hanjiang River

CLC Number: