Remote Sensing Detection of Wheat Stripe Rust by Synergized Solar-Induced Chlorophyll Fluorescence and Differential Spectral Index

摘要

Remote sensing detection of wheat stripe rust is important for agriculture management. In order to improve detection accuracy of the disease severity of wheat stripe rust, a detection method based on solar-induced chlorophyll fluorescence combined with differential spectral index was proposed in the paper. This method makes full use of the advantages of reflectance spectroscopy in detecting biochemical parameters and the advantages of chlorophyll fluorescence in photosynthetic physiology diagnosis. A characteristic dataset was collected from 13 differential spectral indices sensitive to the severity of wheat stripe rust, and different chlorophyll fluorescence values extracted by measuring radiance or reflectance respectively. The dataset was then processed using two different methods—partial least squares(PLS), and BP neural network—to carry out the remote sensing detection of wheat stripe rust severity. The results showed that: (1) The models based on the solar-induced chlorophyll fluorescence combined with spectral indices are more accurate than those based on differential spectral index. (2) The disease severity prediction model of wheat stripe rust constructed by BP neural network is better than PLS approaches. The research results of this paper have important significance for improving the accuracy of remote sensing detection of wheat stripe rust severity, and provide new information and a theoretical framework for remote sensing detection of other crop diseases.