Using Hyperspectral Indices to Diagnose Severity of Winter Wheat Stripe Rust

Hyperspectral remote sensing data, the spectral resolution of which is less than 10 nm, include a mass of diagnosing spectrum information and can be used to non-destructively detect disease stresses of vegetation. A field experiment had been done to collect the canopy spectral reflectance of winter wheat. Canopy reflectance spectra were surveyed, and the disease index (DI) were gained by the field survey in the same regions. The results shows that while DI increases, the spectral reflectance of the canopy gradually increases in the red valley region. And the coefficient of vegetation index SD_r/SD_g and DI is the largest. Here, SDg and SDr denote the sum of 1st derivative within green edge and red edge respectively. Linear and non-linear regression methods are used to build the inversion models. Testing results indicate that the models consisted of SD_r/SD _g has the highest predictable precision. So the conclusion of this paper is that the model is the best one for inversion about severity of winter wheat yellow rust by hyperspectra. The conclusion has important mean to guide people to cure the disease of crops and increase yields of crops and ensure security of food supplies