Estimating grassland coverage based on hyperspectral remote sensing in the northern Tianshan Mountains

By taking the Gan’gou Township in Urumqi County on the northern slope of the Tianshan Mountains as the study area, and using an America SVC HR-768 portable spectroradiometer, we collected hyperspectral data and corresponding grassland coverage data at 25 sites. Using these data, we analysed the correlation between grassland coverage and the original spectrum and the first-order differential spectrum and the hyperspectral characteristic variables. We also constructed a hyperspectral estimation model of grassland coverage by using regression methods based on the hyperspectral position variables, the hyperspectral area variables, and the hyperspectral vegetation index variables, and evaluated the accuracy of simulation models. The results showed that there was a strong correlation between grassland coverage and vegetation canopy reflectance in the spectral ranges of 354-704 nm, 1 420-1 481 nm, and 1 904-2 512 nm; the estimation models based on first-order differential spectrum and hyperspectral vegetation index variables can invert grassland coverage better. Model testing indicated that the first-derivate model based on 560 nm y=-384.153x+72.096 can be considered to be the best estimation model of grassland coverage, for which the root mean square error was 7.344% and the estimation accuracy was 90.343%.