Species analysis and identification of spectral characteristics on Seriphidium transiliense desert grassland

In this research, the portable spectrometer was used to measure the spectral characteristics of 9 plants on the northern slope of the Tianshan mountains middle Seriphidium transiliense desert grassland. The first derivative, smoothed, normalization, and Rr-Rn of spectral characteristics were analyzed to identify and establish discriminant model for different plants. The results showed that the original spectral reflectance of different plants have some differences in specific bands, a first derivative can accurate extract the red edge position of plant characteristic parameters, and smooth process have an important role in removing noise, and the normalization in favor of the spectral characteristics of plants. However, difficult exists by using Rr-Rn in terms of the grass plants distinction. Using the Rn, REP, Rg, MSAVI, CRAI five spectral indies, the established discriminant model can discriminant of nine kinds of plants and the accuracy reached 90.5%.