Study of the spectral characters–chlorophyll inversion model of Sabina vulgaris in the Mu Us Sandy Land

As the dominant shrub community plant in Mu Us Sandy Land, S. vulgaris is key factor of ecological environment restoration It great significance to explore estimation and inversion content based on spectrum for evaluation intervention Land. The SVC HR-1024 portable feature spectrometer SPAD 502 chlorophyll meter were used study Land vulgaris. best band screened by correlation matrix method, vegetation index Structural Equation Modeling model, then model established different mathematical modeling methods. Results revealed that indices correlated, combining six 610–690nm 700–940 nm bands with highest correlation. In selection optimal index, NDVI, ratio mNDVI perform are suitable subsequent modeling. Of four models, partial least squares had fitting effect (R 2 > 0.91). univariate linear regression simplest processing procedure, but its accuracy was unstable = 0.1–0.9). multivariate stepwise also appropriate 0.8). stability BP neural network not high. Compare methods, PLS have their own advantages, higher, you can make a choice according demand as late method.