Quantitative model of foliar dustfall content using hyperspectral remote sensing

By analyzing hyperspectral features of elm foliar dustfall content (FDC), a models of hyperspectral monitoring was built. Relationship between hyperspectral parameters and FDC was investigated by using regression analysis method. The results showed that FDC increased spectral reflectance in the visible band while decreased it in the near infrared band. Foliar dust didnt affect the "three edge" position but significantly affected its amplitudes and areas. FDC of elm was badly predicted with the models based on spectrum index or "three edge" parameter. Models based on multivariate linear regression, principal component regression and partial least squares regression can predict FDC primely. The model with 1st derivative value as variables was the best one for estimating FDC by the hyperspectral. Predictive correlation coefficient, predictive root mean square error, and the ratio of sample standard deviation to predictive root mean square error of this model were 0.92, 1.06, and 8.2, respectively.