Provisional Spectral Analysis Sample Clauses

Provisional Spectral Analysis. Six plots were selected as training sites to develop a base for deciduous vs. conifer signatures. The plots were based on two factors. First, the number of trees in the sample plot are relatively well-predicted by the TDA, which provides a more stable base for individual comparison. Second, the tree composition is either conifer or deciduous-dominant. These sample plots provided a basis on how to construct the modelling approach. Several rudimentary parameters from ITOs were evaluated in order to determine systematic differences between conifer and deciduous compositions. These include: - Mean NDVI, Red, Green, Blue, and NIR spectral values - The standard deviation of NDVI, Red, Green, Blue, and NIR - Intensity, Hue, Brightness Many of these variables show no statistical significance between conifer and deciduous data sets, or demonstrate collinearity within regression analyses. The predictors that do show promise are the Red, Green, and Intensity variables. Deciduous trees show a far greater variation within the Red and Greed spectral values [Appendix E]. To explore this, the standard deviations of these bands were used as broad thresholds to classify respective tree objects as conifer or deciduous. The following figures serve as visual comparisons. The table shows the tree object counts for respective standard deviation values.