ABSTRACT
Derivative hyperspectral vegetation indices (dHVIs) are extensively examined for their capabilities in characterizing forest biophysical and biochemical parameters at different spatial scales. Pros and cons of these dHVIs are explicitly revealed based on simulated datasets generated from radiative transfer models (virtual experiments) as well as measured datasets. Results suggest that derivative hyperspectral VIs generally perform better than those based on the original reflectance but are more sensitive to noise, especially for higher-order derivative indices. With hyperspectral remote sensing data becoming abundant, first-order derivative VIs are probably a good choice for balancing accuracy and robustness in retrieving both biophysical and biochemical parameters and hence may play an increasing role in forest dynamics monitoring.
