An evaluation of the relative value of spectral and phenological information for tree crown classification of digital images in the Eastern Deciduous Forest.

摘要

Digitized, multitemporal, small format 35 mm aerial photographs were acquired, processed, and classified to determine the relative value of spectral and phenological information for tree crown classification of digital images of the Eastern Deciduous Forest. The one-hectare study site, located in a second-growth forest 15 km east of Morgantown, West Virginia, was photographed from a light aircraft ten times from May to October 1997 using both true color and false color film. The negatives and positive transparencies were scanned, and then rectified, enhanced, and classified using ERDAS Imagine. Differences in the timing of phenologic events between tree species, specifically leaf development and flushing, seasonal leaf characteristics, and leaf senescence, made it possible using this imagery to separate spectrally four deciduous tree species, namely Liriodendron tulipifera, Acer rubrum, Quercus rubra, and Quercus alba , from the surrounding vegetation at the study site. Optimally timed photography acquired during peak autumn colors provided the best single date of imagery while photography from spring leaf-out was the second-best. The best individual band of data for tree species discrimination was the blue band pass. Classifications using all four spectral bands (Red, Green, Blue, and Infrared) and four (05/23/97, 06/23/97, 10/11/97, and 10/30/97) of the nine dates provided the best classification accuracies. Variable canopy illumination made digital classification of individual trees complex. A Likelihood Ratio test of the data set revealed that the number of spectral bands included in the classification procedure (spectral resolution) significantly influenced the ability to correctly identify tree species. The number of dates (temporal resolution) became statistically significant only when the tree covariate was introduced into the model, effectively removing the effect of tree-to-tree variation on classification.