SUPPORT OF MULTISPECTRAL VERY HIGH RESOLUTION REMOTELY SENSED IMAGERY FOR OLD-GROWTH BEECH FOREST DETECTION

Abstract

In the Mediterranean basin human activity has modified landscapes for millennia, nevertheless there are few remote forest areas relatively untouched long enough from direct
anthropogenic disturbance to develop old-growth attributes. The aim of this note is to assess the potential of QuickBird (QB) satellite multispectral imagery for detecting old-growth forest
stands, considering as case study a Mediterranean beech forest in central Italy. The segmentation-based analysis of QB image proved to be a promising tool to detect scaledependent
pattern of forest structural heterogeneity. Values of remotely sensed attributes are compared in old-growth and not-old-growth stands: the statistical analysis showed that oldgrowthness is associated to the variability of multispectral reflectance from the image objects (polygons). Green band variability, notably, expressed by Ratio_band_2 has proven to be helpful for predicting old-growthness.