Using pixel and object based IKONOS image analysis for studying decay in silver fir stands

Abstract

The problem of old silver fir stands decay has appeared in the last decades of the last century with symptoms such as defoliation, wood decay and parasites attacks. The problems in monitoring this phenomenon is that the defoliation and mistletoe attack appear on the crown top, triggering the coronation process - tree develop branches on their lower stem part in order to resist to the defoliation. This reaction makes the attack difficult to notice on aerial or satellite images, due to the reflection of light on the lower branches. The objective of the article is to find a research methodology for identifying and even mapping the mistletoe attack phenomenon at a tree or stand level. For a tree-level analysis, a comparison between the data obtained in a 1 ha sample plot and the data extracted from an IKONOS satellite image has been used. The simple spectral response is less correlated at tree level with the defoliation and mistletoe attacks. We found a very significant correlation in the infrared channel but with low correlation coefficients. Better results have been obtained in a stand-level analysis. In order to improve the separability, the methodology for extracting the remote sensing data did not resume only to the mean spectral response, but we also performed a variability and texture analysis. The obtained correlation coefficients were around 0.7, very significant for the data used in the research. The results obtained with the texture analysis are also related to the biometric characteristics of the stands, mistletoe attacks occurring in stands with a low canopy closure index and is usually accompanied by distorted crowns. The model of forest health analysis should also be tested in similar conditions in order to validate and apply on a large scale inventory.

Keywords

• forest health, silver fir, ikonos, image texture