Use of High Resolution Satellite Images in the Forest Inventory and Mapping of Piemonte Region (Italy)

Abstract

The present work was carried out in the framework of the Forest Territorial Plans (PFT) of Piemonte region (North-western Italy). These Plans considered the different components of the territory (forests, grasslands, natural vegetation) analysed by means of a forest inventory and land cover maps at 1:25.000 scale. The general aim of the project is to integrate the traditional survey with the Earth Observation techniques, developing a tool useful to produce land cover and forest maps and providing a reference base for monitoring the future changes. A part of this project focused on the possible use of high-resolution satellite images (Ikonos II) for verifying and updating the forest maps. In order to achieve such a goal a multispectral Ikonos II image was acquired on the study area, an hilly terrain with highly fragmented forest cover. A first phase of ground truth definition by means of specific surveys aiming at recognising the different spectral classes was then carried out. The following steps were the developing of a visual interpretation of the image and a supervised classification applying an object-based approach. The results of these procedures were compared with the forest map defined by means of aerial photos interpretation and ground survey. The experimental application of high-resolution satellite images to forest mapping showed that the different homogeneous areas in the image corresponded to the main differences in cover density and dominant tree species composition (oak, chestnut, etc.), but sometimes they also reflected differences in the average age and sylvicultural system (coppice, high forest) of the forest stands. Another important conclusion of the study is that per pixel classification methods are not suitable for automated analysis of Ikonos images and therefore a new approach, based on a preliminary segmentation of the image, is needed.