Objects-based Image Analysis for Mapping Natura 2000 Habitats to Improve Forest Management

Abstract

Natura 2000 is a European network of protected areas established under the Habitats Directive (92/43/EEC). According to the Habitats Directive, habitat maps must be periodically updated, which requires the development of cost- and time-efficient mapping practices. In this study, we propose a methodology for habitat mapping using very high spatial resolution (QuickBird) images with Object-Based Image Analysis (OBIA). We classified five segmentation levels: level 5 incorporated the prior knowledge of the study area into the analysis; level 4 and 1 were used to identify arable areas and land covers, respectively. The information contained in levels 1, 4, and 5 was then combined to classify plant species in level 2. Finally, habitats were classified in level 3 using level 2 class-related features. The habitat map obtained had an overall accuracy of 86.3 percent. Classification accuracies were higher for tree-and pasture-dominated habitats than for shrub-dominated habitats.