Mapping invasive woody plants in Azores Protected Areas by using very high-resolution multispectral imagery

Abstract

We assessed the effectiveness of very high spatial resolution IKONOS imagery for mapping a top invasive woody plant, Pittosporum undulatum, in a Protected Area in S.Miguel Island. We developed a segmentation-based classification scheme. A strong separability between most important land cover classes and a high accuracy in supervised classification maps was achieved. Overall separability improved significantly after the training data depuration process. Support Vector Machine and Maximum Likelihood's supervised classifiers showed a strong agreement and a good accuracy at land-cover class level, especially with P. undulatum. This approach was confirmed as a cost-effective method to map woody plant invaders in Azores Protected Areas.