A GIS-based assessment of land cover change in an Alpine protected areas

The European semi-natural landscape was markedly changed by the simultaneous effect of different processes. In this context aerial photos are still a valuable means to detect this change delineating the fine-scale pattern over large areas. Furthermore this remote sensed data are available in decades where other remote sensed data were still not available. Conversely data availability, software cost, image preparation and data mining affect the effective usability of this supports. In this case study we used only free and open source geographical software (GFOSS) focusing on image preparation to improve data mining phase. Aerial images of a mountainous protected area were orthorectified, smoothed and similar group of neighbouring pixels were grouped in meaningfull objects trough a semi-automatic unsupervised parameter optimization procedure. Pixel groups were manually labeleld creating a spatial explicit database. Using a priori defined minimum mapping unit (MMU), the different landscape configurations were compared showing a change in protected area. Implemented method ensured a strong repeatability and suitability over different aerial images and represented scene, but there are strong limitation in the use of this remote sensed data as data availability, an enormous amount of work for data pre-process and the need to easily automatize the classification step.