Ecological multicriteria evaluation as fuzzy prior probability supporting forest type mapping on a regional scale

Sound ecosystem management depends on accurate, complete, and concise information regarding the extent, condition and productivity of natural resources. Distinctively, in order to define the elementary patch units within natural and seminatural forest ecosystems according to a "per habitat" approach, suitable reference can be made to forest types. The basic concept is to provide forest managers with an operational tool for discrimination of types of forests under different ecological-management conditions. High resolution remotely sensed data are not able, alone, to provide enough information for such a classification on a regional level. The present study deals with an experimentation carried out in central Italy for forest types supervised mapping based on the integration of multitemporal Landsat imagery (7 ETM+ and 5 TM) and spatial ecological information. The purpose of the study was to assess the effect of the inclusion of the prior probability information derived by forest type suitability analysis into the classification process, with the aim of developing the best classification method to automatically map forest types by remotely sensed images. MultiCriteria Evaluation (MCE) methodology was used to guide a modified maximum likelihood algorithm based on fuzzy logic. Classification accuracy is evaluated against a visually photointerpreted forest type map of the same area.