Un modello di valutazione della funzionalità protettiva del bosco per la pianificazione forestale: la componente stabilità dei versanti rispetto ai fenomeni franosi superficiali

Abstract: The evaluation of forest protective function has been divided into four main branches according to the different types of instability phenomena (landslide events, erosion, floods, avalanches). This paper presents the first module related to landslide events and will be followed by others. Two opposing factors have been considered: instability tendency of a selected land unit and protective function of the vegetation. While the first factor tends to promote landslide events, the second one opposes their occurrence. An expert evaluation of these aspects allows us to derive some qualitative indexes. For each territorial unit, these indexes express the protection value of vegetation, the degree of management restrictions and suitable improvements to the forest cover. Both propension and protective functional character are influenced by many and different parameters which require multidisciplinary competences to correctly evaluate them. Such skills are not easily found in technicians in charge of forest management. The present paper aims to provide a decision making support tool; based on neural network models, it should be able to interpret and simulate the expert knowledge, extending to the “standard” forest technician the opportunity to perform such a kind of evaluation. The neural network training required the identification and characterization of explanatory variables related to both aspects, and the subsequent expert definition of the respective datasets of real classification examples. The descriptive variables were chosen considering the information availability and its compatibility with GIS techniques as well. Model performances have been validated by testing the whole procedure in two sites situated in Antrona valley (Piemonte) and Acqualagna (Marche). The results are discussed in detail and put in evidence a good accordance with both field survey and general conceptual assumptions. Future developments will involve similar analysis and modelling applied to other instability categories. An overall index for the protective function of the forest cover will also be produced.