出版社：Grupo de Métodos Cuantitativos, SIG y Teledetección, A.G.E
摘要：This study focuses on wildfire susceptibility mapping in Madeira Island (Autonomous Region of Madeira, Portugal) and achieves this by using information value method and data processing in a Quantum Geographic Information System (QGIS) Plug-in programmed in Python. By means of a QGIS Plug-in a sensitivity analysis to the predisposition factors used in the construction of models of susceptibility to wildfires in Madeira Island was performed. For this analysis the burnt areas from 2006 to 2013 were related with some widely used variables in wildfire susceptibility modelling, namely land use, elevation, slopes, precipitation, humidity and temperature. The overall performance achieved by the information value method was assessed through success rate curves. The susceptibility models obtained with different combinations of variables present a very similar and unsatisfactory explanatory capacity. The different variable combination provided by the analysis demonstrated that land use and elevation play a major role in wildfire susceptibility explanatory capacity in Madeira Island. Additionally, the development of the QGIS Plug-in allowed to obtain quick and reliable results, proving to be extremely useful in the sensitivity analyses performed.
