Mapping the Causes of Forest Fires in Portugal by Clustering Analysis

This paper presents a spatial characterization of the distribution at district level of the forest fire events that occurred in mainland Portugal between 1996 and 2015 and whose causes were investigated. We further examine the breakdown of the causes of these forest fires over this period. Results supported by relevant validated statistics show that of the total fire events recorded, 94.4% were identified as an effective occurrence, of which 22.2% had burned an area greater than 1 ha, and of these only 42.1% were investigated. False alarms or fires without a recorded burning area are more significant in the districts of Aveiro, Lisbon and Porto, the biggest municipalities. Of the fires whose causes were investigated, the largest number of recorded events were in NE regions (49.0%), followed by NW regions (41.7%), and finally in the rest of the country (9.3%). Taking into account the ratio between the investigated fires and the total number of fires and the behavior profile produced for cluster analysis, a different panorama is brought to light, with the center and south regions showing greater effort to investigate the fires. A thorough analysis of the causes and motivations of the ignition of these forest fire occurrences showed that human activity, either deliberate (20.4%) or negligent (29.9%), outweigh natural phenomena (0.6%). Reactivations (14.6%) and Unknown (34.5%) causes decreased as time passed, whereas negligent and deliberate causes increased. However, these results could change if the percentage of unknown information in relation to the origin of the forest fires is considerable. The outcome of this research will support an efficient management related to fire mitigation and suppression including establishing preventive actions to reduce the occurrence of forest fires and emphasize the need to improve the procedure for recording forest fire events in Portugal, especially in relation to identifying their causeThis research was partially funded by CERENA (strategic project FCT-UID/ECI/04028/2019) and CEGOT (FCT-UID/GEO/04084/2013) and partially co-financed by the European Regional Development Fund (ERDF) through the COMPETE 2020—Operational Programme Competitiveness and Internationalization (POCI).