A Probabilistic Multi-Source Remote Sensing Approach to Evaluate Extreme Precursory Drought Conditions of a Wildfire Event in Central Chile

Abstract

Forest fires are a major issue worldwide, and especially in Mediterranean ecosystems where the frequency, extension and severity of wildfire events have increased related to longer and more intense droughts. Open access remote sensing and climate datasets make it possible to describe in detail the precursory environmental conditions triggering major fire events under drought conditions. In this study, a probabilistic methodological approach is proposed and tested to evaluate extreme drought conditions prior to the occurrence of a wildfire in Central Chile, an area suffering an unprecedented prolonged drought. Using 21 years of monthly records of gridded climate and remotely sensed vegetation water status data, we detected that vegetation at the ground level, by means of fine and dead fuel moisture (FDFM), and canopy level, by means of the enhanced vegetation index (EVI) were extremely dry for a period of about 8 months prior to the fire event, showing records that fall into the 2.5% of the lowest values recorded in 21 years. These extremely dry conditions of the vegetation, consequence of low air humidity and precipitation, favored the ignition and horizontal and vertical propagation of this major wildfire. Post fire, we found high severity values for the native vegetation affected by the fire, with dNBR values >0.44 3 days after the fire and significant damage to the Mediterranean sclerophyllous and deciduous forest present in the burned area. The proposed probabilistic model is presented as an innovation and an alternative to evaluate not only anomalies of the meteorological and vegetation indices that promote the generation of extreme events, but also how unusual or extreme these conditions are. This is achieved by placing the abnormal values in the context of the reference historical frequency distribution of all available records, in this case, more than 20 years of remote sensing and climate data. This methodology can be widely applied by fire researchers to identify critical precursory fire conditions in different ecosystems and define environmental indicators of fire risk.