hide
Free keywords:
-
Abstract:
Forest fires pose a growing threat worldwide, causing damage to ecosystems and releasing significant amounts of carbon. We analyze a national-scale forest fire susceptibility over the past two decades at a sub-decadal level in Nepal. We utilized earth observations and the Random Forest machine learning algorithm within the Google Earth Engine framework to analyze forest fire susceptibility on both spatial and temporal scales. A range of terrain- and climate-related variables were used to train and validate the random forest machine-learning model. Our results show that ongoing and projected changes in weather, land-use and human interventions will likely impact the severity and extent of forest fires in the nation. We estimate that forest fires could potentially release more than 170 million tons of soil organic carbon and 325 million tons of above-ground wood carbon with parallel biodiversity loss in Nepal alone, thus requiring forest management and fire mitigation efforts in the region.
