Modelling forest fire behaviour and mapping carbon emission in the Ludikhola watershed, Gorkha district, Nepal

Forest fires have become intense and more frequent in the last few decades all over the world. The profound impacts of forest fires have on atmospheric chemistry, biogeochemical cycling and ecosystem structure have led to the need to understand their behavior. Forest conditions, topography and weather greatly influence forest fire behavior and thus determine fire severity and consequently carbon emission. The estimation of carbon emission from forest fires is crucial for improving our understanding of the carbon cycle dynamics in order to develop strategies to curb the global warming - climate change problem. Previous studies have shown that modeling forest fire behavior can accurately give a good estimate of carbon emissions. The concentration of forest fire emissions varies spatially from one fire event to another as these emissions are a function of fire intensity, amount of biomass burned, and prevailing weather conditions. Therefore, this study is aiming at modeling forest fire behavior, and developing a method to estimate carbon emission in the rugged terrain of Nepal using 3 fire events that occurred in April 2008. A state of the art fire behavior model, FARSITE was used to simulate fire behavior in a spatially and temporally explicit manner taking into account the fuel, topography and prevailing weather in the Ludikhola watershed situated in the Gorkha district of Nepal. A WindNinja model was used to derive local winds influenced by vegetation and topography in the area. A combined approach involving a FARSITE model output parameter i.e. fire line intensity and a carbon emission estimation model developed in this research. The model was used to estimate fire induced carbon emission. The simulations were validated with the real mapped fire scar. The simulations using both uniform and spatially varying wind data, estimated the size of the real burned area with accuracies ranging between 78% and 96% for the three fire events were analysed.