A time-dependent stochastic grassland fire ignition probability model for Hulun Buir Grassland of China

Grassland fire is one of the most important disturbance factors in the natural ecosystems. This paper focuses on the spatial distribution of long-term grassland fire patterns in the Hulun Buir Grassland located in the northeast of Inner Mongolia Autonomous Region in China. The density or ratio of ignition can reflect the relationship between grassland fire and different ignition factors. Based on the relationship between the density or ratio of ignition in different range of each ignition factor and grassland fire events, an ignition probability model was developed by using binary logistic regression function and its overall accuracy averaged up to 81.7%. Meanwhile it was found that daily relative humidity, daily temperature, elevation, vegetation type, distance to county-level road, distance to town are more important determinants of spatial distribution of fire ignitions. Using Monte Carlo method, we developed a time-dependent stochastic ignition probability model based on the distribution of inter-annual daily relative humidity and daily temperature. Through this model, it is possible to estimate the spatial patterns of ignition probability for grassland fire, which will be helpful to the quantitative evaluation of grassland fire risk and its management in the future.