Predicting Grassland Fire-Occurrence Probability in Inner Mongolia Autonomous Region, China

Fires greatly threaten the grassland ecosystem, human life, and economic development. However, since limited research focuses on fire prediction, it is necessary to find a better method predict probability of grassland-fire occurrence. Multiple environmental variables impact After selecting natural based remote sensing data anthropogenic variables, we built regression models probability, taking into account historical points in Inner Mongolia. We arrived at three methods identify drivers probability: global logistic regression, geographically weighted random forest. According results, forest model had best predictive effect. Nine selected by exercised spatially unbalanced influence fires. The all showed that meteorological factors normalized difference vegetation index (NDVI) were great importance In Mongolia, fires occurring different areas indicated varying responses influencing drivers, differed their geographical conditions fire-prevention periods. Thus, management strategy local should be advocated, existing fire-monitoring systems original improved adding fuels increase accuracy.