BLM REA COP 2010 Current Probability of Human-Caused Fire Occurrence

This dataset shows an estimate of the probability of human-caused fire occurrence, based on 30 years of occurrence data using a MaxEnt model based on several factors including distance to roads, urban areas, vegetation type, and climate. Model performance was poor to fair, with an AUC of 0.674 Significant predictive factors include distance to recreation areas, annual precipitation, distance to roads, summer precipitation, and existing vegetation type. Caution should be exercised in interpreting this dataset, as it is based on an association between landscape factors and the locations of fire occurrences. This dataset does not provide information about the likely outcome of a fire. The location data in this dataset has inherent sampling bias (only includes reported fires, and is generally limited to public lands), and locations may not be precise. Factors were selected based on assumed relationships to patterns of fire occurrence that were also easy to obtain or process; these factors do not necessarily have causal influences on where fires actually occur. The specfic association to site-characteristics may not be particularly precise, but instead serves as a measure integrating both human access to a particular part of the landscape, and the chances of ignitions occurring where humans do have access.