These fire ignition sample points from 1980-2019 were used to train and test a fire ignition probability model for Santa Barbara county. We integrated multiple datasets to capture fire activity patterns over the last 40 years. For our fire ignition risk model, we used ignition points from the National Interagency Fire Program Analysis Fire-Occurrence Database within the study area, for the period 1992-2015. We augmented this database with other sources covering federal land to capture ignition points from 1980-1991 and 2016-2019: Federal Wildland Fire Occurrence Data were available for 1980-1991 and 2016, while FIRESTAT Fire Occurrence – Yearly Update covered 2017-2019. Ignition points from these 3 datasets within our area of interest totaled 2302 points, which we thinned to 500m minimum nearest neighbor distance (based on testing of best distance in the northern Sierra Nevada - Southern Cascades; Syphard et al. 2018) to increase spatial independence and reduce spatial autocorrelation and model performance inflation (Veloz 2009, Boria et al. 2014). This left 1353 points, which we divided into model training points (80%, n = 1081) and evaluation points (20%, n = 272).
We provide advanced conservation science, technology, and planning to empower our partners in solving the world’s critical ecological challenges