As the wildland/urban interface increases and the climate changes, the potential risk of fire damage to private property poses a major threat in the eastern United States. Knowledge of site-specific conditions that increase risk of wildland fires can be vital to managers and municipal agencies. Through risk mapping, we hope to identify areas that local agencies can focus efforts to reduce the risk of intense fires. Available soil moisture has been shown to provide a wealth of information about ecosystems, including long-term water-holding capacity and habitat suitability for ecosystem community types. An Integrated Moisture Index (IMI), a metric calculated from digital elevation models (DEM) and soil survey data, combined with long-term drought indicators (Palmer Drought Severity Index and Standardized Precipitation Index) is being used to investigate risk conditions of wildland fires in New Jersey, Ohio, and Pennsylvania.
Results/Conclusions
Fine-scale environmental data, along with records of drought indicators and reported wildfires, were used to inform a statistical model about conditions that potentially promote fires. RandomForest, a robust decision-tree based ensemble technique, was used to predict the risk of wildland fires. Preliminary results from PDSI data indicate that for the period 1895 to 1999 a greater number of droughts were experienced in New Jersey and Pennsylvania as compared to Ohio. However, during the recent decade 2000 to 2009, conditions have been considerably wetter or near normal for all three states. Although few droughts recently occurred, more were recorded in New Jersey and Pennsylvania than in Ohio, which corresponded with a higher occurrence of wildfires reported by New Jersey and Pennsylvania than compared to Ohio for this period.