MODELING WILDFIRE SUSCEPTIBILITY IN ARKANSAS USING GIS-BASED MULTIPLE REGRESSION AND RANDOM FOREST

Wildfire is a common natural hazard that is influenced by both natural and anthropogenic factors. Contemporary changes in the landscape and the climatic conditions have resulted in a dramatic increase in the fire frequency and intensity in the United States. Arkansas is a natural state with 56% of its area covered by forest. Each year between 1981 and 2018, about 1,000 wildfires occurred and burned more than 10,000 acres in Arkansas. This study employs Multiple Linear Regression (MLR) and Random Forest (RF) methods to address the common natural and anthropogenic factors that influence wildfires and ultimately to model fire susceptibility in Arkansas. To investigate the relationship between the explanatory variables and the fire density, MLR has been applied to the 15 variables known to contribute to Oklahoma's fire density, and then the identified significant variables have been incorporated into RF to train the model for detecting wildfire-prone areas in Arkansas. Oklahoma’s severe wildfires have occurred under similar climatic conditions to Arkansas and RF has a higher predictive ability compared to MLR, thus they have been used for training the model. Among the 15 explored variables, potential evapotranspiration, soil moisture, Palmer Drought Severity Index, and dry season precipitation are found to be the most significant factors contributing to the fire density. The obtained R-square values from RF are significant, with 0.99 for the training regression and 0.92 for the validation. Our results show that the Ouachita National Forest and the Ozark Forest, in west-central and west Arkansas, respectively, have the highest susceptibility to wildfires. The southern part of Arkansas has low-to-moderate fire susceptibility, while the eastern part of the state has the lowest fire susceptibility. These outcomes will certainly support Arkansas’ fire preparedness plan to reduce its economic loss and save lives.