| In fire prone ecosystems, wildfire plays an important part in the distribution structure and floristic composition of many vegetation communities. Wildfires also cause loss of natural resources, endangered species, human lives and property. A fire can be considered either as a prescribed burn, which serves management goals, or a wildfire, which tends to be unwanted and may require certain measures to be taken to control it. If a wildfire shows no foreseeable risk to human life and severe ecological effects, it may be monitored under conditions of prescription. The ability to understand, and even predict the fire direction, location, spread rate, and burned areas of wildfires is critical for wildfire management decision making and effective suppression implementation.;Current analytical techniques for wildfire growth prediction fail to address the following issues, (1) spatial and temporal variability, (2) neighborhood effects, and (3) influences of probability of fire occurrence on wildfire spread. This study addresses these problems by using an ARC/INFO Geographic Information Systems (GIS), spatial autocorrelation analysis, and logistic regression to create a probability model of fire occurrence and estimate the probability of fire occurrence in the Bee Canyon of San Jacinto Mountain where a major fire occurred in 1996. Based on a grid-based GIS and the empirical fire growth data, a probabilistic grid automation model is constructed to simulate wildfire propagation in an environment of heterogeneous conditions. This probabilistic grid automation model modifies the probability value according to wind direction, wind speed, slope gradient, and relative location to the burning cell. The grid automation model simulates the fire progression by comparing the modified probability value with generated random numbers to determine fire growth direction and location.;The grid automation model further incorporates Rothermel's model to estimate the fire spread rate and predict fire spread distribution depending upon predefined time. The simulated patterns of fire spread distribution are compared with the empirical fire propagation data collected from the Bee Canyon fire. The results indicate that the grid automation model could provide useful information to help the management decision making and suppression strategy of wildfires in a heterogeneous landscape. |
