ABSTRACT OF THE DISSERTATION Simulating the Spread of Wildfires Using Geographic Information Systems and Remote Sensing By Jianping Xu Dissertation Director: Assistant Professor Richard G. Lathrop, Jr. Fires burn a large number of acres of wild land annually. Ecologically, recurrent fires are essential for maintaining certain kinds of vegetation communities as well as an important component in nutrient cycling. However, from a management point of view, wildfires also threaten human development and valuable timber resources. The overall goal of this dissertation research is to improve the modeling of wildfire behavior, to better understand the interaction of fire behavior/environment interactions and assist fire fighting operations. The current status of research on fire behavior and its spatial modeling is reviewed. The inadequacies of current methods of modeling spread phenomena in raster-based geographic information systems (GIS) are discussed and a new method is developed. Based on the new spatial modeling approach a wildfire spread simulation model, WiSpS, is developed. WiSpS consists of two modules: a fire media module and a fire motion module. The media module, based on the U.S. Forest Service BEHAVE system, produces four mapped outputs: the forward rate of spread (ROS) of fire; the direction of the forward ROS; the perpendicular ROS; and the maximum potential spotting distance. The media module serves as input to the motion module. The latter is essentially a shortest path algorithm that has been adapted for a raster GIS environment. A non-adjacent cell link approach has been invented to solve the problem of inadequate fire shape simulation. The motion module provides a visual display of fire spread and a map of burned locations with elapsed times. WiSpS directly simulates surface fires and spotting. WiSpS can be configured to handle crown fires, multiple phase fires, and to trace fire spread paths. WiSpS has been tested against five major fires in the New Jersey Pinelands and one in the northern Rocky Mountains using primarily remotely sensed data along with documentary data. Without considering fire fighting activities, of the total area covered by either the actual burn or the simulated bum, 65.2% was both burned and simulated, 19.8% was burned but not simulated, and 15% was not burned but simulated. This dissertation research shows that wildfire spread can be graphically simulated with reasonable accuracy and that modeling with a raster data format has great potential.