Wildfires are a common and devastating event that impacts ecosystems and communities yearly. Fire experts rely on simulations to better understand how to mitigate the damage they cause and respond to live threats. Many available fire simulation tools and libraries do not produce simulation results fast enough to be used with multiple parameter sets during an active fire and are difficult to integrate with other applications. To address this gap, we propose a GPU-based fire simulation and visualization tool: vFirelib. Using a GPGPU (General Purpose Graphics Processing Unit) framework, we can parallelize the fire spread computations and achieve a 20X computation speedup over a sequential implementation of a fire spread using a widely used fire spread model. To facilitate the integration with other applications, we implemented a wrapper including RESTful APIs to provide fire simulation as a service. In this paper, two examples are illustrated how to simulate wildfire scenarios and visualize results: a web-based application, and a 3D virtual reality application.
