VFire: Immersive wildfire simulation and visualization

Abstract

The destruction caused by wildfires has led to the development of various models that try to predict the effects of this phenomenon. However, as the computational complexity of these models increases, their utility for real-time applications diminishes. Fortunately, the burgeoning processing power of the graphics processing unit can not only mitigate these concerns but also allow for high-fidelity visualization. We present VFire, an immersive wildfire simulation and visualization system. Users are placed in a virtual environment generated from real-world data regarding topology and vegetation. There they can simulate wildfires in real-time under various conditions. They can then experiment with various suppression techniques, such as fire breaks and water drops. The simulation is performed on the graphics card, which then provides visualization of the results. The system is intended to train fire chiefs in planning containment efforts and to educate firefighters, policymakers, and the general public about wildfire behavior and the effects of preventative measures.

Introduction

Every year, wildfires destroy millions of acres of land and cost millions if not billions of dollars to control. From 2000 to 2002, over 18 million acres were burned, over 2000 structures were destroyed, and over 3.4 billion dollars were expended just for suppression efforts. Beyond the immediate damage caused by wildfires, there are also lingering effects that are not only environmental, but social and economical as well [1].

To better understand wildfires, scientists have developed methods for modeling wildfire behavior. These models vary in terms of the types of spreading behaviors simulated and the simulation domain used [2], [3]. Some models simulate the spread of fire across surface fuels while others simulate fire propagating into the canopies and possibly moving from tree top to tree top without interacting with the surface fuels. Still others simulate embers being lofted through the air and igniting spot fires potentially vast distances away from the source. These models take into account a variety of factors including wind, weather conditions, fuel types, and slope.

Visualization of wildfire behavior can provide a number of benefits. First of all this allows scientists to verify the accuracy of these models by comparing the results of an actual fire with the output of a simulated version. Once the model is validated, it can then be used to predict not only the behavior of an existing fire, but also the consequences of preventative measures, such as vegetation thinning and prescribed burns.

Displaying these predictions in a visually informative manner allows for community planners and/or city officials to better educate the public on existing fire hazards. Furthermore, enabling interactive manipulation of the simulation along with the visualization allows for training of fire bosses with respect to resource allocation and fire behavior. While experimentation would be dangerous and costly to perform in a real-life situation, these risks can be mitigated by simulating untested approaches first.

Virtual reality (VR) technology allows users to immerse themselves in their data. Stereoscopy provides depth information that is usually lost with standard desktop displays. This depth information enables the creation of 3D user interfaces, which can provide a more intuitive form of interaction. Additionally, combining depth information with high-fidelity graphics can allow an observer to better compare a simulated fire to a historic fire.

VFire is a wildfire simulation and visualization tool built for an immersive virtual environment. Users are able to load data about a geographical region and then experiment on this region by starting fires, manipulating fuels, and altering weather conditions. Fire simulation is performed on the graphics processing unit (GPU) to enable real-time simulation and visualization.

The remainder of this paper is structured as follows. Section 2 provides background information, while Section 3 gives an overview of the system. Section 4 outlines how the fire simulation is performed on the GPU and Section 5 describes the methods used to visualize the simulation. Section 6 discusses the interaction of the user with the system while Section 7 details the current state of the system. Finally, Section 8 offers closing thoughts and avenues for future research.