Wildfires are becoming more common. Largely due to climate change, the increasing occurrence of wildfires means we need to step up the tools we use to prevent damage or loss of life – one helpful tool is prediction models. The prediction models of the past haven’t been extremely accurate, but new research has shown promise.
Wildfire Prediction Model That Is 58-86% Accurate
Researchers have developed a new wildfire prediction model that is designed to help engineers make better decisions when fighting wildfires. The model, developed by engineers at Colorado State University, was tested on several real-world fires and proved to be accurate in predicting the probability of a building surviving a wildfire.
The model was tested on the 2018 Camp Fire and 2020 Glass Fire in California. Through this testing, the model proved that it could estimate which buildings will survive with 58-64% accuracy.
The model developers treated wildfire like a disease and uses graph theory, which is used to study the spread of disease.
The model is still in its early stages, but it could prove to be a valuable tool in the fight against wildfires. The researchers are now working on refining and improving the model. With some recent tweaking, they were able to increase the accuracy of the predictions to 86%
They believe that this could help save lives – not only by allowing firefighters to make more informed decisions, but also by helping people prepare for wildfires ahead of time.
User-Friendly Fire Risk Assessment Tools
The team at Colorado State University isn’t the only group of researchers looking to predict where wildfires will strike. A group of visiting students to the University of California, Berkeley has dedicated themselves to developing a prediction model using machine learning.
The visiting students, Sarthak Arora and Sai Venkat Ram Mallela, are focused on creating open source fire risk assessment tools. They want this tool to be a free resource for fire services, city planners, and developers when making decisions that impact fire risk.
The model considers lots of factors, like types of trees, slope, elevation, housing types and information, accessibility by firefighters, and traffic. The team is currently testing their model. In a recent test of their prediction model, they were able to predict where a fire would happen with 93% accuracy.
How Do Prediction Models Help Fight Fires?
At face value, the functionality of these models is pretty obvious. They can help direct firefighters on where to put their resources. But the useability of these models goes beyond just that.
Prediction models help us understand the various factors that affect a building’s ability to withstand fire. For example, it is important to know what materials are used in the construction of a building, as well as the type of construction that has been used. It is also necessary to understand the layout of a building, as this can play a significant role in its survivability during a wildfire.
These models help us understand what factors make a structure more or less at risk. This can help everyone, from town officials to homeowners, make better decisions to protect their home.