An empirical and dynamic tool for the prediction of forest fire spread using remote sensing and machine learning techniques

Fuel management and predicting flammable areas are the key to managing wildfires. These factors play a vital role in resource allocation, mitigation and recovery efforts. A forest fire can be a major ecological disaster, which has economic, social and environmental impacts on humans and also causes the loss of biodiversity. Therefore, it is important to understand the behaviour of fire ignition and spread so agencies can prevent and mitigate wildfires.

This project will use various models and machine-learning algorithms to develop a system that can be used for forest fire prediction. The fire risk probability model will be developed using a two-step analytic hierarchy process approach with cross-validations of support vector machine (SVM) model outputs. The prediction of fire points will be identified using an SVM model, taking elevation, slope, aspect, soil moisture layer, land surface temperature and vegetation type layers for training the data set. A Weather Research and Forecasting model will be used as a key input for meteorological data in the fire spread model.