|
Assessing the risk of wildfires over the entire globe can be crucial in avoiding harm to wildlife, economy, properties and humans. This is known to be a challenging task. Here, a machine learning model is trained on a dataset composed of remote sensing data variables such as topography, vegetation and weather. The model is able to assess the risk of fire with a spatial resolution of 1000m/pixel. It achieves optimal results compared to other state-of-the-art architectures. Most of the variables in the dataset are found to be critical for the task, while few were disregarded. Particular focus has been given to collecting data across a variety of landscapes. Specifically, samples from Africa, Australia, Asia, Europe, South America and the US are included. This research shows the potential for deploying global wildfire risk assessment applications. |
