A machine-learning (ML) wildfire prediction model uses social media and geophysical satellite data for more accurate real-time forecasting and monitoring. "Instead of having a network of cameras or climate sensors to track a wildfire, you can use a network of social media users or 'human sensors' posting information about a disaster in real time," said researcher Jake Lever from the U.K.'s Imperial College London (ICL), where the model was developed.
Researchers combined Twitter data with satellite data from the Global Fire Atlas to create the Sentimental Wildfires ML model, trained on social and physics wildfires data via the Sentimental Analysis textual content analysis framework. They tested the model using two 2016 datasets from the U.S. and Australia, and their findings imply that social media predicts wildfire activity.
From Imperial College London
View Full Article
No entries found