Presenting in ELLIS-hosted AI for Good Workshop
I am happy to present in a very interesting workshop on July 5th about The role of AI in tackling climate change and its impacts: from science to early warning
My presentation is on the use of deep learning to forecast wildfires at different spatio-temporal scales.
Title: Wildfire danger forecasting at different spatio-temporal scales
Abstract: Climate change exacerbates the occurence of extreme droughts and heatwaves, increasing the frequency and intensity of large wildfires across the globe. Forecasting wildfire danger and uncovering the drivers behind fire events become central for understanding relevant climate-land surface feedback and aiding wildfire management. In this presentation the use of deep learning to predict wildifire patterns across different spatial and temporal scales is discussed. First, we showcase the high predictive skill of models that forecast the next day wildfire danger in the fire-prone Mediterranean, with improved performance over the traditional Fire Weather Index. We highlight the use of explainable Artificial Intelligence to diagnose model attributions and build trust on model predictions. We proceed to model wilfire danger at global scale and sub- seasonal to seasonal forecasting horizon. At such scales, the long spatio-temporal interactions of the Earth System variables need to be captured. We present deep learning models that handle the Earth as a system. Transformer-based architectures in particular, that make use of both local and global context, but also exploit time-series of teleconnection indices, perform best. The presented methodologies pave the way to more robust, accurate, and trustworthy data-driven anticipation of wildfires.
The slide deck will be uploaded here after the talk.