Probabilistic Forecasting of Wildfire Size Using Conditional Flow Matching
This work proposes a probabilistic model using conditional flow matching to predict wildfire size over time, focusing on acreage rather than spatial progression. The model forecasts changes in fire area over 3-hour intervals using current and previous area estimates alongside weather and terrain data. It achieves high accuracy (sMAPE of 0.39) for single steps and provides uncertainty quantification for longer forecasts.