Wildfire Mitigation through Explainable Risk Predictions

The proposed innovation is build upon and improve the wildfire explanation and mitigation prediction capabilities developed in Phase I of this contract. Mitigations are actions like clearing defensible space around buildings or installing fire-resistant vents. Explanations describe the relative importance of each input to the wildfire prediction model, or feature, to the overall wildfire risk; we refer to these explanations as feature contributions. Feature contributions will be used to create localized mitigation guidance that will support more optimal resource allocation and better decision-making during pre-wildfire planning for the insurance industry and the public. In Phase II, we intend to build a more rigorous, machine-learning model to predict mitigation effectiveness. We will also expand the explainability of our model with better machine-generated narratives and risk visualizations in the Terrafuse web-based application and API. We will explore trend analysis as a way of explaining changes in risk caused by significant events. We will aid commercialization by making it easier to integrate our data into third-party platforms. Finally, we will replace NASA MODIS data set, which is nearing end-of-life, with NASA VIIRS data.