Protecting human societies and biodiversity from wildfire risks
Focus on project 1
- First, I attempt to review and compare fire spread models using cellular automata rather than vector-based models (differential equations), on fictional landscapes for ground fires
- Second, I leverage rich data environments to assess actual performances of different models
- Third, I will try to develop additional modules of fire spread, notably on crown fires, which characterize the current fire regimes.
- Fourth, build on recent advances in optimization, computer science and machine learning to increase the accuracy of the model
- Fifth, based on statistics, notably Zero Inflated Poisson Regression, try to predict the number of ignitions, and allocate them on the landscape. This procedure will follow a prescription rule based on the gradient of the regression.
- First, review of living species in the case study forest land
- Second, learn how to model the species and learn the various aspects of biodiversity
- Three, simulate behavior of species considering habitat (amount, quality, connectedness...)
- First, map margins of operations to their impact in terms of modeling component : this can mean literature review, or prospective analysis.
- Second, simulate the impact in terms of burnt area, severity, intensity, and try to come up with an updated version of a burn probability
- Third, get data on costs and map to a cost function.
- Fourth, get data on losses and avoided losses and update burn probabilities