387. A Risk-Averse Multiobjective Stochastic Approach for the Prescribed Burning Problem
Invited abstract in session HB-7: Analytics for Prescribed Burning and Firebreaks Location in Forest Fires Prevention, cluster Use of Analytics in Forest Fires Management.
Thursday, 10:30-12:00Room: CE-210
Authors (first author is the speaker)
1. | Begoña Vitoriano
|
Dept. of Statistics and Operational Research, Interdisciplinary Mathematics Institute, Universidad Complutense de Madrid | |
2. | Javier León
|
Estadística e Investigación Operativa, Universidad Complutense de Madrid | |
3. | John Hearne
|
Mathematical Sciences, RMIT University |
Abstract
Prescribed burning is a widespread practice within forest fire management; it reduces the risk of ignition, but, above all, reduces the risk of fire spreading and makes it easier to control. In this model, the landscape is divided into cells and each cell’s vegetation age is used as a proxy. In particular, the connections between old-age cells and their length are defined as high risk spread links, and the cells’ areas and their ages as a proxy for ignition and wildfire intensity. Unfortunately, modifying the vegetation of a piece of land can negatively affect the existing fauna. Wild animals need vegetation with certain characteristics, and different species will have different requirements. Vegetation age can also be useful for representing animals’ preferences, usually modelled with concave piecewise linear functions. Additionally, the problem of determining where to carry out controlled burnings has great uncertainty, since it is impossible to know in advance how much forest area can be treated during the year, due to the limited time windows when they can occur. To collect all these aspects, a multi-objective stochastic programming model has been developed to determine when and where to carry out prescribed burning, taking into account future plans and landscape evolution. A risk-averse approach is adopted combining CVaR and OWA and solvable by linear programming without adding complexity to the restrictions further than those representing the system itself (following the methodology introduced in León, J., Puerto, J., Vitoriano, B. (2020) 'A risk-aversion approach for the multiobjective stochastic programming problem', Mathematics 8(11), 2026). The model is applied to a realistic case located in Andalusia, Spain.
Keywords
- Applications, Agriculture and Forestry
- Climate and Disaster Risk Management
- Multi-Criteria Decision Analysis
Status: accepted
Back to the list of papers