EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2821. Data-driven optimization of wildfire resource deployment
Invited abstract in session TD-31: Analytics for Combinatorial Problems from Health Care to the Food Industry, stream Analytics.
Tuesday, 14:30-16:00Room: 046 (building: 208)
Authors (first author is the speaker)
| 1. | Mostafa Rezaei
|
| Information and Operations Management, ESCP Business School | |
| 2. | Yasser Zeinali
|
| University of Alberta | |
| 3. | ILBIN LEE
|
| University of Alberta |
Abstract
Wildfires impose significant health, environmental, and social costs annually. This paper presents the development of a predictive and prescriptive framework for deploying three wildfire suppression resources: air tankers, helicopters, and firefighters. Leveraging fifteen years of historical data on wildfires in Alberta, Canada, we initially train machine learning models to predict both the daily number of fire occurrences and the required hours of suppression resources. Subsequently, we utilize the forecasts in an optimization procedure to ascertain the necessary number of suppression resources at various bases, aiming to minimize response time. The optimization procedure employs a queueing model to compute the response time probabilities for fires. Given resource shortages, the historical level of resource usage serves as a lower bound for the required resources. To address this, we incorporate survival analysis techniques into the forecasting models to account for data censoring. Our results can guide wildfire managers and decision-makers in enhancing the acquisition and allocation of fire suppression resources.
Keywords
- Analytics and Data Science
- Forecasting
- Forestry Management
Status: accepted
Back to the list of papers