Resilient Forest Road Networks for Wildfire Management

The Problem

```

In Mediterranean countries, wildfires are among the most serious environmental threats. They destroy ecosystems, threaten rural communities, and generate significant economic and social losses.

One of the key factors determining the effectiveness of wildfire suppression is road accessibility. Firefighters must be able to reach affected areas safely and rapidly in order to control fires before they spread.

Many forest road systems were originally designed for forest management activities such as silvicultural treatments, monitoring, and timber extraction. During wildfire events, however, these roads become critical access routes to remote areas.

Existing road networks often present important limitations:

• Some forest regions remain difficult or impossible to access.
• Many routes consist of narrow roads unsuitable for emergency operations.
• Dead-end roads can become dangerous traps under extreme fire conditions.
• Road systems are rarely designed with wildfire resilience as a primary objective.

Designing road networks that minimize environmental impact and construction costs while maximizing accessibility, safety, and resilience represents a challenging mathematical and engineering problem.

```

The Modeling Approach

```

The project aims to develop innovative optimization models and algorithms capable of simultaneously:

• Minimizing road construction and maintenance costs.
• Reducing environmental impacts on forest ecosystems.
• Guaranteeing access to all relevant forest areas under normal conditions.
• Maintaining functionality when parts of the network become unusable during wildfires.

To achieve these goals, the project will employ advanced mixed-integer programming models capable of handling uncertainty in wildfire occurrence and propagation.

Rather than relying on detailed probability distributions, the methodology uses robust optimization techniques that remain effective under a broad range of possible fire scenarios.

Multiple wildfire scenarios will be generated and analyzed through simulation. These scenarios will support the application of distributionally robust optimization, a state-of-the-art framework in operations research designed to produce solutions that remain reliable even under highly uncertain conditions.

The project combines:

• Mathematical optimization.
• Network design models.
• Algorithm development.
• Machine learning techniques.
• Wildfire simulation methodologies.

Together, these tools will provide a robust scientific foundation for future applications in forest planning and wildfire management.

```

The Team

```

The project brings together researchers from CEMS.UL with complementary expertise in optimization, forestry, and mathematical modeling.

• Miguel Constantino – Department of Mathematical Sciences, Faculty of Sciences, University of Lisbon.
• Agostinho Agra – Department of Mathematical Sciences, Faculty of Sciences, University of Lisbon.
• Marta Mesquita – Instituto Superior de Agronomia, University of Lisbon.

The team has extensive experience in mathematical optimization applied to forestry and environmental systems, including mixed-integer programming, network design, and robust optimization.

```

Our Needs

```

To successfully carry out the project, we plan to recruit:

• One Master's student to develop and calibrate the wildfire simulation framework.
• One Postdoctoral Researcher to formulate, implement, and test optimization models and solution algorithms.

Financial support is sought to fund these research positions, provide computational resources, and support the dissemination of scientific results.

```

Expected Impact

```

By supporting this initiative, partners contribute to research that combines mathematics, environmental sustainability, and wildfire resilience.

The outcomes of this project have the potential to improve wildfire preparedness, protect forest ecosystems, enhance the safety of rural communities, and support more sustainable forest management strategies for future generations.

```