Thesis

Many European countries have experienced unprecedented wildfires in the past decade, causing damage to ecosystems and infrastructure, as well as civilian deaths.

Insurance is a key tool for managing the many risks associated with climate change, particularly regarding private forests that are vulnerable to storms and wildfires. In France, this market remains limited in terms of both supply and demand, raising questions about its ability to adapt to these risks and about the role of public authorities. The scarcity of available data makes modeling essential for better understanding and guiding this market. Modeling could therefore be of great assistance.

In the past 20 years, European forests have suffered significant timber losses due to storms, wildfires, and bark beetle infestations, all of which are intensifying as a result of climate change. These cascading dynamics threaten the supply of forest goods and services, weakening the timber sector and compromising national commitments to carbon sequestration.

Sustainable forest management is increasingly threatened by a growing number of natural hazards, whose frequency and intensity are intensifying as a result of climate change. These disturbances—drought, parasites, wildfires, and more—interact with one another, creating cascading dynamics that increase uncertainty for forest owners. Understanding these mechanisms and the resulting behavioral aspects of decision-making is becoming essential for adapting forest management in a context of multiple and interconnected risks.

The intensification of human activities is degrading the quality of our water resources, particularly through the excessive influx of nutrients from agricultural and industrial watersheds. This accumulation contributes to the eutrophication of lakes and the proliferation of cyanobacteria. This phenomenon raises concerns on several fronts: ecological, economic (due to the closure of fishing and swimming areas) and public health. The purpose of this thesis is to develop modeling tools to manage and anticipate such situations.

This study aims to improve the early detection of food crisis risks by using text sources, an area that remains largely unexplored. In particular, it focuses on assessing and representing risk factors that are often diffuse or delayed over time. This innovative methodological approach is ultimately intended to strengthen the effectiveness of existing early warning systems.