Résumé
A risk assessment for disasters is usually composed of hazard, vulnerability and exposure variables, which are hardly studied and modeled simultaneously. In volcanology, it remains ambitious to anticipate risk trajectories of pre- and post-eruption regimes. The interdependencies and feedback loops of the system's components, between geological, ecological, social and economic ones, give rise to trade-offs and synergies that should be disentangled for supporting decision-makers and helping local communities to face the risks. We developed here an innovative discrete-event and possibilistic model based on a dynamical network representation to assess volcanological multi-risk and long term post-eruption impacts of such a multifactorial system. We illustrated our method with the region around Mount Meru (Northern Tanzania), a strato-volcano with various eruption styles, located in a growing economic and touristic region (>1 M.inh.). We used qualitative and rule-based Petri nets, largely unused in environmental sciences, for an integrated assessment of the overall system dynamics and associated risks. As a central result, we showed that the region could recover from a blast eruption, irrespective of the timescale. Our study highlights the fact that agriculture and pastoralism remain key activities to favour the recovery of this region. Yet, as soon as subsidies from governmental and non-governmental organizations are lacking, the modeled region remains isolated from national and international activities and shifts to rural dynamics. Our case study can equip environmental risk assessment with innovative models, new dynamical indices (e.g. desirable and non-desirable trajectories), and rigorous reasoning for an ultimate integrated management of social-ecological systems at stake.