Une aide à la décision pour le choix des interventions en zone inondable A decision aid methodology for the selection of intervention in floodplain analysis

Résumé

Cet article a été réalisé dans le cadre d'un projet de trois ans visant à développer une méthodologie d'analyse, de prévision et de contrôle des risques d'inondation au Québec. Il présente une approche concrète pour calculer les impacts potentiels et le risque d'inondation et utiliser ces résultats afin d'évaluer la situation du risque local, de décider si les impacts doivent être minimisés et de choisir les moyens d'intervention appropriés. Le risque d'inondation est considéré comme étant le produit de la probabilité d'occurrence des crues et des conséquences occasionnées par ces événements. Les pertes de vies potentielles et les dommages directs sont évalués en simulant les niveaux d'eau de différents scénarios d'inondation à partir d'un modèle d'écoulement unidimensionnel non permanent, et en intégrant ces résultats à un logiciel géoréférencé de calcul des dommages d'inondation. L'analyse des impacts et du risque calculé permet de dresser un portrait du montant des dommages annuels potentiels sur les sites habités le long du cours d'eau et un portrait de l'évolution des impacts en fonction de l'amplitude des crues. Cette analyse mène à l'identification des sites où existe un risque jugé inacceptable selon des critères préétablis. Pour chaque site où des interventions sont justifiées par le niveau de risque, des scénarios de minimisation des impacts tenant compte des mécanismes d'inondation sont élaborés et ensuite simulés afin d'en mesurer l'efficacité. Un exemple d'application à un site de la rivière Châteauguay illustre la méthode et les gains pouvant découler de son utilisation.

Mots-clés:

• Inondations,
• risque,
• dommages,
• crues,
• décision

Abstract

Each year, several rivers in Quebec are responsible for severe flooding and these events generate major socio-economic impacts. The frequency and magnitude of these episodes highlight the existence of a real flood risk. Using global information concerning level and extent of flood risk, authorities would be more likely to make appropriate decisions in the management of flood risk. This article results from a three year project aimed at developing a methodology for the analysis, forecasting and control of flood risk in Quebec. It suggests a concrete approach for the evaluation of the potential impact of floods in order to obtain a better knowledge of local risk in inhabited areas and exploits there results to evaluate the acceptability of the calculated risk and to plan appropriate risk minimisation interventions.

Risk is defined as the product of the mathematical expectation of a specified occurrence with the expected consequences of the event. In floodplain studies, flood risk is the probability of the occurrence at a given flood multiplied by the expected consequences resulting from this event. Different types of consequences may be observed, clearly the easiest to evaluate being direct or material damages and potential loss of life. The risk calculated using the proposed definition is attributable in variable proportions to the frequency of the floods and the amount of damages. A given calculated global risk on a site could be the result of frequent floods, each causing moderate damage or of a single (or more) extreme event, with very low probability of occurrence, but causing severe damage. Risk associated with rare events could be considered as an acceptable risk, a risk we decide to live with, since the resources available to prevent flood damage are often limited and a decision is taken to optimise the allocation of these resources. The flood level corresponding to the limit between acceptable and unacceptable risk must be determined by the population concerned and be based on a good knowledge of the risk situation.

The proposed methodology to evaluate and minimise flood risk for a site localised in a river flood plain involves six steps:

1. the realisation of a hydrologic frequency analysis to determine the amplitude of the floods associated with the flood frequency,

2. the hydraulic simulation of floods to predict water level and velocity in the stream for each scenario,

3. the assessment of direct damage and potential loss of life for each flood simulated,

4. the calculation of risk,

5. the risk analysis considering the limit of acceptable risk and

6. the choice and planning of appropriate intervention to eliminate unacceptable risk.

This approach has been applied to the study of a site along the Châteauguay River, a tributary of the St Laurence River, a river that experiences flood events every two years or so. Seven flood scenarios (the 2, 3, 10, 20, 100, 1 000, and 10 000 year flood) are used to evaluate the risk for a site localised in the village of Huntingdon. Hydraulic characteristics, water level and velocity, associated with each flood scenario are determined using the DAMBRK model, a one-dimensional unsteady flow model. The results are incorporated in DOMINO, a geo-referenced software calculating flood impacts. This software allows the user to create a three-dimensional numerical model of the site based on topographic information. The superposition of hydraulic results provides the flow depth at any point within the site. Damage is evaluated by integrating the municipal roll number of Huntingdon, which provide the site location and value of each building, and gives an estimate of the population threatened by each flood event at the site. These results of direct damages are used to calculate the risk related to each flood event simulated on the Huntingdon site. For this application, the unacceptable risk has generally been agreed to be the risk resulting from the 20 year flood, or more frequent floods, for the material damages and to the number of potential losses of life associated with the 100 year flood or more frequent flood. The analysis indicates that an unacceptable risk of 23 993 $ per year for material damages and potential loss of life of 50 persons exist. Different site scale interventions to eliminate this risk have been simulated and proven to be efficient only if complemented with a few local modifications to the more exposed buildings.

This approach may be extended to the study of any river because it takes into consideration local hydrologic and hydraulic conditions. It has the advantage of being based on existing information and to be automated, which limits the time and resources required to obtain the base data and perform the necessary simulations.

Keywords:

• Risk,
• floods,
• decision making,
• damages,
• flood plains