Watershed Prioritization Tool for Flood Risk Management

Following the floods of 2017, the government of Québec launched a reflection on flood risk management and planned a new mapping of areas at risk.

In this context, the ministère de l’Environnement, de la Lutte contre les changements climatiques, de la Faune et des Parcs (MELCCFP) set up the INFO-Crue project, which aims to map flood risk for southern Québec on a platform that is readily accessible to public decision-makers. Given the magnitude of the mapping task, it was necessary to determine the order in which the watersheds should be mapped. Moreover, the prioritization needing to be performed had to be based on scientific foundations and coordinated in keeping with governmental guidelines in flood risk management. Thus, the Université du Québec à Rimouski (UQAR) was mandated to develop a decision support tool for the planning process. The project team was inspired by the Analytic Hierarchy Process (AHP), a multi-criterion method, and developed a hierarchical criteria breakdown tree in collaboration with an interdisciplinary team representing various departments and organizations involved in flood risk management. In addition to the production of a web-based decision support tool, this project has made it possible to establish a summary portrait of the features and risks of the overall watersheds of southern Québec in the context of climate change.

Understanding and Assessing Impacts

At the beginning of the project, a thematic committee was set up to structure the project and to ensure that it was in line with the governmental guidelines regarding flooding. These guidelines namely included the protection of people, the mitigation of damages to material goods and the protection of natural ecosystem functions.

The first step in the process was to identify scientific literature and expert knowledge in order to establish and weight the prioritization criteria with the project team. A hierarchical structure with four core components and a range of sub-components and criteria was selected. These core components are the elements of risk established in the literature:

  1. Risk,
  2. Vulnerability
  3. Historical consequences
  4. Resilience

Based on historical data, the “risk” component allows us to validate the occurrence of flooding risk for each watershed. The notion of vulnerability is often associated with the sensitivity and readiness of the natural and humanized territory to suffer damage. The “historical consequences” component quantifies the actual impacts of past floods. Finally, measures to mitigate risks and improve community capacities to cope with the risk are reflected in the “resilience” component.

Furthermore, the approach required defining the concept of flooding that would be used, since it implies not only the natural phenomenon of a flood, but also its occasional impact on the humanized environment. According to the Ministry of Public Safety (MPS), flood risk is defined as the probability that a natural or man-made phenomenon or risk will occur in an area where material and human issues are present. The flood risk is therefore the product of the natural hazard itself (the flood), but also of the sensitivity of the issues that may be affected by this risk.

Identifying Actions

Through several work meetings, the project team began with a hierarchical breakdown of goals based on governmental guidelines. A top-down breakdown approach inspired by the AHP method was selected. Meetings were held with the thematic committee to weight the criteria using this approach, which is unique in its ability to structure a complex, multi-criteria issue in a hierarchical manner. This model includes in detail all the goals stemming from the governmental guidelines and breaks them down into sub-components that help explain each objective. A list of criteria is then created for each sub-component. The scores were then calculated, reviewed according to the consistency indices and validated by the thematic team.

In conjunction with this prioritization process, for each proposed criterion, the project team was also asked to consider a comprehensive list of indicators related to the availability of data for each of them. From this comprehensive list, a simplified list of indicators consisting only of those for which data would be available was considered. Data availability was assessed based on the purpose of the project and the future users of the final tool; for example, data should be available at the watershed scale and easily accessible to local government agencies.

Implementation

Following the conceptual development of the method, a computerized watershed prioritization tool has been developed. This tool is comprised of a series of integrated functionalities that cover the complete prioritization process by combining the AHP and TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) methods. It includes a knowledge base consisting of the hierarchy of risk analysis components and the pair comparisons resulting from the consultation exercise. The tool developed is flexible since it offers the possibility of selecting only certain elements of the component hierarchy in an analysis exercise. In order to verify the proper functioning of the developed tool, tests were conducted throughout the development process. The level of granularity of the selection tool is high. Indeed, it makes it possible not only to select components and sub-components, but also criteria and indicators.

In order to offer maximum flexibility in the use and rollout of the tool, it has been developed as a web application, based on development languages under open source licences (code/open) which ensure their sustainability.

Outcomes and Monitoring Progress

The end result of this project is the creation of a watershed prioritization tool. This tool is comprised of a series of integrated functionalities that cover the complete prioritization process by combining the AHP and TOPSIS methods. It includes a knowledge base consisting of the hierarchy of risk analysis components and the pair comparisons resulting from the consultation exercise. Since its launch with future users in the summer of 2019, this tool has contributed to the prioritization of watersheds to be mapped. This helped protect the population against flooding events and support adaptation to climate change through the development of a reproducible methodology that allows the acquisition of knowledge for better decision-making in the face of climate issues.

The project team considers that machine learning and deep learning in particular would allow for a better use of the collected data, in the context of an eventual improvement of the tool. These methods would address the large volume of data and the unconventional nature of this data that has not been considered in conventional hydraulic models, which requires appropriate models to fully exploit it. Deep learning algorithms would allow rapid processing of this data, while integrating human expertise in the development of the models.

Resources

Link to Full Case Study (in French only)