Abstract

Fluvial hazards of river mobility and flooding are often problematic for road infrastructure and need to be considered in the planning process. The extent of river and road infrastructure networks and their tendency to be close to each other creates a need to be able to identify the most dangerous areas quickly and cost-effectively. In this study we propose a novel methodology utilizing random forest machine learning methods and hydro geomorphic expertise to provide easily interpretable fine scale fluvial hazard predictions for large fluvial networks. The developed tools provided these predictions at reference points every 100 meters along the fluvial network of three watersheds within the province of Quebec, Canada and used variables focused on river conditions and to proxy hydro geomorphic processes such as sediment transport. Training/validation data was collected in four forms: field data, results from hydraulic and erosion models, government infrastructure databases, and hydro geomorphic evaluations using the 1-m DEM and satellite/historical imagery. First a subset of the reference points was manually classified then divided into training (75%) and validation (25%) datasets. Then the training dataset was used to train supervised random forest models. The validation dataset combined with extensive validation indices indicated the models were capable of accurately predicting the potential for hazards to occur. Metrics are extracted from the model to determine which variables are most important to predict each hazard. Finally, a methodology is proposed for a top-down hazard analysis of extensive fluvial networks to identify the most at-risk infrastructure/communities.