Hydraulic flood routing with minimal channel data: Peace River, Canada

Abstract

Hydrologic flood routing models have been, and continue to be, the primary tool of the flood forecaster. However, any advancement in our ability to model a wider variety of flow scenarios, including extreme flood events (for which no calibration may be available), dam break floods, or ice-related events, necessitates the use of deterministic (hydraulic) models. A more fundamental advantage of hydraulic flood routing models over hydrologic models, in terms of less dynamic events, is that output describing flood hydrographs between gauge sites is produced. Such output is valuable in flow forecasting, and as input to the hydraulic analyses required for floodplain delineation. To date, hydraulic flood routing models have not gained widespread use for two key reasons. First, they present a particularly challenging numerical problem. Second, they are seen to be data intensive, requiring geometric data over the entire modelled reach. The former problem is no longer the primary concern, as recent research has led to the development of numerous robust computational schemes. The intensive data requirements of hydraulic models are much more limiting from a practical perspective, as flood routing typically involves very long reaches and the cost of obtaining sufficient cross section data is generally prohibitive. In this investigation, the reliability of a hydraulic flood routing model based on limited cross section survey data is evaluated for the case of the Peace River in British Columbia and Alberta. Based on the successful results of these investigations, it is concluded that a reliable hydraulic flood routing model can be developed with limited field data supplemented with topographic map data. Key words: flood routing, St. Venant equations, Peace River, characteristic-dissipative Galerkin scheme, finite element method.