Year of Publication: 1996
The objective of this study was to develop a preliminary hydraulic flood routing model of the Peace River, between the Bennett Dam, in British Columbia, and Peace Point in Wood Buffalo National Park, Alberta. Although Alberta Environmental Protection hydrologists have successfully developed a hydrologic flood routing model of this reach using the “SSARR” model, output from models of this type is limited to discharge hydrographs at select sites. As several of the other components of the Northern River Basins Study (NRBS) require estimates of flow parameters not available from hydrologic models, such as stage and velocity, as well as discharge hydrographs at intermediate sites, a hydraulic flood routing model was needed. An additional advantage of this hydraulic flood routing model is that it has the ability to provide for an evaluation of the effects of ice on the propagation of flood hydrographs and, being fully dynamic, it can be used to route extreme events such as dam break floods and surges resulting from ice jam releases.
The project began with the development of a geometric data base describing the study reach. Under the terms of reference for this study, only available data (collected by other agencies) were used. This included surveyed cross section data and National Topographic Survey (N.T.S.) mapping. B.C. Hydro cross sections surveys extended from the dam to the B.C7Alberta Border. Localized cross section surveys were available downstream of the border at the Dunvegan Bridge, the town of Peace River, Fort Vermilion and at Peace Point The largest break in available cross section surveys extended from the town of Peace River to Fort Vermilion, a distance of more than 400 km. Because of this paucity of data, an approximate model of the channel geometry had to be developed from other data sources, in particular: water surface slopes and channel top widths obtained from 1:250,000 scale NTS maps. Given the approximate nature of the geometric model and the fact that the hydraulic model was based on a one-dimensional approximation, a rectangular channel section was assumed. Comparisons to actual surveys confirmed that the surveyed river cross sections were well represented by this classical wide, rectangular channel approximation. The final geometric model consists of more than 1100 computational nodes describing channel width, effective bed elevation and channel roughness. No consideration of flood plain geometry could be provided at this stage, due to the limited field data available.
The hydraulic flood routing model used was the cdg-lD finite element model developed at the University of Alberta by F. Hicks and P. Steffler. The model provides for a solution of the fully dynamic, one-dimensional open channel flow equations (modified St. Venant equations). Although the model is capable of handling highly dynamic flood events (such as dam break floods or surges resulting from ice jam releases), the test scenarios examined for this preliminary study were simpler "diffusive" waves. As outlined in the terms of reference for this study, because of the exploratory nature of the research, the specific range of test scenarios was modified in consultation with Mr. John Taggart Alberta Environmental Protection (AEP) and Dr. Terry Prowse, Hydraulics/Hydrology/Sediment Project Leader, NRBS. After an examination of available hydrologic data the final range of tests was reduced to two events, one moderate and one large, specifically the 1980 spring runoff event and the 1987 summer flood event Evaluations of non-regulated flows and the effects of ice on flood propagation were not considered warranted, given the lack of recorded data and the extensive gaps in surveyed geometry. However, it was intended that the model should provide enough information to assess where further surveys are needed, to facilitate future tests of this type.
The only calibration parameter involved in the development of the hydraulic model was the channel resistance coefficient, specifically Mannings n. Initial values for the parameter were based on the data provided by Kellerhals, Neill and Bray (1972) for 1:2 year flood events at gauge sites. Agreement between measured and computed flood hydrographs was good for both the moderate (1980) and extreme (1987) flood events and no further refinement, or "calibration" of the model was considered warranted until additional data is obtained in the unsurveyed reaches. It is stressed that because of the limited data available (both in terms of these unsurveyed reaches and the lack of overbank geometry) the model is still somewhat empirical.