Author(s): Nguyen Thu Hien
Linked Author(s):
Keywords: Flood computation; Compound channels; Roughness coefficient; Main channel; Flood plains; Calibration
Abstract: In flood computation, many rivers have compound cross-sections and the roughness values in main channel and flood plains are considerably different. In this study, the inverse problem of identifying the roughness coefficient (Manning’s n) for compound channels is investigated using sythetic data. The values of roughness coefficients in the main channel and flood plains are identified as two different parameters using an optimization method (auto-calibration). The writer adopted the well known Preissmann’s four-point different scheme to solve the Saint-Venant equations. The optimisation process involves minimising the square errors in observed values and simulated ones using the Powell algorithm. The results show that when there are no errors (free-noise) in the observed data the identified roughness coefficients are very acurate. However, when observed data with errors (it is the case in paractice), the identified parameters are biased especially for the floodplain roughness coefficients. The errors of indentified roughness coefficients were analysed with some different senarios of peak flows, ratios of the widths and the depths of the main channels and floodplains with differrent error (noise) levels in observed data. The results indicate that the values of main channel roughness coeffients are much more sensitive and less biased from the true values than the floodplain ones. This implicates that the main channel roughness coefficients should be paid more attention in calibration process in the compound channels.
Year: 2014