Impulsively Started Flow in a Submarine Canyon: Comparison of Results from Laboratory and Numerical Models

Abstract

Intercomparisons have been made of results from laboratory experiments and a numerical model for the flow in the vicinity of an idealized submarine canyon located along an otherwise continuous shelf. Motion in the rotating and continuously stratified fluid was impulsively generated by suddenly changing the period of rotation, so that the resulting flow occurred with the coastline either on the left (upwelling favorable) or right (downwelling favorable) when facing downstream. A principal purpose of the study was to further develop the notion that laboratory experiments can be effectively utilized to provide datasets to benchmark the development of numerical models. Laboratory data are of two types: velocity fields on three horizontal planes at numerous times, and time series of isopycnal movement in the canyon area. Comparison of numerical and laboratory results shows that values for bottom friction and interior mixing in the numerical model are crucial. Once those friction/mixing parameters are set, ?skill? statistics using observed and predicted horizontal velocity components indicate that the high quality of the numerical model description is maintained over the full measurement period. Two principal features of the circulation are early (<one rotation period) along-canyon flow followed by generation of a canyon vortex. In up- (down-) welling cases, the cyclone (anticyclone) develops along the upstream edge of the canyon and then advects into the canyon interior without significant local vortex stretching within the canyon itself. Numerical results for the case of an extra slow rotation rate change show that vortex creation is not an artifact of the fast rate of rotation change. The canyon vortices extend from just slightly above shelf depth to the deepest part of the canyon; the intensities of the up- and downwelling vortices are asymmetric with respect to the direction of forcing at shelf level, but basically symmetric deeper in the canyon. Upper column vorticity generation by stretching over the canyon rim and flow separation around the canyon headlands could explain this upper water column asymmetric response. The symmetric response in the lower water column is shown to be related to the flow separation only. Overall, the results demonstrate that laboratory and numerical experiments work hand in hand to decipher the complexities of circulation and hydrography undergoing rapid change in a model coastal canyon.