| description abstract | An integrated set of laboratory and numerical-model experiments has been conducted to understand the development of residual circulation surrounding a coastal canyon and to explore further the degree to which laboratory experiments can provide useful benchmark datasets for numerical models of the coastal ocean. The use of an idealized shear-stress boundary condition along the coastal floor in the numerical model gives good quantitative agreement with the laboratory results for the zeroth-order, time-dependent flow and good qualitative agreement for the higher-order [i.e., O(Ro), where Ro (the Rossby number) is small] time-mean flow. The quantitative agreement for the latter, however, is not within estimates of laboratory uncertainties. It is shown that the use of a no-slip condition along the floor improves the model away from the canyon boundaries, but the enhanced viscosities needed to obtain numerical stability give boundary layers that are too wide along the coastline. The laboratory and numerical-model results are used to investigate the trends of a number of flow diagnostics with changes in the governing parameters. A scaling argument to estimate the characteristic strength of the horizontal component of the time-mean or residual velocity U1 leads to the relation U1/u0 ? [Ro(hS/ hD)?1Ro?1tBu?1/2Ek?1/2], where u0 is the amplitude of the oscillatory background flow at the shelfbreak level, (hS/hD) is the ratio of the depth of the shelf to that of the deep ocean, Rot is the temporal Rossby number, Bu is the Burger number, and Ek is the Ekman number. Laboratory and numerical data support this scaling. The model-to-model comparisons indicate that, for the range of parameters investigated, upwelling dominates the residual flow patterns in the vicinity of the shelf break and above. This study supports the notion that a closely coupled laboratory?numerical model investigation can lead to results that are more reliable than those obtained by either approach alone. | |
