Assimilation of moored velocity data in a model of coastal wind-driven circulation off Oregon: Multivariate capabilities

Abstract

Horizontal current measurements from an array of moored acoustic Doppler profilers are assimilated sequentially into a model of coastal wind-driven circulation off Oregon during the upwelling season of May-August 2001. Model results are compared against independent moored and ship survey data to document a positive effect of velocity data assimilation (DA) on other oceanic variables of interest such as the sea surface height (SSH), temperature, potential density, surface salinity, and near-bottom turbulence parameters. Significant improvement is achieved for the nearshore SSH even when data are assimilated from only two moorings at an alongshore distance of 50 km from the SSH verification site. At 45°N, in an area of simple shelf bathymetry with relatively small alongshore variations, the model (even without DA) provides a good description of the isopycnal structure on a cross-shore section. At 44.2°N, over complicated bathymetry, velocity DA may improve the slope of isopycnals but at the same time not necessarily the density values themselves. Data assimilation based on a time-invariant representation of the forecast error covariance may inhibit spatial variability on horizontal scales smaller than the assumed forecast error decorrelation scale. An experiment involving assimilation of both velocity and moored salinity measurements demonstrates that moored velocity DA improves transport of buoyant surface water. The level of improvement in the near-bottom turbulent dissipation and bottom stress found with the DA model indicates that it is suitable for future studies of spatial and temporal variability in the bottom boundary layer off Oregon. Copyright 2005 by the American Geophysical Union.