Abstract
This note studies the response of a simple linear baroclinic coastal-upwelling model to fluctuating longshore winds. Correlations between wind stress and velocities are computed explicitly. It is shown that these correlations depend primarily upon the wind-stress spectrum and that, for a realistic spectrum, the wind stress leads the longshore velocity by approximately one day. The computed correlations agree remarkably well with observations and dismiss the belief that time lags ought to be directly related to the local inertial period, i.e., some fraction of the local pendulum day.
