Surface currents measured by high-frequency (HF) radars in southern San Diego are addressed from three perspectives: technical issues, physical interpretations, and environmental applications. Objective mapping (also known as optimal interpolation (OI)) is applied to the surface vector current using both observed and idealized covariance matrices. The mapping produces smooth fields and can fill in missing data. The covariance matrices calculated from the raw observations of surface currents show a roughly exponential form instead of the Gaussian shape which is often assumed.
The OI methods have been extended to map vector current directly from the radial velocities as an alternative to un-weighted least-squares fitting (UWLS), which has been the default method for the HF radar community. OI uses the expected covariance function in place of the arbitrary, discontinuous correlation function used in UWLS. Moreover, the OI approach reduces inconsistency along baselines between stations and provides superior uncertainty measures for the estimated current field.
In order to refine the covariance estimates and maps, the surface currents are decomposed according to their driving forces: tides, wind, low frequency pressure gradients, and several continuous frequency bands. The locally wind-driven currents are calculated by regression of the shore station winds on the observed surface currents to estimate the wind impulse response function. The response of surface currents to the wind in a coastal region is anisotropic due to the anisotropic bottom/coastline friction, pressure gradient, and boundary conditions. The frictional momentum balance between the gradient of sea surface elevation and the low frequency band currents is also considered. The spatial correlations of the components of surface currents exhibit a mix of Gaussian and exponential shapes with varying decorrelation length scales.
Finally, a data-driven model of the fate and transport of the plumes from three local discharges in southern San Diego has been developed using surface current observations. The statistical model calculates particle trajectories which are compared with water quality samplings, and the skill of an alarm for low water quality is evaluated using receiver operating characteristic (ROC) analysis.