We present a method for evaluating characteristics of wind vector retrieval errors from polarimetric radiometer measurements. We focus on wind direction errors due to ambiguous solutions.

A computer simulation is used to analyze errors in high-frequency (HF) radar ocean surface current measurements. Two pointing algorithms used for current extraction, a direction finding approach using MUltiple SIgnal Characterization (MUSIC) developed by Schmidt (1986), and conventional beam forming, are compared in terms of the effect of variations in sea state parameters on current measurement error. The radar system parameters used in the simulation were taken from the University of Michigan's multi-frequency coastal radar (MCR), which operates on four frequencies from 4.8 to 21.8 MHz and employs an eight-element linear phased array for its receive antenna. Results show MUSIC direction finding to be applicable to phased array systems and to have a better sensitivity to sharp current features, but larger random error than traditional beam forming methods. Also, for cases where beam forming errors are dominated by beam width or low signal to noise ratio, results show MUSIC to be a viable alternative to beam forming.

A simulation of the University of Michigan's Multifrequency Coastal Radar (MCR) has been applied under various scenarios to compare the performance of beam forming and direction finding techniques for bearing determination. The MCR system is a phased array radar that operates at four distinct frequencies in the high frequency (HF) range in order to measure ocean surface currents at several depths. These simulations indicate a decrease in performance of the system using either algorithm for lower frequencies. The simulations also demonstrate an empirical means of generating improved estimates based on the MUSIC algorithm, and indicate advantages of applying the MUSIC algorithm in regions of large horizontal divergence

Several HF radar systems were used to map ocean surface currents in the third Chesapeake Outflow Plume Experiment (COPE-3) conducted just outside the mouth of Chesapeake Bay, off Virginia Beach, VA during October and November 1997. A multifrequency radar system recently constructed by the University of Michigan, an Ocean Surface Current Radar (OSCR) system operated by the University of Miami, and a SeaSonde system developed by Cedar Ocean Sensors, Ltd. All were deployed at nearly the same locations and operated nearly simultaneously. Each of the three systems consisted of two radars separated by about 20 km. Although all three systems are similar in their use of Doppler processing of first-order resonant backscatter of surface-wave HF energy to estimate ocean surface currents, there are significant differences between the systems. The multifrequency system operates simultaneously on four frequencies between 4.8 and 22 MHz and uses omnidirectional transmitting antennas and an array of 8 wideband loop receiving elements operated in either a beam-formation or a direction-finding mode. The use of multiple radar frequencies allows estimates of ocean current to be made at different depths; the effective depth is about 0.04 to 0.08 ocean wavelength, depending on the shape of the vertical current profile. The OSCR system operates on a single frequency and uses a broad-beam transmit antenna and an array of up to 16 whip receiving antennas operated in a beam-formation mode. The SeaSonde operates on a single frequency using an omnidirectional transmit antenna and a set of compact, colocated receiving loops and whip operated in a direction-finding mode.

This paper addresses the application of HF radar systems to the detection and monitoring of shipping activity. The presence of ships within HF radar swaths often hinders the use of the data for the measurement of ocean surface currents. Data collected south of Chesapeake Bay mouth during the third Chesapeake Outfall Plume Experiment (COPE-3) this past fall from both CODAR SeaSonde and multifrequency HF radar systems illustrate the presence of large tanker ships. Comparisons of data from both sets of radar instruments with ancillary observations illustrate the ability of high-frequency radar to detect and track ships within a near coastal region. An approach is suggested to remove the ship echoes from multifrequency HF radar data. Tanker ship radar cross sections are measured to be in the range of 40 dB to 60 dBsm based on a comparison of ship echo power to ocean echo power in the Bragg region.

Simulation techniques are applied to evaluate the performance of HF radar surface current algorithms. The simulation assumes that the ocean echo is due to first order Bragg scatter from a large number of differential area elements in an annular ring of the ocean surface, and generates, to first order, the voltage time series data which would be sampled from an eight element phased array, at four frequencies from 4.8 to 21.8 MHz. The geometry and frequencies match those of a pair of existing pulsed radar systems currently deployed at Monterey Bay. The simulation addresses the problem of estimation of accuracy and error bounds for the ocean radar system. The authors examine physically plausible surface current profiles with random fluctuations in the amplitude of the scattered signal as a function of position on the ocean surface. Deviations between simulation input and algorithm output as a function of the parameters input to the simulation are examined for both direction finding and beam forming algorithms