Evaluation of Alternative Direction-of-Arrival Methods for Oceanographic HF Radars

The majority of ocean current measuring HF radars obtain the direction of arrival (DOA) of signals backscattered from the ocean with the multiple signal classification (MUSIC) algorithm. These radars often operate under conditions including low signal-to-noise ratio (SNR), low numbers of data samples (aka snapshots), and with the number of signal sources approaching or exceeding the number of receive antenna elements. Improving the accuracy and coverage of the radar data in these situations would improve data produced by radar networks such as the U.S. Integrated Ocean Observing System, revealing new understanding of coastal ocean dynamics. This paper presents an evaluation of DOA techniques developed over the last 10-20 years, for application to oceanographic HF radars using a simulation-based approach. Simulations performed using three commonly used receive arrays suggest that the use of maximum-likelihood estimation by alternating projection (MLE-AP) leads to similar accuracy, with improvement in coverage due to the higher number of DOA solutions obtained when compared to MUSIC. These advances come at a higher computational cost, though the difference is manageable. The analysis also illustrates the need to identify and evaluate signal detection methods (i.e., methods to identify the number of simultaneous source bearings) to work in conjunction with MLE-AP. Overall, these results suggest improvements in the data coverage of ocean current maps produced by HF radars, and thus in the many practical applications employing them such as spill response, and search and rescue.