Improving the uncertainty of ocean surface wind and wave parameters estimated by a single HF radar system

A 16 Rx element linear array HF radar system (LERA III), working at 27.75 MHz and 300kHz in bandwidth, was installed at the north of Taichung harbor at the western coast of Taiwan in November 2018. This LERA system is low-cost, compact in size, easy to set-up, and maintain. The purpose of present studies is to implement algorithms for retrieving wave and wind fields and assess the system performance in terms of operational mode. For inter-comparison, the long-term in-situ wave data measured by an AWAC was adopted. Wind data were measured from a coastal wind gauge. The inter-comparisons between radar data and in-situ data were carried out on seasonal basis, including severe sea states during winter monsoon and passage of typhoon as well as calm seas during spring.

For the data processing, the Doppler-range spectrum for each azimuth direction was extracted by using the classical beam-forming technique and then provided as level 1 product for further analysis. Regarding the method for retrieving wave parameters, formulations directly derived from Barrick’s assumption was implemented. In those formulas, wave parameters are calculated based on the ratio of the 2^nd order component multiplied by the coupling coefficient function to the 1^st order component in the Doppler spectrum. It means that no empirical constants were included. Initially, Wyatt’s (1999, 2011) and Walsh & Howell’s (1993) methods were applied to determine the lower and higher bounds that separate the 1^st and 2^nd order component. For wind speed inversion, Dexter & Theodorides’s (1982) method was adopted. The Bragg wave direction was used as a proxy to the direction of the wind field.

It is found that when using Wyatt’s (1999, 2011) method, the wave height and period results often lead to bias estimations for severe sea-state, and with the presence of highly variable surface current. In order to improve the accuracy, adaptive methods for the identification of spectra component areas is crucial. In this study, an alternative method is proposed. This method is developed based on the concept proposed by Kirincich (2017), which includes the pretreatment of Doppler-rang spectrums, marker-controlled watershed segmentation, and an image processing technique. In this research, we will demonstrate the advantages of using the new method for wave and wind field retrieval. From comparative studies, the error indexes based on the sea truth data are discussed. It is found that the accuracy would be improved using the proposed method, especially for the cases of varying current fields, severe sea state, and noisy radio background.

Key words: high-frequency surface wave radar; phased array antennas; significant wave height, wave period, marker-controlled watershed segmentation (MCWS) techniques.