Numerical Investigation Of The Uncertainty Of Typhoon Wave Parameters Retrieval Using HF Radar

The aim of this study was to identify the sources of uncertainties of typhoon wave field monitoring using HF radar and to quantitatively assess the bias of the wave parameters, such as significant wave height and mean period, retrieved under various conditions. The strategy was to apply a purely numerical simulation of the Doppler-range spectra and then compare the estimation results to the target. For the quantitative investigation, a numerical test-bed was established. The test-bed was used to simulate the Doppler spectra based on Barrick's (1972) theory by using the input of the directional wave spectra. The typhoon wave spectra were hindcasted using a third-generation wave model. Initially, the uncertainty source from Barrick's theory was identified when comparing the deviations of the estimation results obtained from the idealized case of steady and homogenous fields. The accuracy of the wave height and the mean period was found to be significantly influenced by the angle between the radar-looking direction and the wave direction. Furthermore, two conceptual numerical experiments were designed to evaluate the uncertainties owing to the rotation and the translation of the typhoon wind fields, respectively. The first design focussed on the upper-right quadrant around the maximum wind radius of the typhoon using a virtual radar network that moved along with the typhoon. The second one was a more realistic design in that the virtual radar stations were located on the coastline. The scatter index of the wave height estimated from the second design was found to be approximately twice larger than that obtained using the first design. It was 25% larger for the uncertainty of the mean period. This demonstrated that except for the error from theory, the uncertainty of the estimated wave parameters in type 1 was influenced by the change in the wave generated by the wind field, while that of type 2 was affected by the complicated typhoon wavefield, including the mixed wind waves and swells. The results showed that the error of 0.02 of the scatter index in the case of the wave height could be identified even when no system noise was considered. This error was attributed to the simplification of the coupling coefficient and the weighting function in Barrick's theory. The error was direction dependent and non-negligible. For the typhoon cases, the heterogeneity and rapid changes in the spatial distribution of the wavefield under the influence of the rotating wind fields were the challenging factors for the HF radar wave parameter retrieval. The error increased further under such conditions.