The relationship between a SAR image and a directional wave spectrum was clarified by many studies, and the various inverse analysis methods for estimating a directional wave spectrum from a SAR image were studied. However, it was difficult to estimate a directional wave spectrum from a SAR image because of the so-called inverse problem. For solving this problem, a new estimation method based on the Bayesian theorem was proposed in this study. In this method, the inverse problem for estimating a directional wave spectrum from a two-dimensional wavenumber spectrum of a SAR image was reduced to a problem of a search for the minimum value. For searching a minimum value properly, a priori condition in the Bayesian theorem was considered. From this viewpoint, the smooth change of estimated directional wave spectrum was regarded as a priori condition in this study. In addition, for deciding the minimum value objectively,
ABIC was used. To verify the accuracy of the estimation method proposed in this study, a two-dimensional wavenumber spectrum of a SAR image was calculated from the theoretical directional wave spectrum, then, a directional wave spectrum was estimated from a two-dimensional wavenumber spectrum by this estimation method. As the result of comparison between the theoretical directional wave spectrum and the estimated directional wave spectrum, they showed good agreement. Finally, a directional wave spectrum was inversely estimated from a EERS1-SAR image. Simultaneous observations were made with wave-gage array, and the subsequent comparison showed good agreement with the result obtained from the EERS1-SAR image.
抄録全体を表示