Recognition And Numerical Simulation Of Oceanic Internal Waves Based On Synthetic Aperture Radar

Due to the characteristics of the Synthetic Aperture Radar(SAR),which can work in different periods and different weather conditions,SAR has become an important tool in observing oceanic internal waves.The recognition of oceanic internal waves are not only beneficial to discovering the oceanic internal waves phenomenon earlier but also conducive to the safety of ocean navigation and the development and utilization of marine resources.The recognitions of oceanic internal waves from SAR images are the basis of the further study for oceanic internal waves parameter inversions.Since the oceanic internal waves are imaged as light and dark stripes in the SAR images,the geometric characteristics of the stripes indirectly represent the parameter information of oceanic internal waves.Hence,with the challenges and opportunities brought by the big data of ocean remote sensing,how to better identify oceanic internal waves from the SAR images and get parameter information from oceanic internal waves is the hotspot of oceanic internal waves currently in the remote sensing research.To achieve this,it is necessary to identify and segment information of oceanic internal waves from the SAR images,and obtain the locations of the light and dark stripes of oceanic internal waves,and then inverse the parameters of oceanic internal waves based on their characteristics.Since SAR image data can not only be obtained from satellites,it can also be obtained by simulation.The data acquired by these two methods has their own advantages and disadvantages,and can be analyzed through the quantitative details of the data assimilation.In order to better analyze the errors existing in the identification and segmentation algorithm,it is necessary to use the SAR simulation images.Therefore,to further evaluate the accuracy of the segmentation and inversion value of oceanic internal waves,it is necessary to carry out numerical simulation and SAR imaging simulation of oceanic internal waves based on theoretical research and numerical simulation,then obtain the segmentation and inversion of oceanic internal waves with SAR images obtained by the simulation,and analyze the accuracy of the method with the given parameters of oceanic internal wave in simulation.Therefore,the following studies have been carried out.Firstly,a method for detecting oceanic internal waves in SAR images based on the boundary characteristics of remote sensing images is established Through the Canny operator detection,the contour in the SAR images is identified,and then the length,direction and area ratio of the oceanic internal waves stripes are utilized.By screening the contour,and finally get the oceanic internal waves stripes.Secondly,an automatic detection and identification method of oceanic wave stripes in SAR images based on support vector machine(SVM)and feature fusion is established.By using histogram of orientated gradients(HOG),Grey level co-occurrence matrix(GLCM),and fractal dimension(FD)to extract the SAR images features,and then use SVM classifies the SAR images and obtain the SAR image containing oceanic internal waves.Combined with the characteristics of the oceanic internal waves stripes,the locations of oceanic internal waves are finally acquired.Thirdly,a method for semantic segmentation of oceanic internal waves stripes in SAR images based on SegNet is established.Based on the SegNet,the semantic segmentation method can not only automatically identify whether the SAR images contains oceanic internal waves,but also obtain the respective locations of the light and dark stripes of the oceanic internal waves in the SAR images,including the relative locations.The information can be used to distinguish the moment when oceanic internal waves undergo polarity conversion.In addition,the width error of the identified stripe is also analyzed.Fourthly,a method for instance segmentation of oceanic internal waves stripes in SAR images based on Mask R-CNN is established.Based on Mask R-CNN,the instant segmentation method can better identify the light and dark stripes of oceanic internal waves,and distinguish them with colours and labels.In addition,the method can accurately identify the relative locations of the light and dark stripes,the width and the direction of each of the light and dark stripes,and calculate the stripe width and angle at anywhere of light and dark stripes.Finally,based on potential flow theory,a theoretical model of oceanic internal waves propagation in a two layer with two different water surface boundary conditions,namely the rigid assumption and the free surface assumption,is derived and the numerical model under the rigid assumption is established.With the rigid lid,the oceanic internal waves are simulated and analyzed,then compare the numerical results with the analytical solution and obtain the surface layer flow for SAR imaging simulation.