Science Service System

Summary of Proposal OCE0059

TitleShip classification and directional wave spectrum analysis on wave fields in Korean coastal waters
Investigator Yang, Chan Su - Korea ocean Research and Development Insitute, Korea Ocean Satellite Center
Team Members
Research Scientist Shanmugam, Palanisamy - Korea Ocean Research and Development Institute, Ocean Satellite Research Group
SummaryThis study will develop methodology to detect marine transports (e.g. ships) and characterze the ocean wavefields. The main objectives are as follows, 1. Characterizing type and heterogeneity of the backscattering mechanisms 2. Identification of ships in a range of sea states through multi-look processing method. 3. Development of algorithms/methods for characterizing ocean wave filed through multi-look processing. 4.Validation of the developed algorithms using field data
Detailed reportIn order to identify ships in the vast sea a novel design is proposed which is done by means of integration of Synthetic Aperture Radar (SAR) image and Automatic Identification System (AIS) data. That was performed through 4 main steps: matching of time, position, size and speed. Ships, which do not provide the AIS information, can also be detected by this integrated process. The results of that study will contribute to design a Near-Real-Time (NRT) operational system for ship detection, identification, and classification by SARs through graphical screen visualized results; thereby would be useful for coast guards as an early warning system. As for directional wave spectrum analysis on wave fields, spherical wave parameters were estimated which appears in SAR image of coast region. The ocean wave parameters were retrieved by determining the dominant wavelengths, and then estimating wave slopes and heights assuming that the spherical waves were linear and progressive. These azimuth travelling waves are much affected by the velocity bunching mechanism and it is difficult to estimate the wave parameters for these affected areas in SAR imagery. To compensate these effects, the velocity bunching ratio (VBR) based on modulation transfer function (MTF) was compared with the intensity ratio for neighbor area in the radial direction in order to assign the spherical wave properties for azimuthally travelling waves. Dispersion relation provides good estimates for the wave heights for all the selected sub-image areas in the range of 1m to 2m. VBR based on MTF was found to be 0.78 at wave height of 1.36m, while the intensity-based VBR was 0.69 which corresponds to the height of 1.75m. The sea-bottom effects were found to be the reason of velocity bunching accounts for azimuthally travelling spherical waves and the subsequent difference. To know a relation between SAR signals and ocean surface phenomena, a numerical simulation technique for computing and analysis of the radar backscattering from the ocean surface covered by oil spill is presented using a microwave scattering model and Monte-Carlo simulation. That numerical computation agreed well with the theoretical models and the SAR measurements. The reduction of the backscattering coefficients due to oil spills on sea surfaces has been analyzed for various surface roughness spectra, oil layer thickness, frequencies and incidence angles. The preliminary results can be applied to detect of oil spills on sea surfaces. Through these studies, the researchers have developed algorithms which can be applied to TerraSAR-X images for effective monitoring and surveillance of ocean. In particular, the TerraSAR-X data was expended in conjugation with SAR images, numerical model and ground-based data. Therefore, it can be expected that the developed algorithms and methods could be suitably utilized in the marine environmental safety, management, as well as would be potentially helpful in further research. This report is organized based on published papers which are described here as following studies: Study I: Integration of SAR and AIS for Ship recognition Ship detection by SAR in coastal waters and ports has several problems to overcome. Therefore through this study a new approach is presented to integrate the data assessed through ship-borne AIS transponders and space-borne SAR to design and operate a real-time system for ship detection in coastal waters as well as in the open sea in order to discriminate and identify ships. As an example, a large number of AIS signals are transmitted from ships, and signals from different ships sometimes overlap, resulting in the loss of some signals. This causes difficulty in identifying the AIS signals corresponding to particular ships in the SAR image. Radar and AIS have a short transmission range for the exchange the vessel’s identity information. However, satellite based SAR provides a large area of coverage for a particular region. The ships’ information such as size, position and speed can be estimated for the targets without identity information. Key objectives of the study were: - To detect the ships’ position in TerraSAR-X HH-polarization image datasets (SpotLight mode) over Incheon Port in South Korea and Tokyo Bay in Japan using thresholding, morphological and concatenating operations on different dates of acquisition. - AIS data sorting and estimation of Dead Reckoning (DR) positions. - To perform three consequent criteria of ‘Position Matching’, ‘Size Matching’, and ‘Speed Matching’ between SAR and AIS based ships in order to achieve integrated output results with an estimate of their respective error reports along with the compensation of the azimuth position shift. - Generate graphical screen results. AIS data has the potential to contribute to the design of detection algorithms for ship identification in SAR images over wide areas when combined with various automated ship detection methods which have already been developed. In this study a TerraSAR-X image HH-polarization (SpotLight mode) dataset was used. Ships’ lengths were extracted from SAR imagery, and AIS data was used for validation. Based on the previous AIS signal reports, DR positions were estimated for the further matching steps; ships were detected, identified and well-matched. However, ship not validated through AIS was also detected in the SAR image and assessed with a certain level of confidence in terms of position and size. The research resulted in following outcomes: - The ‘Time Matching’ criterion was used to determine DR positions. - The ‘Position Matching’ criterion was used and measurement accuracy was found to be suitable based on the nearest neighbor re-sampling method. - The ‘Size Matching’ and ‘Speed Matching’ criteria gave a measurement error of less than 20%. SAR-derived ships’ hull signatures and matching with AIS-based ship sizes is the most important finding here. Ships was detected, identified and matched accurately with unidentified ships acquired over the whole image area. Therefore, this work will contribute to the design of an operational system using a real-time satellite-based monitoring system like SAR, and ground based monitoring system like AIS and radar for ship monitoring in coastal regions with very dense traffic levels. Study II: Retrieval of Spherical Ocean Wave Parameters Using RADARSAT-2 SAR Sensor Observed at Chukk, Micronesia The purpose of this study is to estimate the spherical wave parameters in SAR image acquired over the coast of Chukk. Through this study outcome, a simple technique is proposed to extract the spherical wave parameters along their propagation from a start point up to the certain distance in spherical and radial manner with the consideration of velocity bunching mechanism for azimuth traveling waves. Here a simple technique is proposed to estimate the spherical wave heights for the 16 different sub-image areas (256×256 pixels) extracted from RADARSAT-2 SAR sensor data over the coast of Chukk Island. Fast Fourier Transform (FFT) analysis is carried out over all the sub-image areas in order to determine the wavelengths and based on these measurements, wave slopes has been estimated for the all the sub sequent areas using dispersion relation for different water depth conditions obtained through navigational chart. Estimation of wave heights is performed using wave lengths and slopes. The retrieval of ocean wave parameters consists of two main stages: 1. To determine the dominant wavelengths by FFT over 16 sub-image areas, and 2. To estimate wave slopes and heights using dispersion relationship under various water wave conditions. It is assumed that the spherical waves are linear and progressive. These types of waves have the range and azimuth components traveling in radial directions. The azimuth travelling waves are more affected by the velocity bunching mechanism and it is difficult to estimate the wave parameters for these affected areas in SAR imagery. In order to compensate these effects, the velocity bunching ratio (VBR) based on modulation transfer function (MTF) was compared with the intensity ratio for neighbor area in the radial direction in order to assign the spherical wave properties for azimuthally travelling waves. Dispersion relation provides the good estimates for the wave heights for all the selected sub-image areas in the range of 1m to 2m. VBR based on MTF was found to be 0.78 at wave height of 1.36m, while the intensity-based VBR was 0.69 which corresponds to the height of 1.75m. It can be said that the velocity bunching accounts for azimuthally travelling spherical waves and the difference results from the sea-bottom effects. Velocity bunching relationship for the sub-image area P has also been compared with the sub-image area K based on the consideration of wave properties of K. The obtained velocity bunching ratio was found to be 0.78 at the wave height of 1.36m for the subimage P. The ratio of image intensity to mean image intensity is also being estimated for sub-image area P and it was found to be 0.69. This value is then further compared with the obtained VBR through the consideration of the VBR obtained through modulating transfer function for area P. The comparison result showed the closeness in the value of VBR for sub-image area P with both the methods. Further, it can be concluded that the wave parameters for sub-image area P is same as that of area K. Hence, the estimated wave height for sub-image area P can be assigned as 1.36m. In future, we will estimate the wave heights using polarization ratio of radar cross sections for HH- and VV-polarization data in order to estimate sea surface slope by pixel based analysis in order to test and validate the results with in-situ measurements. Also, plan is to be decided to estimate the wave parameters using the theory of radar cross section and velocity bunching effect for the spherically generated waves. Study III: An analysis of the radar backscatter from oil-covered sea surfaces using moment method and Monte-Carlo simulation: preliminary results In the analysis, a one-dimensional rough sea surface is numerically generated with an ocean waveheight spectrum for a given wind velocity. A two-layered medium is then generated by adding a thin oil layer on the simulated rough sea surface. The electric fields backscattered from the sea surface with two-layered medium are computed with the method of moments, and the backscattering coefficients are statistically obtained with N independent samples for each oil-spilled surface using the Monte-Carlo technique for various conditions of surface roughness, oil-layer thickness, frequency, polarization and incidence angle. The numerical simulation results are compared with theoretical models for clean sea surfaces and SAR images of an oil-spilled sea surface caused by the Hebei Spirit oil tanker in 2007. Further, conditions for better oil spill extraction are sought by the numerical simulation on the effects of wind speed and oil-layer thickness at different incidence angles on the backscattering coefficients. The numerical results were compared with the small perturbation method and a new physical optics models for clean sea surfaces, and the real data of the 2007 Hebei Spirit oil tanker were acquired by the ENVISAT ASAR and the TerraSAR-X. The numerical computation agreed well with the theoretical models and the SAR measurements. The reduction of the backscattering coefficients due to oil-spills on sea surfaces has been analyzed using the new numerical technique for various surface roughness spectra, oil layer thickness, frequencies and incidence angles. It was shown quantitatively that the reduction of the radar backscatter increases as the thickness of oil slick increases, and decreases as the wind speed increases. The results will allow better detections of oil spills on sea surfaces using SAR images.

Back to list of proposals

© DLR 2004-2016