Doppler parameter estimation from SAR (Synthetic Aperture Radar) for velocity measurements: Sea surface current and ship velocity

Abstract

Doppler parameter from SAR (Synthetic Aperture Radar) is among the most effective tools for velocity measurements. The physical principle for estimating radial velocity utilizes the Doppler shift extracted from SAR data. The Doppler shift is caused by the relative motion between a sensor and target. The Doppler parameter can be used as an important source for velocity measurements in various applications including oceanography, geology, civilian, and military fields. Based on the type of SAR products and Doppler parameter estimation techniques, specific applications including distributed and artificial moving target velocities can be estimated with the velocity range. In this thesis, I examined a distributed target velocity, a sea surface current, using SAR raw data and an artificial target velocity, a ship, using SAR SLC data of a single channel SAR system. <br/> First, the Doppler parameter in the received radar signal from the SAR raw data was used to retrieve the sea surface velocity. Sea surface velocity is derived by calculating Doppler shift anomalies between predicted and estimated Doppler centroids. The predicted Doppler centroid is defined as the Doppler centroid, in which it is assumed that the target does not move, and it is calculated based on improved geometric parameters of a satellite including the satellites orbit, beam pointing direction, and attitude with respect to the rotations of Earth. I used an established model with improved parameters (including slant range distance, look angle, and hour angle) and an iterative fitting procedure. The fitting procedure included a global fitting method and an attitude control algorithm for correct biases. The estimated Doppler shift that represents the actual Doppler centroid in the situation of real SAR data acquisition can be directly extracted from ScanSAR raw signal data calculated by applying the adjusted Average Cross Correlation Coefficient (ACCC). The characteristics of sea surface velocities under hurricane conditions were investigated using RADARSAT-1 ScanSAR Doppler centroid measurements. Five different hurricanes (i.e., Typhoon Xangsane, Hurricane Dean, Hurricane Ivan, Hurricane Lili, and Hurricane Kyle) and sequential acquisitions of two cases (Hurricane Lili and Hurricane Kyle) were selected to investigate the contribution of wind-induced waves to Doppler velocities and compared to in-situ measurements of drifting buoys. The results indicate that hurricane-generated seas and associated winds and waves appear to differ from those of the ordinary sea state. This leads to lower estimates of Doppler velocities than the expected estimates and that are significantly closer to the sea surface current velocities. In this respect, our study is the first attempt to characterize Doppler velocities influenced by tropical cyclones using different and sequential RADARSAT-1 ScanSAR data.<br/> Second, we investigated the potential for automatic calculation of ship velocity using the azimuth offset between ship and wake in SAR SLC imagery. The azimuth offset between ship and wake is proportional to the Doppler shift effect of the back-scattered signal, and is thus related to the radial velocity of the moving target. Our methods automatically identified ships and wakes from TanDEM-X SLC images using convolutional neural networks (CNN), a deep learning technique. An accurate reference point between the ships and wakes was identified using Radon transforms and edge filtering. Additionally, ship velocity was estimated using the along-track interferometry (ATI) phase due to the Doppler shift effect. Using the Korea Strait as a test site estimating moving ship velocity using TanDEM-X data, we compared the accuracy of the ship and wake detection rate with Automatic Identification System (AIS) data. We also compared the processing results from the azimuth offset to those from the ATI and in-situ measurements of AIS to determine the feasibility of estimating moving target velocity. <br/> Thus, the estimation of a Doppler parameter, such as Doppler centroid measurements, azimuth offset, and ATI, can lead to the effective extraction of velocity in various applications including sea surface current and ship velocity.