Retrieval Of Surface Current From SAR Imagery Using Doppler Shift Method

Measurements of surface current are important to estimate the global transport of salt and heat which regulates the world’s climate. At a regional scale, current have a profound impact on the coastal and shelf ecosystems. So deepening our understanding of surface current is helpful to monitor the changes of environment, maintain coastal safety as well as support the development of fishing and so on. Satellite remote sensing has proved to be an cost-effective means of observing the ocean surface currents, among which, satellite synthetic aperture radar (SAR) is one of the important sensors in terms of its full-day monitoring and high resolution.The basic principle of Doppler method to retrieve surface current is that the Doppler shift anomaly extracted from SAR echoes is linearly connected to range component of current speed. Recently, researches on mesoscale and submesoscale oceanic processes using Doppler method have been gradually a hot topic. In this thesis, the Lagrange model with linked components (LMLC) for the nonlinear stochastic water waves is employed to simulate the Doppler shift and bandwidth of surface motion. The derivations are based on the two-scale model under the hypothesis that the slopes and velocities of surface waves are mutually independent and follow Gaussian distribution. The effect of tilt and hydrodynamic modulation of large-scale waves on small-scale capillary waves are both taken into consideration. The simulations are consistent with the previous observations and show that the HH-pol Doppler shift is larger than that of VV-pol due to the tilt modulation. And the fact that the relaxation rate in hydrodynamic modulation is not zero makes the asymmetry between upwind and downwind. The comparisons with empirical model C-band Doppler (CDOP) prove the simulation is reasonable.The processing steps and error corrections to estimate Doppler shift from ASAR WSM data are given in detail. The error source includes azimuth error, predicted Doppler shift error and residual error. Experiments demonstrated that the RMS of Doppler shift over land area improved a lot after error corrections. The current maps of Agulhas Current retrieved from WSM imagery show that the current core agree well with that from geostrophic current, but the range velocity can better portray mesoscale and submesoscale features due to its high spatial resolution. Further, range velocity and SST map from GHRSST display similar patterns of surface circulations over Agulhas Current. An interesting phenomenon is that the high current velocity corresponds to large SST gradient and its theoretical explanation needs further discussion. It should be noted that the Natal Pulse, large solitary meander which move downstream in the Agulhas Current, can also be monitored by WSM imagery. The range velocity map is able to provide fine structure of Natal Pulse especially its coastal characteristics and would support the further researches on understanding its evolution scheme.A refined algorithm is proposed based on the ASAR IMS imagery, which replace the Doppler estimator with Sign Doppler Estimator (SDE). The SDE puts equal weight on light and dark targets, so the azimuth error can be reduced largely and neglected. As a result, there will be two terms of error source and improve speed of this algorithm. Measurements of oceanic surface current using range velocity over Estuary of Yangtze River are examined to prove the feasibility of this algorithm. The comparisons with tide table demonstrate the retrieved current velocity is reasonable. It is essential to monitor the Yangtze River because altimetry data cannot be employed as corrupted by land contamination as well as tide.This thesis presents the Doppler method to observe surface current and experiments over Agulhas Current and Yangtze River. The direct comparisons with both geostrophic current and SST patterns show that this method is potential to measure surface current quantitatively. The theoretical derivation is consistent with the empirical model, which could support further research on Doppler shift.