THESIS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Retrieval of ocean surface winds and currents using satellite synthetic aperture radar and infrared radiometry

This thesis focuses on the development, analysis and evaluation of methods for the retrieval of ocean surface winds with synthetic aperture radar (SAR) and surface currents with infrared (IR) radiometer sensors. The SAR wind speed retrievals are based on geophysical model functions using wind directions obtained with two approaches. The first approach determines wind directions from the spatial detection of image features. This method detects small-scale variations in the wind directions. This work includes a validation against in-situ data and an assessment of the retrievals at different resolutions with respect to wind data from a scatterometer sensor and a weather prediction model (WPM). The assessment results show that the detection of small-scale wind features is improved by using the wind direction retrievals from the SAR data. In polar regions, however, the approach is limited by the stability of the marine atmospheric boundary layer, which results in a lack of features for the wind direction retrieval. Thus, the second approach uses wind directions from a WPM to produce wind maps of the Amundsen Sea in Antarctica. Time series of wind maps from both SAR and WPM data were correlated with components of the deep water velocities measured by an acoustic Doppler current profiler at 72.5S and 116.3W during 2010 and 2011. This evaluation showed that the SAR wind data have correlations up to RSAR = 0.71, with larger statistical significance than wind data from a WPM (RWPM = 0.41). Thus, SAR surface winds can be highly correlated with variations in the warm deep water currents, which in previous studies has been shown to be related to the basal melting of the ice shelves in the Amundsen Sea. A method for surface current retrieval with IR sensors was developed from the maximum cross correlation (MCC) technique, which correlates two IR images of the same region acquired at different times and assumes that the current field is produced by the horizontal advection of surface temperatures measured by the IR sensors. The method was evaluated in cloud free days and was qualitatively compared with surface current data from a WPM. The results show that the MCC retrievals are largely in agreement with the modeled currents for 3 h of time delay between the IR images. However, for longer time delays (6 to 10 h), the agreement of the current fields depreciates. In summary, this work resulted in the development and quality assessment of algorithms