| In this thesis, two analysis techniques are demonstrated to quantify atmospheric signatures detected by SAR. First, a generalized algorithm for remotely sensing the current-relative wind direction from ERS-1 and RADARSAT images of the ocean surface is presented. It is demonstrated that one-dimensional spectra of SAR wind images can, if uncontaminated by other phenomena, closely resemble spectra derived from time series of wind speed. Specifically, uncontaminated SAR wind spectra for the convective boundary layer exhibit mesoscale power, a spectral gap, and an inertial subrange obeying the --5/3 power law of Kolmogorov. Moreover, the available in situ data suggest that the SAR wind speed images examined here are at least moderately well calibrated. This multi-parameter agreement suggests that well-tuned SAR wind speed retrieval algorithms are indeed capable of retrieving accurate values on pixel scales greater than 150 m under conditions of minimal oceanographic contamination. A procedure is described to estimate the wind direction by maximizing the ratio of power in the inertial subrange spectra from two orthogonal directions. The direction with the least amount of power is along-wind. Two ERS-1 images are analyzed and the results are presented.; Second, an image containing the signatures of nocturnal drainage flow forced exit jets in the Chesapeake Bay is analyzed in order to gain insight into the dynamics of these types of boundary layer flows. The link between basin terrain characteristics and exit jet length is established with the square root of the basin area and the gap width sufficient to account for 90% of the variance in exit jets. In addition to the observational study, a simple two-dimensional shallow fluid model with surface drag is used to simulate these structures. From these simulations, predictive equations for exit jet length and maximum jet velocity are developed. These predictive equations account for more than 99% of the variance in both exit jet length and maximum velocity. |
