The TOPEX/POSEIDON Project is a joint U.S. and French mission to develop and operate an Earth orbiting satellite capable of making accurate measurements of the mean sea level in a way that allows the study of ocean dynamics. The understanding of ocean dynamics is very important in order to study events such as El Nino. Soon after the launch of the TOPEX satellite, some unusually high, but localized, values of the ocean's radar cross section, sigma-0, were observed by scientists at the NASA Goddard Space Flight Center. These phenomena have been referred to as sigma-0 blooms, and are accompanied by an increase in noise in the significant wave height (SWH) and altitude measurements. Since approximately 5% of all data recorded by the satellite contains sigma-0 blooms, it is important to understand their causes so that corrective measures can be taken by NASA. This thesis investigates two possible origins of the sigma-0 blooms: a surface containing a step discontinuity in sigma-0, and a surface containing slick or calm areas. Models corresponding to the theoretical returns from these two types of surfaces are developed and studied. / Master of Science

RapidScat is a spaceborne wind scatterometer mounted on the International Space Station (ISS). The RapidScat mission lasted from September 2014 to November 2016. RapidScat enables the measurement of diurnal patterns of sigma-0 measurements. This capability is possible because of the non-sun-synchronous orbit of the ISS, in which the local time of day (LTOD) of sigma-0 measurements gradually shifts over time in any given location. The ISS platform is a relatively unstable platform for wind scatterometers. Because of the varying attitude of the ISS, RapidScat experiences a constant variation of its pointing vector. Variations of the pointing vector cause variations in the incidence angle of the measurement on the ground, which has a direct effect on sigma-0 measurements. In order to mitigate sigma-0 variations caused by incidence angle and LTOD, the dependence of on these parameters is modeled in order to enable a normalization procedure for sigma-0 . These models of sigma-0 dependence are determined in part by comparing RapidScat data with other active Ku-band instruments. The normalization procedure is designed to adjust the mean value of sigma-0 to be constant across the full range of significant parameter values to match the mean of sigma-0 at a particular nominal parameter value. The normalization procedure is tested both in simulation and with real sigma-0 measurements. The simulated normalization procedure is effective at modeling and removing sigma-0 dependence on incidence angle and LTOD over a homogeneous region. The variance in simulated images is reduced by the normalization procedure. The normalization procedure also reduces variance in real backscatter images of the Amazon and an arbitrary region in East Africa.

An algorithm has been developed that enables improved the resolution wind estimates from SeaWinds data. This thesis presents the development of three key portions of the high resolution wind retrieval algorithm: Compositing individual σ-0 measurements and Kp, Retrieved wind bias correction, and ambiguity selection for high resolution winds. The high resolution winds produced by this algorithm are expected to become a useful resource for scientists and engineers studying the ocean winds. The high resolution wind retrieval algorithm allows wind to be retrieved much closer to land than is available from the low resolution winds estimated from the same scatterometer by the Jet Propulsion Laboratory. The high resolution winds allow features such as the eye of hurricanes to be seen with much greater detail than was previously possible.

The presence of rain over oceans interferes with the measurement of sea surface wind speed and direction from the Sea Winds scatterometer and as a result wind measurements contain biases in rain regions. In past research at the Central Florida Remote Sensing Lab, it has been observed that rain has multi-fractal behavior. In this report we present an algorithm to detect the presence of rain so that rain regions are flagged. The forward and aft views of the horizontal polarization σ0 are used for the extraction of textural information with the help of multi-fractals. A single negated multi-fractal exponent is computed to discriminate between wind and rain. Pixels with exponent value above a threshold are classified as rain pixels and those that do not meet the threshold are further examined with the help of correlation of the multi-fractal exponent within a predefined neighborhood of individual pixels. It was observed that the rain has less correlation within a neighborhood compared to wind. This property is utilized for reactivation of the pixels that fall below a certain threshold of correlation. An advantage of the algorithm is that it requires no training, that is, once a threshold is set, it does not need any further adjustments. Validation results are presented through comparison with the Tropical Rainfall Measurement Mission Microwave Imager (TMI) 2A12 rain retrieval product for one whole day. The results show that the algorithm is efficient in suppressing non-rain (wind) pixels. Also algorithm deficiencies are discussed, for high wind speed regions. Comparisons with other proposed approaches will also be presented.