Statistical Analysis of Reflected GNSS Signal Off Sea Surfaces From a Coastal Scenario

Abstract

This article presents the statistical analysis of the reflected global navigation satellite system-reflectometry (GNSS-R) signal from a coastal experiment, including the non-Gaussianity, probability distribution functions, autocorrelations, and fractal dimensions of the speckle and texture components. The results clearly show that the amplitude distribution is modeled well by a Weibull model. The texture component of the reflected GNSS signal has a log-normal distribution. Due to the presence of the coherent and non-coherent components, the phase of the reflected signal is not uniformly distributed with $\left [{{-\pi, \pi }}\right]$ . The autocorrelation functions (ACFs) of the speckle and texture components both are Gaussian-shaped, with the correlation times on the order of hundreds of milliseconds and a few seconds, respectively. Some statistical properties of the reflected GNSS signal depend on GNSS-R geometry and sea state; therefore, once the influence of GNSS-R geometry is corrected, they can be used to determine sea state. The speckle and texture correlation times of the reflected GNSS signal, as an example, are used to retrieve wind speed. The speckle and texture correlation times provide retrieved wind speeds with root mean square errors (RMSEs) of 1.66 and 1.75 m/s. When a minimum variance estimator is used to fuse two retrieved wind speeds, the RMSE is reduced to 1.46 m/s. The work is helpful for developing a GNSS signal scattering model over the sea surface and further studies on coastal GNSS-R to monitor sea state and maritime target.