Spectral-Based Wind Vector Retrieval From Low-Altitude Multistatic GNSS-Reflectometry

This article investigates the spectral characteristics of reflected global navigation satellite system (GNSS) signals in low-altitude scenarios. The spectral shift and spectral width are defined to characterize the spectra. The results demonstrate that both the auto-spectrum and cross-spectrum exhibit comparable responses to variations in wind speed, wind direction, elevation angle, azimuth angle, swell, and wave age. The spectral widths are sensitive to wind speed, enabling their utilization in retrieving wind speed. The spectral shift shows symmetric dependencies on wind direction, leading to the potential of wind direction retrieval. GNSS-reflectometry (GNSS-R) geometric angles, swell, inverse wave age (IWA), signal-to-noise ratio (SNR), and receiver velocity influence the spectral properties; thus, their effects must be minimized during the retrieval of wind speed and direction. Data from two coastal scenarios and one shipborne experiment are used to validate the capability of spectral features in retrieving wind speed and direction. Surrounding environmental factors, such as coastline, sea state, and ship velocity, significantly impact wind speed and direction retrievals. When these factors are excluded or accounted for in the retrieval models, a notable improvement in retrieval performance can be achieved. In one of the coastal experiments, the auto-spectral and cross-spectral widths yield root mean square errors (RMSEs) of 1.98 and 2.04 m/s, respectively. The auto-spectral and cross-spectral shifts provide retrieved wind directions with RMSEs of 27.80° and 32.39°, respectively, with a 180° ambiguity.