Statistical Analysis of Surface Reflectivity with GNSS Reflected Signals from a Mixed Ice and Water Surface

This paper analyzes the surface reflectivity (SR) over time of GPS L1 signals scattered off a partially frozen lake surface. The SR for a given GPS satellite is compared with the mean red value (MRV) of pixels inside the Fresnel zone taken by collocated cameras, with the MRV serving as a truth reference for water (low MRV) or ice (high MRV). GPS signals are collected during a data campaign at Lake Michigan in Chicago, at a time when the lake surface consisted of a mixture of ice and water. A portable ground-based sensor suite is set up at the lake front to collect both surface-scattered GPS signals and independent validation data (lidar and camera) from the surface. The camera pixel locations and the Fresnel zone of the GPS satellites are mapped into an east-north-up (ENU) coordinate system centered at the reflected antenna. The MRV of the Fresnel zone is computed for each satellite. Satellites whose Fresnel zones sweep across both water and ice over the lake surface are selected. To process the GPS reflected signals of chosen satellites, a software defined receiver is customized. In order to find the SNR, one ms coherent correlation power is summed incoherently for one minute for reflected signals. The distance from the satellites to reflection point (Fresnel zone) and from the reflection point to the reflected antenna are computed in order to estimate the SR. Then the SR is correlated with the mean red value. High correlation is seen between SR and MRV for PRN 26, while for PRN 16 only low-to-moderate correlation is found. The low correlation is attributed to a lack of overlap between the Fresnel zone of PRN 16 and the cameras’ fields of view.