On the Impact of Sea State on GNSS-R Polarimetric Observations

We investigate a long-term ground-based GNSS-R dataset to evaluate the effect of sea state on the polarization of the reflected signals. The dataset consists of one-year polarimetric observations recorded at Onsala space observatory in Sweden in 2016 using right- and left-handed circular polarization (RHCP and LHCP) antennas. One up-looking antenna to receive direct signal and two side-looking antennas to collect reflections are installed at about 3 meters above sea level. The data is collocated with the measurements from a nearby tide-gauge and meteorological station.

We focus on precise power estimation using a polarimetric processor based on Lomb–Scargle periodogram at precisely observed sea levels. The processor converts 0.1 Hz coherent in-phase and quadrature correlation sums provided by a reflectometry receiver to power estimates of the direct and reflected signals. The power estimates are reduced to three power ratios, i.e. cross-, co-, and cross to co-polarization. A model, describing the elevation dependent power loss due to sea surface roughness, is then utilized to invert the calculated power ratios to the standard deviation of sea surface height.

Analysis of about 14000 events found in the dataset (~40 continuous tracks per day) shows a fair agreement with the wind speeds as an indicator of the sea state. Although an increasing sensitivity to sea state is observed for all the power ratios at elevation angles above 10 degrees, the measurements from the co-polar link seem to be less affected by the surface roughness. The results reveal that the existing model cannot predict the effect of sea surface roughness in a comprehensive way. The different response of RHCP and LHCP observations to roughness is evident, however, the polarization dependence is not covered by the model. The deviations from the model are particularly clear at lowest elevations (<5 deg) where the roughness effect is expected to vanish. The results indicate that roughness also affect observations at lowest elevation angles. In this elevation range the expected dominance of the RHCP component above the LHCP component is not observed.  A different approach is required to model the influence of sea state in GNSS-R. The increasing amount of reflectometry data may allow to retrieve an empirical relation between coherent reflection power and sea state in future investigation.