Comparative analysis of GNSS-IR sea-level retrieval methods

A global navigation satellite system interferometric reflectometry (GNSS-IR) technique has been confirmed to retrieve sea levels using signal-to-noise ratio (SNR) data. To investigate the suitability of different GNSS-IR sea-level retrieval methods, several strategies were tested during the data process, including: the whole-arc Lomb-Scargle periodogram (whole-arc LSP) method and the window LSP (WinLSP) method, when using spectral analysis on SNR data; the tidal harmonic analysis (THA) and dynamic SNR method, when using dynamic correction on retrievals; and the moving-window smoothing method, proposed for application upon processing of retrievals. Furthermore, the THA method is improved by segmented SNR data, and the moving-window smoothing method with robust local weighted regression (RLOWESS) and a Savitzky–Golay (SG) filter was adopted for better serviceability. One-month data from the SC02 station were used to test all the strategies by comparing with local tide-gauge records. HKQT station further verified the usability of the moving-window smoothing method. The results confirmed that the WinLSP method can obtain more retrievals, implying higher temporal resolution, but the whole-arc LSP method achieves better precision. The former is easily susceptible to the sampling rate of the SNR. The correction results of the dynamic SNR method are better than those of the segmented THA method. When the former is applied to retrievals from the WinLSP method, the retrievals of all windows from multiple SNR arcs should be adjusted simultaneously. The moving-window smoothing method based on RLOWESS and the SG filter is more applicable to the retrievals corrected by the segmented THA method, and the smoothing effect of the two smoothing methods makes little difference. In addition, wind speed more than 20 m s −1 seriously affects the effect of GNSS-IR sea-level retrieval. The GNSS-IR technique has the potential for centimeter-level sea level retrieval.