Water Level Retrieval Using a posteriori Residual of GNSS Pseudorange and Carrier-Phase Observations

In global navigation satellite system reflectometry (GNSS-R) applications, water level variations could be determined by analyzing one of the three types of data: signal-to-noise ratios (SNRs), combinations of carrier-phase observations, and combinations of pseudorange and carrier-phase observations. In this study, a new GNSS-R method based on the analysis of a posteriori residual of undifferenced and uncombined (UC) precise point positioning (PPP) and ionospheric-free (IF) PPP was proposed. Double peaks were found in the Lomb-Scargle periodogram (LSP) estimates of dual-frequency carrier-phase residuals generated with UC PPP, the pseudorange and carrier-phase residuals generated with IF PPP, and the detrended S2 time series. To solve this problem, we present a strict quality control strategy to screen the correct retrievals. Two experimental datasets from a dam and a bridge-deformation monitoring system, respectively, were used to evaluate the performance of this new method in water level retrieval. The results show that the water level retrievals from the pseudorange and carrier-phase residuals generated by the UC PPP and IF PPP positioning method show a good agreement with the in situ references, with correlation coefficient exceeding 0.95 and 0.97 in the two experiments, which is comparable to the SNR method. The retrievals showed a root-mean-square error (RMSE) of 6-7 cm in the reservoir experiment and 13-16 cm in the Ganjiang river experiment. Meanwhile, the interfrequency bias for about 0.2 m was found in the Xilongchi reservoir water level retrieval experiment; however, it is not the case for the Ganjiang river water level retrieval experiment. The new method enables the simultaneous determination of displacements and water level variation and is, thus, applicable for in situ displacement monitoring of civil engineering structures, such as dam and bridge.