Improving near real-time GNSS-IR sea level retrievals with robust outlier detection

The Global Navigation Satellite Systems - Interferometric Reflectometry (GNSS-IR) method has been utilized for nearly fifteen years as an alternative and cost-effective approach to determine hydrological parameters such as sea level, snow depth, and soil moisture through the analysis of signal-to-noise ratio (SNR) data. Most GNSS-IR studies to date rely on archived data and post-processed results. However, the potential for near real-time GNSS-IR analysis is increasingly being explored. In this study, high-rate GNSS archive data, sampled at 1-second intervals and stored in 15-minute files, were processed in a simulated near real-time workflow. Every 15 minutes, new data were added to the analysis, focusing exclusively on the most recent 60 minutes of observations. A novel approach for detecting outliers in near real-time GNSS-IR estimates was also proposed. The median-based robust outlier detection (ROD) method, previously validated for post-processed GNSS-IR snow depth results, was adapted and applied to near real-time GNSS-IR data. A 30-day dataset of multi-GNSS, multi-frequency SNR observations from the Portland (PTLD) GNSS station in Australia, collected in November 2024, was analyzed. The near real-time GNSS-IR results were validated using sea level measurements from the PORL tide gauge station. The results demonstrate that the modified ROD approach can be used to identify outliers in near real-time GNSS-IR sea level retrievals.