[Paper Review] PARIS Altimetry with L1 Frequency Data from the Bridge 2 Experiment

Giulio Ruffini, Marco Caparrini+1 more|ArXiv.org|Dec 16, 2002

Soil Moisture and Remote Sensing5 references3 citations

TL;DR

This paper presents a novel application of GPS-Reflectometry (GNSS-R) for altimetry using L1 frequency signals from the Bridge-2 campaign, employing the PARIS phase altimetry technique via the Parfait algorithm. It achieves sub-centimeter precision (better than 2 cm) in measuring sea surface height changes over a bridge, demonstrating high-accuracy tide estimation without signal fading models or signal rejection based on SNR.

ABSTRACT

A portion of 20 minutes of the GPS signals collected during the Bridge 2 experimental campaign, performed by ESA, have been processed. An innovative algorithm called Parfait, developed by Starlab and implemented within Starlab's GNSS-R Software package STARLIGHT (STARLab Interferometric Gnss Toolkit), has been successfully used with this set of data. A comparison with tide values independently collected and with differential GPS processed data has been performed. We report a successful PARIS phase altimetric measure of the Zeeland Brug over the sea surface with a rapidly changing tide, with a precision better than 2 cm.

Motivation & Objective

• To evaluate the feasibility of PARIS phase altimetry using L1 frequency GPS signals from the Bridge-2 experiment.
• To assess the performance of the Parfait algorithm in retrieving phase information from reflected GNSS signals for altimetric applications.
• To compare the altimetric results with independent tide gauge measurements and differential GPS data to validate accuracy.
• To investigate the impact of instrumental bias and signal fading on phase-based height estimation.
• To determine whether high-precision altimetry can be achieved without relying on signal fading models or SNR-based satellite filtering.

Proposed method

• The Parfait algorithm, implemented in the STARLIGHT GNSS-R software package, was used to extract the in-phase and quadrature components of the reflected GPS signal by generating a local quadrature signal through I/Q mixing.
• The complex signal was reconstructed from in-phase (I) and quadrature (Q) components, enabling phase estimation of the reflected signal relative to the direct signal.
• A least-squares fit was applied to the phase history against the sine of satellite elevation to estimate bridge height and instrumental bias.
• Integer ambiguity resolution was performed over a constrained search space around an initial guess to resolve cycle ambiguities and improve height estimation.
• The algorithm processed 20 minutes of data from two segments (A1 and A2), using coherent integration over 1 ms intervals and assuming constant navigation bits during integration.
• The final height estimate was derived by interpolating the bias-corrected phase data and comparing it to tide gauge measurements, with optimal bias and time delay adjustments applied.

Experimental results

Research questions

• RQ1 Can PARIS phase altimetry achieve sub-centimeter precision using only L1 frequency GPS signals from a ground-based experiment?
• RQ2 How does the Parfait algorithm perform in estimating the phase of reflected GNSS signals without modeling signal fading effects?
• RQ3 To what extent can instrumental bias and signal-to-noise ratio (SNR) variations affect the accuracy of altimetric height estimation?
• RQ4 Is the phase-based height estimation consistent with independent tide gauge measurements over a short time window with rapidly changing tides?
• RQ5 Can the Parfait method provide accurate altimetry without rejecting satellites based on SNR or fading characteristics?

Key findings

• The Parfait algorithm successfully retrieved PARIS phase altimetry with a precision better than 2 cm, achieving a standard deviation of 0.35 cm relative to tide gauge measurements during the first 10 minutes of data (part A1).
• The final bridge height estimation was 18.82 m with an instrumental bias of -0.45 cm, refined from an initial guess of [0 0 1 1 2 3] to [0 0 2 2 4 5] through ambiguity resolution.
• For part A2, the standard deviation of the interpolated height estimate relative to tide was 0.84 cm, with a final height estimate of 17.41 m and bias of -0.08 cm.
• When fitting both data segments to the tide curve, the best agreement was achieved with a 3-minute and 12-second time delay and a bias of 39.13 cm, yielding a standard deviation of 0.893 cm.
• The comparison between estimated height changes in A1 and A2 showed agreement within approximately 4 cm, confirming consistency with measured tide variations.
• The method demonstrated robustness by achieving high-precision altimetry without relying on signal fading models or SNR-based satellite filtering, even with low SNR and frequent fading conditions.