Source-linked AI summary
GipsyX/RTGx, A New Tool Set for Space Geodetic Operations and Research
Willy Bertiger, Yoaz Bar-Sever, Angie Dorsey, Bruce Haines, Nate Harvey, Dan Hemberger, Michael Heflin, Wenwen Lu, Mark Miller, Angelyn W. Moore, Dave Murphy, Paul Ries, Larry Romans, Aurore Sibois, Ant Sibthorpe, Bela Szilagyi, Michele Vallisneri, Pascal Willis
TL;DR
Existing space-geodetic software needed modernization and broader capabilities for high-accuracy GNSS and related applications. The paper presents GipsyX/RTGx, a unified redesign combining real-time and post-processing functions. It reports deployment across JPL’s GNSS operations and research community, while noting convergence and post-processing scope boundaries.
Problem
Few software packages provide the capabilities needed for centimeter- and millimeter-level GNSS accuracy and for broader space-geodetic processing.
Method
GipsyX/RTGx modernizes JPL’s software architecture by unifying RTG real-time processing with GIPSY post-processing flexibility.
Results
GipsyX/RTGx underlies all GNSS orbit-determination operations at JPL and powers geodetic analyses and science operations worldwide.
Takeaways & Limitations
The software provides a widely used unified tool set for orbit determination, positioning, geodesy, and remote sensing applications.
Takeaways & Limitations
Real-time solutions can require days to fully converge, while RTG lacked post-processing capabilities required by many NASA science applications.
Abstract
from arXiv · showhide
GipsyX/RTGx is the Jet Propulsion Laboratory's (JPL) next generation software package for positioning, navigation, timing, and Earth science using measurements from three geodetic techniques: Global Navigation Satellite Systems (GNSS), Satellite Laser Ranging (SLR), and Doppler Orbitography and Radiopositioning Integrated by Satellite (DORIS); with Very Long Baseline Interferometry (VLBI) under development. The software facilitates combined estimation of geodetic and geophysical parameters using a Kalman filter approach on real or simulated data in both post-processing and in real-time. The estimated parameters include station coordinates and velocities, satellite orbits and clocks, Earth orientation, ionospheric and tropospheric delays. The software is also capable of full realization of a dynamic terrestrial reference through analysis and combination of time series of ground station coordinates. We present some key aspects of its new architecture, and describe some of its major applications, including Real-time orbit determination and ephemeris predictions in the U.S. Air Force Next Generation GPS Operational Control Segment (OCX), as well as in JPL's Global Differential GPS (GDGPS) System, supporting User Range Error (URE) of $<$ 5 cm RMS; precision post-processing GNSS orbit determination, including JPL's contributions to the International GNSS Service (IGS) with URE in the 2 cm RMS range; Precise point positioning (PPP) with ambiguity resolution, both statically and kinematically, for geodetic applications with 2 mm horizontal, and 6.5 mm vertical repeatability for static positioning; Operational orbit and clock determination for Low Earth Orbiting (LEO) satellites, such as NASA's Gravity Recovery and Climate Experiment (GRACE) mission with GRACE relative clock alignment at the 20 ps level.
1. Introduction
GipsyX/RTGx is JPL’s fourth major redesign of GNSS data-analysis software, created by modernizing and unifying earlier GIPSY-OASIS and RTG capabilities. The paper presents its architecture and performance across orbit determination, positioning, geodesy, and remote sensing.
- GipsyX/RTGx was developed to address the need for software capable of obtaining high accuracy across GNSS applications and of modeling other space-geodetic techniques.
- The software set supports applications spanning orbit determination and positioning operations, geodesy, and remote sensing.
- GipsyX/RTGx is JPL’s 4th major redesign of its GNSS data analysis software.
- The redesign combines RTG’s agile real-time processing with GIPSY’s post-processing flexibility in a new architecture.
- By 2014, GipsyX/RTGx had replaced RTG for most GDGPS real-time GNSS orbit-determination operations.
- By 2017, it had replaced GIPSY for all post-processed GPS orbit-determination operations at JPL, including reprocessing tracking data back to the early 1990s.
2. Software Design, Overview
GipsyX/RTGx redesigns JPL’s geodetic software around C++ computational components, Python3 scripting, and a unified architecture. It preserves common-use precision and accuracy while improving usability, performance, and support for larger and new analysis problems.
- Architecture: GipsyX/RTGx replaces GIPSY’s FORTRAN-based computational structure with C++ and uses Python3 as its sole scripting language.The redesign targets easier development and maintenance while avoiding the prior mixture of scripting languages.
- Performance: Almost a factor of three increase in performance is reported for a moderate size problem.The authors attribute the improvement to optimization and parallel computation, while noting that the test problem is too small to realize large multicore speedups.
- Capabilities: GipsyX/RTGx supports very large adjustments combining a gravity field, GNSS orbits, GRACE orbits, and additional low Earth orbiters.These problems are described as only being handleable by GipsyX/RTGx among the compared software capabilities.
- Usability and continuity: For common GIPSY and GipsyX/RTGx use cases, the authors expect almost identical precision and accuracy with a much improved user interface.The redesign aims to preserve existing analytical performance while adding usability and new use cases.
- Capabilities: The software adds hot starts, GNSS phase ambiguity resolution, inertial measurement modeling, and reference-frame determination with Earth-station time-series analysis.These capabilities cover both real-time and post-processing workflows where stated.
- Software organization: A single main executable, rtgx, replaces many individual GIPSY executable modules, while Python3 utilities support time-series and data-processing tasks.The distribution contains approximately 140 executable programs, including the main tracking-data analysis executable and smaller utilities.
3. User interface - the input tree
The input tree is a hierarchical text-based interface for configuring executables with complex and finely controllable user inputs. Its roots, branches, and leaves organize configuration data in a Python-like indentation structure.
- Configuration: Several executables use the tree interface when they require more complicated and finely controllable user inputs.The interface is intended to provide structured configuration for these executables.
- Tree structure: A tree is a text file with hierarchical structure identified using Python-like indentation.The hierarchy consists of roots, branches, and leaves.
- Tree structure: Roots are the highest level, branches organize intermediate structure, and leaves are lowest-level elements that generally contain specific data.Leaves have no branches and carry the concrete configuration values.
4. Main C++ Software Modules/Classes
GipsyX/RTGx organizes orbit integration, measurement editing, filtering, smoothing, and signal modeling into reusable computational modules. These modules represent geodetic observations, spacecraft dynamics, Earth deformation, and propagation effects for parameter estimation.
- Core modules: The software separates orbit integration, filtering, smoothing, and data editing into broad computational modules.Filtering fits the linearized model forward in time, while smoothing fits it over all time.
- Data editing: GNSS data editing detects phase breaks and gross outliers using turbo-edit and continuity checks on phase combinations.Turbo-edit uses pseudorange−phase averages, while continuity checks use low-degree polynomial fits.
- Data editing: Dual-frequency phase differences suppress receiver and transmitter clock jumps, while polynomial removal reduces slowly varying ionospheric signals.Single differences across satellites are also available as an option.
- Orbit integration: Orbit modeling combines spherical-harmonic Earth gravity with tidal, relativistic, atmospheric-drag, albedo, and solar-radiation-pressure forces.Radiation-pressure models support arbitrary panel configurations and JPL-specific empirical GPS models.
- Orbit integration: A general-purpose adaptive ODE solver integrates satellite states and parameter partial derivatives while detecting force-model discontinuities.The solver includes multistep and embedded Runge–Kutta methods and handles shadow transitions and thruster firings.
- Signal model: The signal model represents GNSS range through reception and transmission times, tropospheric and ionospheric delays, clock effects, and geometric propagation.Phase observations add phase-bias and ambiguity terms, while multi-GNSS code biases are estimated by constellation with one reference constellation.
- Earth and station modeling: Earth-based observations require modeling crustal deformation and inertial Earth orientation, with polar motion and hour angle among the adjustable parameters.GipsyX/RTGx implements the IERS standards for these models.
5. Main executable, rtgx
The primary executable, rtgx, contains the main processing loop, while its detailed operation is documented separately in an electronic supplement.
- rtgx executable: The main loop of the primary executable is described in electronic supplement 2, “rtgx main loop.”The supplied passage identifies the supplement as the location of the detailed description.
6. Sample Use Cases, Accuracy, Precision
JPL distributes multiple GNSS product types for operational use and IGS combination, with GipsyX supporting validated production, Earth-rotation estimation, and multi-constellation analysis. Reported comparisons show millimeter- to micrometer-scale agreement for Earth orientation and continued operational availability after the software transition.
- Operational products: JPL delivers fiducial-free, no-net-rotation, and ITRF-tied GNSS products for geodetic use and combination by the IGS.The product classes differ in their terrestrial-frame constraints and transformation files.
- Multi-constellation products: Four-constellation products routinely process GPS, BeiDou, Galileo, and GLONASS data, with an average of 102 satellites.The products include consistent orbital states, transmitter clocks, spacecraft attitude, Earth rotation parameters, and widelane phase biases.
- Operational products: January 29, 2017 marked JPL’s seamless transition from GIPSY-generated to GipsyX-generated GNSS analysis-center products.The full 1994–2018 span was subsequently reprocessed using GipsyX consistently with IGS repro2 standards.
- Product evaluation: Orbit and clock precision is assessed from five-hour daily overlaps, whereas accuracy is compared with combined final IGS orbit and clock solutions.The IGS Final product is a weighted average of contributing analysis centers, including JPL.
- Earth orientation: 20 µas polar-motion differences correspond to 0.6 mm at Earth’s surface, while LOD differences reach 16 µs/day or 1.2 mm over one day.These values are reported for JPL Final no-net-rotation solutions relative to the combined IGS solution.
- Earth orientation: 27–49 µas, 26–44 µas, and 13–18 µs/day are the reported scatter ranges for JPL Xp, Yp, and LOD relative to six other IGS analysis centers.The comparison covers January 29, 2017 through April 20, 2019.
6.2. Real-time GNSS Orbit and Clock solutions, GNSS Differential Corrections
RTGx provides real-time GNSS orbit, clock, positioning, and differential-correction capabilities through coupled and decoupled filtering. Its reported performance includes centimeter-level URE and positioning, while snapshots support rapid filter restarts without loss of convergence.
- Differential corrections: Orbit states are generated every 60 seconds and clock solutions at 1 Hz, with latency never exceeding 6 seconds relative to measurement epochs.GDGPS also distributes differential corrections supporting real-time ambiguity resolution.
- Real-time orbit and clocks: Real-time URE relative to post-processed solutions is typically 5 cm RMS.URE combines radial, cross-track, along-track, and clock errors using the paper’s stated approximation.
- Robust real-time operations: Decoupled partitions isolate satellite or station estimates so unhealthy satellites can be monitored without influencing other estimated parameters.The GDGPS System uses this arrangement to protect the remaining constellation from mismodeling, including maneuver-related errors.
- Real-time positioning: RTGx produces real-time time series for hundreds of ground sites and makes them available for global natural-hazard monitoring.The sites are estimated kinematically within the GDGPS orbit-determination system.
- Real-time orbit and clocks: Integer ambiguity resolution lowers real-time URE relative to float solutions, approaching 5 cm RMS with 90 or more tracking sites.The comparison uses GDGPS data and JPL daily post-processed solutions as references.
- Real-time positioning: Kinematic ground-site positioning within the real-time orbit filter provides few-centimeter accuracy at 1 Hz.The filter retains temporal correlations among all parameters, unlike forward-filtering point positioning.
- Robust real-time operations: Snapshot-based hot starts let RTGx restart with instant convergence and rapidly correct past errors before catching up to continuous operation.With identical post-snapshot data, the restarted solution is identical to the continuously running filter.
6.3. Precise Point Positioning with GNSS
GipsyX/RTGx evaluates precise point positioning across 59 ITRF2014 stations using several reference-frame constraints and GNSS processing models. Static solutions achieve millimeter-level repeatability, while kinematic performance depends strongly on tropospheric treatment and station/time-series design.
- Static PPP: NF, NNR, and NNRTS constraints produced nearly identical results in the static positioning analysis.The processing included separate PPP runs and Helmert transformations where required to express results in the reference frame.
- Static PPP: 0.4 mm was the average offset associated with omitting IONEX corrections, although IONEX use did not improve repeatability in this test.The authors still recommend IONEX for consistency with Final POD models and because other analyses found possible z-component offsets without it.
- Static PPP: 2, 2, and 6 mm ENV frame repeatability was obtained for east, north, and vertical components, respectively.The vertical result was slightly worse than the earlier 6.0 mm value, while the east and north results were almost identical; station selection differences were cited as a possible explanation.
- Interpretation and limitations: Repeatability estimates can be biased by short time periods, inhomogeneous station distributions, and seasonal loading signals.Vertical residuals were largely explained by atmospheric and hydrological loading, while horizontal displacements showed seasonality associated with effects such as snow cover.
- Kinematic PPP: Kinematic PPP was tested every 5 minutes for 45 ITRF2014 stations, with vertical repeatability improving significantly when static-solution tropospheric parameters were fixed.The standard kinematic setup used a loose 1 m/√s position random walk and a constrained zenith delay to reduce correlation with vertical position.
6.4. Precise orbit and clock determination of low Earth orbiters
GipsyX/RTGx extends LEO precise orbit determination beyond earlier GIPSY capability by supporting newer GNSS constellations, DORIS, SLR, reduced-dynamic processing, and embedded real-time POD. Tests with Jason-2, LAGEOS, GRACE, and GRACE-FO demonstrate centimeter- to millimeter-level orbit and clock performance.
- Capabilities: GipsyX/RTGx supports precise post-processing, reduced-dynamic orbit determination, and embedded real-time POD for LEO satellites.It retains GIPSY’s GPS-based post-processed LEO capability while adding newer GNSS constellations, DORIS, and SLR.
- GRACE POD: 1.6 mm was the average daily baseline standard deviation for GRACE GPS POD relative to KBR measurements.The processing used 10-second GPS data, antenna calibrations, reduced dynamics, and bias fixing; the result improved on a reported 10 mm comparison without bias fixing.
- GRACE time synchronization: 20.3 ps was the median GRACE relative-clock precision, significantly below the 160 ps requirement.The double-difference overlap metric removes the reference clock and evaluates relative clock precision over the central five hours of six-hour overlaps.
6.5. Multi-Technique Geodesy
GipsyX/RTGx combines GPS and SLR observations at the observation level to connect radiometric and optical geodetic systems. Space ties from Jason-2 and the GRACE tandem reproduce independent ground ties at centimeter-level agreement without relying on ground survey ties.
- Observation-level combination: GPS and SLR data were combined at the observation level in a multi-day solution containing 31 GPS satellites and five dedicated SLR targets.The experiment also included Jason-2 and GRACE A and B, which carry both GPS receivers and SLR corner reflectors.
- Results: Centimeter-level residuals were uniform and precise across GPS and SLR observations, supporting coherence between radiometric and optical measurements.The combined network linked GPS and SLR without ground survey ties through low-Earth-orbit satellite observations.
- Observation-level combination: 45 GPS and 18 SLR ground stations contributed to the combined solution, with satellite and station positions estimated simultaneously alongside Earth orientation.Range biases were estimated for a minority of SLR stations to compensate for systematic SLR errors.
6.6. Building Terrestrial Reference Frames and Fitting Ground Site Time Series
GipsyX provides tools to combine technique-specific solutions into terrestrial reference frames, simulate networks, transform coordinates, and fit station time series. These tools support reference-frame monitoring and scalable estimation of positions, velocities, seasonal terms, and breaks.
- Reference-frame construction: GipsyX combines daily or weekly solutions from individual geodetic techniques and their ties into a single terrestrial reference frame.A frame built from all four geodetic techniques and ties was compared with ITRF2014 and DTRF2014.
- Network simulation: Network simulations generate position, velocity, break, seasonal, and correlated geocenter, scale, or rotational noise models for future-network studies.Simulated solutions can be processed like real solutions to test software and predict future network performance.
- Reference-frame transformations: Daily x-file transformations represent translations, rotations, and scale between reference frames at a specified epoch.The transformations use an input frame to predict coordinates and estimate parameters against GNSS positions.
- Reference-frame transformations: GNSS-only geocenter values have approached IGS14 more closely over time as modeling and data quality improved.The cited x-file comparison reports standard deviations of 4.7 mm, 5.6 mm, 6.8 mm, and 1.7 mm for TX, TY, TZ, and scale, respectively.
- Ground-site time series: Time-series tools estimate positions, velocities, breaks, and seasonal terms with automatic break detection and outlier removal.Break detection uses an exhaustive search with an F-test, while estimation uses the same time-variable square-root information filter as frame combinations.
7. Summary
GipsyX/RTGx is presented as a robust tool for challenging positioning, navigation, timing, and science applications, incorporating lessons from previous software generations. It underlies JPL’s GNSS orbit-determination operations and supports geodetic analyses worldwide.
- GipsyX/RTGx is described as a robust and powerful tool for geodetic data analysis and simulations.Its design incorporates decades of expertise and lessons learned from previous software generations.
- The software incorporates expertise and lessons from previous generations for challenging positioning, navigation, timing, and science applications.
- GipsyX/RTGx now underlies all GNSS orbit-determination operations at JPL.
- Hundreds of academic and research licenses support geodetic analyses and science operations across the globe.