The impact of observation losses on IVS-R1/R4 VLBI sessions

Basic Information

• Paper ID: 2510.11436
• Title: The impact of observation losses on IVS-R1/R4 VLBI sessions
• Authors: Tiege McCarthy, Lucia McCallum (University of Tasmania)
• Classification: physics.geo-ph
• Publication Date: Draft version October 14, 2025
• Paper Link: https://arxiv.org/abs/2510.11436

Abstract

This study analyzes the impact of observation data losses on Earth Orientation Parameters (EOPs) and station position accuracy in geodetic VLBI observation sessions of IVS-R1 and IVS-R4 during the decade 2014-2023. Statistical analysis of 1,030 sessions reveals that on average 25.3% of planned observations fail to enter the final analysis. This results in performance degradation compared to planned versions of: UT1-UTC 18.8%, three-dimensional station positions 19.2%, X/Y nutation 12.1%/11.3%, and x/y polar motion 28.7%/22.9%. The study finds that X/Y nutation estimates exhibit strong robustness to typical observation losses, while other geodetic parameters experience severe performance degradation when observation losses reach 15-19%.

Research Background and Motivation

Problem Definition

Geodetic VLBI is one of four space geodetic techniques, utilizing networks of radio telescopes to observe weak signals from extragalactic radio sources such as quasars to determine Earth orientation parameters and station positions. Each 24-hour observation session has a specially optimized observation schedule, but not all planned observations meet the quality standards required for analysis due to station failures, sensitivity issues, or source effects.

Research Significance

IVS-R1 and IVS-R4 are two core rapid-turnaround session series of the International VLBI Service (IVS), beginning on Mondays and Thursdays respectively, and are the primary VLBI sessions for determining EOPs. These sessions have been operating since 2002, with network participation reaching a peak of 14 stations in 2017.

Existing Limitations

Current observations remain a highly manual process, with many stations requiring operator setup and monitoring. Observation losses often result from operator errors, insufficient performance monitoring, or network communication failures—issues that are largely preventable.

Research Motivation

To understand IVS network efficiency based on long-term statistics, highlight the impact of observation losses on geodetic parameter estimation, provide realistic performance impact scaling factors for future simulation studies, and establish a basis for improving network operations.

Core Contributions

1. First systematic quantitative analysis: Comprehensive statistical analysis of observation losses from 1,030 IVS-R1/R4 sessions over a decade
2. Performance impact assessment: Quantitative analysis of the specific impact of observation losses on the accuracy of various geodetic parameter estimates
3. Critical threshold identification: Determination of critical observation loss thresholds (15-19%) causing severe performance degradation
4. Simulation correction factors: Provision of performance correction factors considering actual observation losses for future simulation studies
5. Series-to-series difference analysis: Revelation of significant differences in performance degradation patterns between R1 and R4 sessions

Methodology Details

Task Definition

Input: Plan files and vgosDB geodetic database for IVS-R1/R4 sessions Output: Observation loss statistics and quantification of their impact on EOP estimation accuracy Constraints: Only observations with quality code ≥5 are considered; baseline data between co-located stations are excluded

Data Processing Pipeline

Sample Selection

• Time Range: Decade 2014-2023
• Session Count: 1,030 sessions (515 R1 + 515 R4)
• Data Source: Plan files and vgosDB database downloaded from CDDIS

Observation Quality Filtering

The study employs two filtering standards:

1. Loose standard: All observations in vgosDB (median observation ratio 0.864)
2. Strict standard: Only observations with quality code ≥5 (median observation ratio 0.747)

Through comparison with IVS analysis reports, the strict standard is selected as the primary analysis basis, as it more closely aligns with the observation ratio (0.693) used in actual analysis.

Simulation Methodology

VieSched++ Simulator

Performance assessment using the Monte Carlo simulator in the VieSched++ software package:

• Observation noise: 25 ps white noise per observation (corresponding to 17.68 ps per station)
• Clock drift: Integrated random walk process with Allan standard deviation of 1×10^-14 s after 50 minutes
• Tropospheric delay: Average turbulence factor Cn = 1.8×10^-7 m^-1/3

Parameter Estimation Strategy

• Clock parameters: Piecewise linear drift at 60-minute intervals
• Tropospheric parameters: Zenith wet delay (30-minute intervals) and atmospheric gradients (180-minute intervals)
• Station coordinates and EOPs: Estimated once per 24-hour session
• Datum realization: All stations participate in datum realization
• Simulation runs: 1,000 runs per scenario for accurate repeatability values

Performance Evaluation Metrics

Relative performance comparison method is adopted:

• Observation quantity: Actual observations / Planned observations
• Repeatability parameter: Planned repeatability / Actual repeatability (smaller values indicate worse performance)
• Critical degradation: Defined as doubled repeatability (relative performance <0.5)

Experimental Setup

Dataset Characteristics

• Network scale: Median of 10 stations
• Observation quantity: Up to 3-fold variation between sessions
• Geographic distribution: Global network with relatively sparse coverage in Southern Hemisphere and East Asia

Evaluation Metrics

Primary assessment of repeatability for the following geodetic parameters:

• UT1-UTC: Earth rotation parameter
• Three-dimensional station coordinates: Station position accuracy
• X/Y nutation: Earth nutation parameters
• x/y polar motion: Polar motion parameters

Statistical Testing Methods

• Mann-Whitney U test: Detection of median differences
• Kolmogorov-Smirnov test: Detection of distribution shape differences
• Significance level: p < 0.05

Experimental Results

Main Results

Observation Loss Statistics

Analysis of observation losses over the decade shows:

• Total observation loss: Median 25.3%
• Processing-related loss: 13.4% (station non-participation or recording errors)
• Quality-related loss: 11.9% (insufficient quality detection)
• Station efficiency: Approximately 87%

Performance Impact Quantification

Median performance losses for each parameter:

Critical Performance Degradation Analysis

Observation loss thresholds at which 10% of sessions experience critical degradation for different parameters:

• X/Y nutation: 51% and 55% observation loss (best performance)
• Three-dimensional station positions: 19% observation loss
• UT1-UTC: 17% observation loss
• X/Y polar motion: 15% observation loss (most susceptible)

Annual Trend Analysis

Observation losses show a deteriorating trend during 2014-2023:

• 2014: Observation ratio 86.1%
• 2023: Observation ratio 59.1%
• Performance of all parameters declines with increasing observation loss

R1 vs. R4 Series Comparison

Statistical test results show:

• No significant difference: Observation quantity, X/Y nutation parameters
• Significant difference: UT1-UTC, three-dimensional coordinates, x/y polar motion parameters
• R4 series disadvantage: Poorer performance in three-dimensional coordinates and polar motion parameters, possibly related to sparse coverage in East Asia and Southern Hemisphere

Related Work

VLBI Geodetic Technology Development

• Traditional S/X dual-frequency system: Technology used in current R1/R4 sessions
• VGOS broadband system: Next-generation VLBI Global Observing System, expected to replace traditional systems
• Automated scheduling: Dynamically scheduled systems under development to reduce observation losses

Performance Optimization Research

• Network geometry optimization: Station layout optimization for specific EOP parameters
• Observation strategies: Scheduling strategies to improve network robustness to observation losses
• Quality control: Improved correlation processing and analysis methods

Conclusions and Discussion

Main Conclusions

1. Severe observation losses: Median 25.3% observation loss over the decade significantly impacts geodetic parameter accuracy
2. Parameter sensitivity differences: X/Y nutation is most robust to observation losses; polar motion parameters are most sensitive
3. Low critical thresholds: 15-19% observation loss can lead to severe performance degradation
4. Series-to-series differences: R4 sessions perform worse on certain parameters, related to network geometry

Limitations

1. Simulation-based: Comparative analysis based on simulations rather than actual observational data
2. Network-specific: Conclusions primarily applicable to global S/X networks; may not apply to VGOS or regional networks
3. Insufficient causal analysis: Lacks in-depth analysis of specific causes of observation losses
4. Rescheduling not considered: Does not evaluate the effectiveness of rescheduling when station failures are known

Future Directions

1. Automated operations: Development of performance monitoring and automatic setup tools
2. Communication improvement: Enhanced communication between correlators, analysts, and station operation teams
3. Robust scheduling: Design of observation strategies more robust to typical observation losses
4. Real-time scheduling: Development of real-time rescheduling capabilities during station failures

In-Depth Evaluation

Strengths

1. Large data scale: Comprehensive analysis of 1,030 sessions over a decade provides a reliable statistical foundation
2. Rigorous methodology: Monte Carlo simulations and multiple statistical tests ensure result reliability
3. High practical value: Provides realistic performance correction factors for future simulation studies
4. Comprehensive analysis: Covers all major geodetic parameters and comparison between two observation series

Weaknesses

1. Lack of mechanism analysis: Insufficient in-depth analysis of specific impact mechanisms of different types of observation losses
2. Limited solutions: Primarily problem diagnosis with relatively simple discussion of improvement measures
3. Insufficient predictive capability: No predictive model or early warning system for observation losses established

Impact

1. Academic contribution: First systematic quantification of VLBI observation loss impacts, filling an important research gap
2. Practical value: Provides important reference for IVS network operation optimization and future system design
3. Methodological reference: Statistical analysis methods can be generalized to other space geodetic techniques

Applicable Scenarios

1. Network operation optimization: Guides IVS network operation improvements and resource allocation
2. Simulation study correction: Provides realistic performance correction for VLBI simulation studies
3. System design reference: Provides empirical data for next-generation system design such as VGOS
4. Quality control standards: Provides basis for establishing observation quality control standards

References

1. Nothnagel et al. (2017) - IVS Service overview
2. Schartner & Böhm (2019) - VieSched++ scheduling software
3. Thomas et al. (2024) - IVS R1/R4 session performance analysis
4. Niell et al. (2018) - VGOS broadband VLBI system demonstration

This paper provides the VLBI geodesy community with important performance benchmark data, which is significant for understanding and improving the operational efficiency of the global VLBI network. The research results not only reveal deficiencies in the current system but also point the direction for future technological development.