Source-linked AI summary
The Parkes Pulsar Timing Array Project: Second data release
M. Kerr, D. J. Reardon, G. Hobbs, R. M. Shannon, R. N. Manchester, S. Dai, C. J. Russell, S. -B. Zhang, W. van Straten, S. Osłowski, A. Parthasarathy, R. Spiewak, M. Bailes, N. D. R. Bhat, A. D. Cameron, W. A. Coles, J. Dempsey, X. Deng, B. Goncharov, J. F Kaczmarek, M. J. Keith, P. D. Lasky, M. E. Lower, B. Preisig, J. M. Sarkissian, L. Toomey, H. Wang, J. Wang, L. Zhang, X. Zhu
TL;DR
Long-term pulsar timing supports studies including nanohertz gravitational-wave searches, while detailed calibration and RFI mitigation matter for the resulting data. This paper releases PPTA DR2 with a provenance-oriented processing pipeline; the Parkes UWL system is expected to reduce observing time for similar results and support a larger observed pulsar sample.
Problem
Pulsar-timing studies depend on understanding timing residuals and systematic errors, while calibration and RFI mitigation affect the data products.
Source: §1 · pp. 1–2Method
The paper presents a public PPTA data release and describes a processing pipeline intended to preserve the methodology and provenance from raw data to arrival times.
Source: §3 · p. 6, §6 · p. 17Results
The DR2 release presents PPTA observations spanning 2004 to 2018, produced with a new pipeline for high-precision pulsar timing.
Source: §6 · p. 17Takeaways & Limitations
The authors expect DR2 to support nanohertz gravitational-wave searches and other studies of solar-system dynamics, terrestrial time standards, and millisecond pulsars.
Source: §6 · p. 17The timing-noise and instrumental-jump models are preliminary and should be reassessed for each application.
Source: §5.1 · p. 16
Abstract
from arXiv · showhide
We describe 14 years of public data from the Parkes Pulsar Timing Array (PPTA), an ongoing project that is producing precise measurements of pulse times of arrival from 26 millisecond pulsars using the 64-m Parkes radio telescope with a cadence of approximately three weeks in three observing bands. A comprehensive description of the pulsar observing systems employed at the telescope since 2004 is provided, including the calibration methodology and an analysis of the stability of system components. We attempt to provide full accounting of the reduction from the raw measured Stokes parameters to pulse times of arrival to aid third parties in reproducing our results. This conversion is encapsulated in a processing pipeline designed to track provenance. Our data products include pulse times of arrival for each of the pulsars along with an initial set of pulsar parameters and noise models. The calibrated pulse profiles and timing template profiles are also available. These data represent almost 21,000 hrs of recorded data spanning over 14 years. After accounting for processes that induce time-correlated noise, 22 of the pulsars have weighted root-mean-square timing residuals of < 1 ${\mu}$s in at least one radio band. The data should allow end users to quickly undertake their own gravitational-wave analyses (for example) without having to understand the intricacies of pulsar polarisation calibration or attain a mastery of radio-frequency interference mitigation as is required when analysing raw data files.
1 INTRODUCTION
Pulsar timing arrays use long-term, precise millisecond-pulsar timing for diverse studies, including searches for nanohertz gravitational waves. This paper releases current PPTA arrival times and describes the pipeline and intermediate products, emphasizing calibration, RFI mitigation, and traceable processing.
1 INTRODUCTION
- Pulsar timing arrays use long-term, precise millisecond-pulsar timing to study residual correlations and pursue science including nanohertz gravitational-wave searches.PTA sensitivity to gravitational waves from binary supermassive black holes complements ground-based detectors targeting stellar-mass-object mergers.Source: §1 · p. 1
- The PPTA contributes southern-sky pulsar observations to international efforts, with the PPTA, EPTA, and NANOGrav combining in the IPTA.Parkes has had almost exclusive access to pulsars south of δ < −40°, and the IPTA's primary goal is detecting nanohertz gravitational waves.Source: §1 · pp. 1–2
- Prior PPTA data releases have supported gravitational-wave searches and constraints as well as studies of pulsars, reference frames, and the interstellar medium.The first major release, DR1, contained data collected through 2011; subsequent work updated noise models and timing parameters.Source: §1 · p. 2
- The paper makes current PPTA pulsar times of arrival available and describes the processing pipeline and intermediate products.It aims to preserve the methodology and reasoning behind processing, and reports systematic errors and their impact.Source: §1 · p. 2
- The paper describes identified systematic errors and their impact, demonstrating the importance of detailed calibration and radio-frequency interference mitigation.Planned follow-up work includes residual-noise studies, gravitational-wave searches, and analyses of individual pulsars.Source: §1 · p. 2
- The release is publicly available, with data products designed for use with Psrchive and tempo2.The paper also outlines its sections on observing systems, processing and calibration, data products, and comparisons with previous releases.Source: §1 · p. 2
2 THE PPTA OBSERVATION SYSTEM AND PULSAR SAMPLE
The paper describes the Parkes observing systems and pulsar sample underlying PPTA DR2, including the observation span, cadence, bands, and selection considerations. It also records how the observing strategy and sample evolved.
2 THE PPTA OBSERVATION SYSTEM AND PULSAR SAMPLE
- The DR2 observations span 2004 February 6 to 2018 April 25, or 14.2 years, except for PSR J0437−4715, which includes early-science data spanning about 15 years.The observations were made with the 64-m Parkes telescope; data end before the new receiver system's installation and commissioning during 2018.Source: §2 · pp. 2, 4
- Before the wide-band system, observations used Parkes multibeam and H-OH receivers, a 10/40-cm dual-band receiver, and PDFB3, PDFB4, or CASPSR signal processors.PDFB3 ceased use after a hardware failure in April 2014; PDFB4 and CASPSR were used thereafter until UWL.Source: §2 · pp. 2, 4
- PPTA pulsar suitability depends partly on obtaining precise arrival times in at least two bands, since dispersion-measure variations can dominate residuals without multiband observations.The release table summarizes pulse, dispersion, orbital, profile-width, and flux-density parameters; flux densities vary with interstellar scintillation, so reported values are long-term averages.Source: §2 · p. 4
- The observing strategy adapted to the evolving pulsar sample and RFI environment, inspecting profiles after about five minutes and switching targets or bands when scintillation or RFI was unfavorable.A pulsar could be re-observed after an interval exceeding the diffractive timescale.Source: §2 · p. 4
- PPTA pulsars were typically observed every two to three weeks across three bands, while pre-2004 legacy 20-cm observations shown in the cadence figure were not reprocessed for DR2.Those earlier observations can be obtained from the M+13 collection if longer data spans are needed.Source: §2 · p. 4, Figure 1 · p. 5
- PSR J1832−0836 is faint at 10 and 40 cm, complicating dispersion-measure correction, yet its three narrow profile components allow rms residuals below 1 µs.Until UWL, PPTA observed it only in the 20-cm band; the data can be combined with multiband IPTA observations.Source: §2 · p. 6
3 THE DATA PROCESSING PIPELINE
The pipeline documents the processing from raw data to pulse times of arrival, with explicit treatment of interference, calibration, instrumental problems, and timing offsets.
3 THE DATA PROCESSING PIPELINE
- The pipeline uses human-readable psrsh scripts to document the processing from archived raw data to final ToA products.The stated goal is to make the reduction transparent; the code and ancillary files are open source.Source: §3 · p. 6
3.1 RFI mitigation
- The 40 cm band moved to 724 MHz after 4G signals caused system nonlinearities; affected sub-integrations are excised and heavily affected observations discarded.At 20 cm, aircraft radar and satellite transmissions can also cause nonlinearities or contaminate bandwidth, leading to discarding affected observations.Source: §3.1 · pp. 6–7
3.2 Calibration Methods
- Calibration uses a cycled noise diode and Hydra A to assess differential gain and phase and flux density; a full receiver model is applied to the 20 cm multibeam system.That system shows substantial feed non-orthogonality and cross-polarisation; other systems receive differential-gain and bandpass calibration.Source: §3.2 · pp. 7, 9
3.3 Identification and Correction of Instrumental Problems
- Rule sets and annotated observation lists guide processing and propagate quality flags; observations with unreliable timing may still be useful for some astrophysical analyses.The identified problems include missing or incorrect metadata and instrumental issues recorded during the observing program.Source: §3.3 · p. 9
3.4 Instrumental Timing Offsets
- Most previously used instrumental delays are accurate to a few tens of nanoseconds; newly identified bandpass-change and saturation delays are small and difficult to isolate.The new delays are recorded as tempo2 JUMP parameters, with observation flags triggering the appropriate correction.Source: §3.4 · pp. 9–10
4 THE PPTA DR2 TIMING DATA SET
The PPTA DR2 timing data set provides calibrated, band-averaged and sub-banded pulse arrival times with timing and noise models. Its residual plots show systematic structure before noise fitting and both frequency-independent and band-dependent features after dispersion-measure variations are removed.
Time reference
- ToAs are referenced to the Observatory time system, with clock corrections transferring them through UTC(PKS), UTC(GPS), TT(TAI) and TT(BIPM18).Daily averaging of GPS offsets gives an estimated reference-time accuracy of a few nanoseconds.Source: §4 · p. 10
ToA generation
- Calibrated profiles are matched to receiver- and band-specific templates to obtain ToAs, while sub-banded ToAs help account for profile evolution and frequency changes.Sub-band counts vary dynamically from 2 to 32; centre-frequency variations can add measurement noise when averaging across a band.Source: §4 · p. 10
Observational properties
- Tables 2 and 3 describe observational properties, including data spans, integration times, ToA counts and uncertainty summaries.Table 2 distinguishes band-averaged and sub-banded ToA counts and their mean and median uncertainties.Source: §4 · pp. 10, 15, Table 2 · p. 11, Table 3 · p. 12
Noise models
- Initial noise models use power laws for frequency-independent red noise and frequency-dependent DM variations, fitted with Fourier components up to a 1/60-day frequency.DM time derivatives are also included in the timing model; additional band-dependent models and event terms are used for specified pulsars.Source: §4 · p. 15
- Exponential DM and magnetospheric events are modelled with chromatic exponential functions and included in the timing model as extra parameters.The function uses event epoch, amplitude, decay timescale and spectral index; it phenomenologically matches the arrival-time function.Source: §4 · p. 15
Residuals
- Before fitting DM variations and frequency-independent red noise, many pulsars show large systematic residuals, and their weighted rms values reflect those variations.Figure 5 displays the 10 cm, 20 cm and 40 cm bands in blue, cyan and red, respectively.Source: Figure 5 · p. 13, §4 · p. 15
- After DM variations are removed, residuals highlight frequency-independent timing noise in some pulsars and remaining band-dependent features in others.For PSR J1939+2134, unexplained 40 cm features may result from interstellar scattering changes that modify the pulse profile.Source: Figure 6 · p. 14, §4 · p. 15
Data products
- The release includes pulse arrival times, initial timing ephemerides and noise models, plus profiles, templates and calibration files used in processing.Select files for high-precision timing are applied to the data collection.Source: §4 · pp. 10, 16
5 DISCUSSION
PPTA DR2 extends the data spans and adds sub-band ToAs alongside band-averaged ToAs, with preliminary timing-noise and instrumental-jump models that require reassessment for specific uses. Known timing events and the value of continued observations shape how the data can support precision timing and gravitational-wave studies.
5.1 The DR2 Data Set
- DR2 adds eight years of observations from 2011 to 2018; most pulsars now have about 14-year spans, and PSR J1909−3744 has twice its DR1 span.The longer spans exceed Jupiter’s orbital period, improving the ability to detect solar-system ephemeris errors.Source: §5.1 · p. 16
- Sub-band ToAs enable pulse-shape, interstellar-medium, flagging, and residual-scatter studies, while their computational cost can favor band-averaged ToAs.For PSR J0437−4715, sub-band files contain 29,316 ToAs; the choice depends on processing goals, pulsar, and available computing power.Source: §5.1 · p. 16
- DR2 timing-noise and instrumental-jump models are preliminary and should be reassessed for each application.The release supplies initial pulsar parameters, jumps, and noise models; some are covariant, so initial Bayesian noise parameters are also provided.Source: §5.1 · p. 16
5.2 Known timing events affecting the DR2 pulsars
- Known events include profile variations, extreme scattering events, a long-period binary candidate, interstellar-medium under-densities, and solar-wind dispersive delays.Solar-wind delays affect PSRs J1022+1002, J1730−2304, and J1824−2452A near the Sun; those epochs need care in precision-timing studies.Source: §5.2 · p. 16
5.3 Continuing the PPTA observations
- A gravitational-wave background detection requires strong evidence for Hellings–Downs correlations across a large pulsar sample timed over long spans.A few best-timed pulsars currently dominate bounds on gravitational-wave amplitude.Source: §5.3 · p. 16
- Pulsars with substantial timing noise remain useful for studying binary evolution, globular-cluster dynamics, interstellar-medium variations, and timing noise itself.J1939+2134’s short-term precision can support searches for burst events and higher-frequency continuous gravitational waves despite red noise limiting its value for long-period background detection.Source: §5.3 · p. 17
- Twelve pulsars’ observations are unique to DR2 because Parkes was the only southern-hemisphere telescope conducting high-precision pulsar timing for most of the covered period.The listed pulsars are J0437−4715, J0711−6830, J1017−7156, J1045−4509, J1125−6014, J1446−4701, J1545−4550, J1603−7202, J1732−5049, J2124−3358, J2129−5721, and J2241−5236.Source: §5.3 · p. 17
- MeerKAT has five times Parkes’s sensitivity in the 20-cm band, but its PTA allocation is 300–400 hours annually versus almost 1,000 for Parkes.Parkes observations should continue for pulsars that are jitter-dominated with both instruments; Parkes and MeerKAT are separated by 130° longitude, enabling extended or continuous coverage.Source: §5.3 · p. 17
6 SUMMARY AND CONCLUSIONS
The paper releases PPTA DR2 observations from 2004 to 2018, processed with a pipeline intended for high-precision pulsar timing. The publicly available release is designed to support reproducible analysis, while further data-set improvements and substantial results are left to future work.
6 SUMMARY AND CONCLUSIONS
- PPTA DR2 provides observations spanning 2004 to 2018, produced with a new pipeline intended for high-precision pulsar timing.The release is publicly available as a stand-alone data set.Source: §6 · p. 17
- DR2 is expected to support gravitational-wave searches, solar-system studies, terrestrial time-standard analyses, and detailed studies of individual millisecond pulsars.The authors report that detailed noise modelling, gravitational-wave bounds, and improved pulsar timing parameters will appear in future papers.Source: §6 · p. 17
- The pipeline’s transparent, provenance-oriented processing is a step toward reproducible analysis of large pulsar data sets from compatible systems.It can work with most IPTA telescope data and emerging systems such as FAST, MeerKAT, and the planned Square Kilometre Array when their file structures are psrchive-compatible.Source: §6 · p. 18
- Incremental improvements may come from combining IPTA data, further RFI excision, and wide-band timing methods.The paper identifies significant quality or extent improvements as a subject for future data releases.Source: §6 · p. 18
- The new UWL receiver covers 704–4032 MHz in one observation and enables at least halving observing time for similar results.An approximately one-year overlap between old and new systems will enable precise measurement of timing offsets for joining new data to DR2.Source: §6 · p. 18
A EXAMPLE OF A PSRSH SCRIPT
The example script documents a PSR J1603−7202 observation’s processing from input file through calibration, intermediate averaging, and final Stokes I profile. Its commands also show how receiver settings, timing-header issues, and band-edge removal are handled.
A EXAMPLE OF A PSRSH SCRIPT
- The script starts from the raw observation, sets feed parameters to ensure calibrated data follow the PSR/IEEE convention, corrects known header-time issues, and removes 5% of band edges.It then loads pulsed-calibrator and flux-calibration information before applying calibration.Source: A EXAMPLE OF A PSRSH SCRIPT · p. 20
- After calibration, the script sets dispersion measure and a common pulsar parameter file, then generates a dynamic spectrum.The parameter file ensures that files for a given pulsar are processed using the same parameters.Source: A EXAMPLE OF A PSRSH SCRIPT · p. 20
- The profile is first averaged to 8 time sub-integrations and 32 frequency channels, then further averaged before output as Stokes I.The script’s intermediate and final files use the extensions dzt8f32, dzTF, and dzTFp.Source: A EXAMPLE OF A PSRSH SCRIPT · p. 20
B EXAMPLE OF A RULE SET USED BY THE PIPELINE
The pipeline’s rule sets document manual RFI removal so that flagging can be reproduced. They also identify catastrophic-RFI observations that should not be processed.
B EXAMPLE OF A RULE SET USED BY THE PIPELINE
- The rule sets make manual RFI flagging reproducible by recording which observations are affected and how they are flagged.The example includes rules for removing channels or sub-integrations and notes that some bands are clean enough to skip automatic RFI zapping.Source: B EXAMPLE OF A RULE SET USED BY THE… · p. 21
- The example lists observation-specific manual flagging rules, including channel removals with -z and sub-integration removals with -w.The displayed entries include rules for multiple pulsars and observations.Source: B EXAMPLE OF A RULE SET USED BY THE… · p. 21
- Catastrophic-RFI rules identify observations where every sub-integration or channel is affected and the data should not be processed.The rule maps to a Python function that treats the listed observations.Source: B EXAMPLE OF A RULE SET USED BY THE… · p. 21
C EXAMPLE OF JUMPS IN PARAMETER FILES
The release’s parameter files provide initial pulsar parameters and jumps associated with observing systems or time ranges, rather than parameters for detailed pulsar studies. The jumps account for instrumental timing offsets, with some held fixed and others fitted.
C EXAMPLE OF JUMPS IN PARAMETER FILES
- The parameter files are not suitable for detailed pulsar studies; they provide initial parameters and jumps tied to observing systems or time ranges.The example shows jump parameters for PSR J0711−6830.Source: C EXAMPLE OF JUMPS IN PARAMETER FILES · p. 22
- A jump’s final-column value marks whether it is precisely measured and held fixed (0) or fitted during timing (1).Table 4 lists both fixed offsets and offsets allowed to vary, and jump values often differ between pulsars.Source: C EXAMPLE OF JUMPS IN PARAMETER FILES · p. 22
- Before MJD 55318, some instrumental offsets depended on filterbank frequency and phase-bin configuration; firmware updates removed that dependence.After MJD 55318, timing offsets are associated with physical system movements or signal-path cable-length changes.Source: C EXAMPLE OF JUMPS IN PARAMETER FILES · p. 23
D EXAMPLE OF PULSE TOAS
The example ToA file uses TEMPO2 format, with observation details on each line. Its required columns identify the file, observing frequency, arrival time, uncertainty, and observing site, while additional flags carry metadata and jump information.
D EXAMPLE OF PULSE TOAS
- The sample lines show repeated observations with metadata including PDFB4 configuration, 10CM band, template, and fitted-jump flag.The supplied example includes several lines for the same observation file with differing arrival times and signal-to-noise values.Source: D EXAMPLE OF PULSE TOAS · p. 23
- Each TEMPO2 ToA line’s first five required columns give filename, observing frequency, arrival time, uncertainty, and site code.The site code is “pks” for the Parkes telescope, and frequency is in MHz, arrival time in MJD, and uncertainty in µs.Source: D EXAMPLE OF PULSE TOAS · p. 23
- Additional ToA columns carry flags for receiver, backend, observing band, bandwidth, channel count, template, signal-to-noise ratio, and fitted jump.Some flags identify groups used for flagging or applying time offsets; their mappings are documented in the public collection’s ruleSets directory.Source: D EXAMPLE OF PULSE TOAS · pp. 23, 25
E EXAMPLE AND USE OF A TEMPO2 SELECT FILE
TEMPO2 select files specify which observations to retain or reject using flags and processing conditions. Their rules can reject flagged observations or apply criteria such as a minimum signal-to-noise ratio and frequency ranges.
E EXAMPLE AND USE OF A TEMPO2 SELECT FILE
- TEMPO2 select files allow observations identified by specific instruments or flags to be excluded from subsequent analysis.The data collection described has already had the select files applied.Source: E EXAMPLE AND USE OF A TEMPO2 SELECT… · p. 25
- The LOGIC command evaluates a condition, while PROCESS carries out a specified processing task.An example rule rejects observations marked with the -unusual_config flag.Source: E EXAMPLE AND USE OF A TEMPO2 SELECT… · p. 25
- Observations with -snr below 15 are rejected, while the example frequency rule passes observations between 600 and 800 MHz.The selection file can be invoked in tempo2 with the -select command-line argument.Source: E EXAMPLE AND USE OF A TEMPO2 SELECT… · p. 25
F AUTHOR CONTRIBUTIONS AND AFFILIATIONS
The author list distinguishes those who led specific work or conducted a large fraction of observations from project participants listed alphabetically.
F AUTHOR CONTRIBUTIONS AND AFFILIATIONS
- The initial non-alphabetical authors led specific aspects of the data release or conducted a large fraction of the observations; the remaining authors are listed alphabetically.The latter group includes people involved in the project during the period covered by the observations.Source: F AUTHOR CONTRIBUTIONS AND AFFILIATIONS · p. 26