Destruction of "Peas in a Pod?" A Candidate Multi-planet System Around the Nearby, Bright Star, HD208487

TL;DR

HD208487 is reexamined with 15+ years of RV data and Bayesian analysis to test the reality of HD208487c and search for additional companions. The authors combine GLS periodograms with the EMPEROR Adaptive Parallel Tempering MCMC to model a multi-planet system, account for instrumental offsets and stellar activity, and verify signals with independent analyses. They find strong evidence for two additional planets, c and d, with c around 923 d (Saturn-mass) and d around 1380 d (super-Neptune), near a 3:2 resonance, and discuss a dynamical history involving resonant chains and scattering. Photometric data show no transits and no activity signals at the planetary periods, and dynamical simulations suggest a plausible formation path from a more ordered resonant chain to the observed architecture; further RVs are needed to confirm a possible fourth signal and solidify the system.”

Abstract

We re-investigate the HD208487 system to test the reality of the proposed HD208487c world. We also search for additional companions using applied Bayesian statistics and 15+ years of new RV data from the HARPS and the PFS instruments that were taken post-discovery of HD208487b. The RV data was analyzed with GLS Periodograms, followed by Bayesian analysis using the EMPEROR code. We scrutinised various stellar activity indices to search for any corresponding peaks in the power spectra, correlations with the RV measurements, or significant signals from a Bayesian analysis methodology. Finally, photometric data was checked to test for any transits or possible activity manifestations that could lead to possible false RV signals or excess noise. Our analysis points towards a candidate second planet in the system, positioned near the period of a previously proposed and subsequently challenged signal. This signal, HD208487c, would relate to a cool Saturn with an orbital period of 923.06 +2.02 -2.76 d and a minimum mass of Mj sini = 0.32 +/- 0.01Mj. Our analysis also gives rise to a newly discovered candidate planet, HD208487d, which would be the result of a cool super-Neptune/sub-Saturn with a period of 1380.13 +19.20 -8.25 d and a minimum mass of Mj sini = 0.15 +/- 0.01Mj. Neither stellar activity indices nor photometric data show signals statistically matching these periods. We have uncovered a candidate three planet system that would consist of an inner gas giant, a central Saturn and an outer super-Neptune/sub-Saturn. A dynamical analysis suggests that gravitational scattering of an initially ordered, equally-spaced system in a long resonant chain of six Neptunes can explain the current proposed architecture of HD208487. More RVs may also shed light on the reality of a fourth Doppler signal uncovered in the data that sits close to the 2:1 period-ratio with signal of HD208487c.

Figures (11)

• Figure 1: All 219 mean-subtracted radial velocity measurements of HD208487, including 48 UCLES (grey), 20 HARPS-TERRA (red), 106 HARPS-TERRA2 (olive), and 45 PFS (teal) observations.
• Figure 2: Top panel: original GLS periodogram of the combined and mean-subtracted RV data (black). The periodogram of the window function (sampling) for the combined data (green) is also shown. Second panel: GLS periodogram after fitting and removing the first detected signal. Third panel: GLS periodogram after fitting and removing both the first and second signals. Bottom panel: GLS periodogram after fitting and removing all three previously detected signals. The blue, dotted horizontal lines represents the 0.1% false alarm probability level estimated from 5000 bootstrap resamplings.
• Figure 3: Top to bottom, left to right: phase-folded, best-fit model for the first, second, third, and complete Keplerian model.
• Figure 4: Top, left to right: final posterior distribution of the period for the first, second, and third Keplerians from the best-fitting parallel-tempered MCMC run. Bottom, left to right: final posterior distribution of the semi-amplitude for the first, second, and third Keplerians from the best-fitting parallel-tempered MCMC run. The numbers at the top of each figure correspond to the mean, variance, skewness and kurtosis, median, and mode, respectively. The solid line represents a Gaussian curve with the same mean and variance.
• Figure 5: GLS periodogram of the available stellar activity indices. From top to bottom: PFS S-index, TERRA S-index, TERRA FWHM, TERRA BIS, TERRA2 S-index, TERRA2 FWHM, TERRA2 BIS, Combined S-index, Combined BIS, Combined FWHM. The green, solid vertical lines show the best-fit periods of the three planetary signals. The blue, dotted horizontal lines represents the 0.1% false alarm probability (FAP) level estimated from 5000 bootsrap resamplings.