RedDots: Multiplanet system around M dwarf GJ 887 in the solar neighborhood

TL;DR

The paper analyzes GJ 887, a nearby M dwarf, with extensive HARPS and ESPRESSO RVs and complementary photometry to disentangle stellar activity from planetary signals. Employing Gaussian-process models with a quasi-periodic kernel and multiple periodogram techniques, the authors confirm a four-planet configuration: b (9.26 d), c (21.78 d), d (50.77 d in the habitable zone), and e (4.4249 d), with a tentative fifth signal at 2.2166 d (f) that remains unconfirmed. They demonstrate that the rotation period is around 38.7–39 d and that stellar activity aliases can mimic or obscure signals, underscoring the need for activity-aware modeling. The study highlights GJ 887 d as a potentially habitable-zone super-Earth and demonstrates the viability and limitations of detecting sub-meter-per-second RV signals in active M-dwarf systems, with implications for future atmospheric characterization and direct-imaging prospects.

Abstract

GJ 887 is a bright M dwarf in the solar neighborhood with two currently reported nontransiting exoplanets with periods of $9~\mathrm{d}$ and $21~\mathrm{d,}$ along with an additional unconfirmed signal at $50~\mathrm{d}$. We reanalyzed the system with 101 new HARPS and 12 new ESPRESSO radial velocities (RVs) secured with a cadence to confirm or refute the origin of the $50~\mathrm{d}$ signal. To do so, we searched for signals related to stellar activity in photometric data and spectroscopic indicators. We modeled the stellar activity in the RVs with Gaussian processes (GPs). With the Bayesian analysis, we confirmed a four-planet model, including the two previously known planets at periods of $9.2619\pm0.0005~\mathrm{d}$ and $21.784\pm0.004~\mathrm{d,}$ as well as two newly confirmed exoplanets: an Earth-mass planet, with a $4.42490\pm0.00014~\mathrm{d}$ period and a sub-meter-per-second amplitude, and a super-Earth with a $50.77\pm0.05~\mathrm{d}$ period located in the habitable zone (HZ). This super-Earth is the second closest planet in the HZ, after Proxima Cen b. We found an additional signal in a 2:1 resonance with the $4.4~\mathrm{d}$ planet at $2.21661\pm0.00010~\mathrm{d}$ with an amplitude of $0.37\pm0.09~\mathrm{m/s}$, which could be related to an additional planet. However, other explanations of its origin are also plausible. This signal remains a candidate, as further investigation is required to confirm its true nature. If the signal is caused by a planet, its minimum mass would be half that of Earth. We measured the stellar rotation period with the characteristic periodic timescale of the GP. We found a period of $38.7\pm0.5~\mathrm{d}$, which is consistent with the rotation period determined from photometry and other activity indices.

Figures (18)

• Figure 1: GLS periodograms of photometric TESS (upper panel) and ASAS (lower panel) data, including 0.1, 0.01, and 0.001 FAP levels in red. Significant peaks with periods higher than the 1 d sampling peak are highlighted in cyan triangles.
• Figure 2: GLS periodograms of HARPS RV data and seven activity indicators described in the plots including 0.1, 0.01, and 0.001 FAP levels. The long-term trend was removed from all time series by subtracting the $600~\mathrm{d}$ average from each observation. The blue line indicates the rotation period determined by photometry. The cyan lines represent signals that we attribute to planets (solid) and candidates (dashed).
• Figure 3: Time series of the active phase of GJ 887 in the RV and in Na D$_1$ and dLW indicators. The spectra were recorded during the 2019 RedDots observing campaign.
• Figure 4: GLS periodogram of RVs 0.1, 0.01 and 0.001 FAP levels, shown in red. The blue line indicates the rotation period determined by photometry and spectroscopy. The cyan lines represent signals that we attribute to planets (solid) and candidates (dashed). The individual panels represent residuals of the respective model in the legend.
• Figure 5: $\ell_1$-periodogram of all HARPS data. The model includes a white and a red noise model corresponding to the best likelihood as well as all signals exceeding a FAP of 0.1. Those signals are highlighted with red dots and their period.