Long-period magnetic activity in the K dwarf GJ 1137 and a new super-Earth on a 9-day orbit

Abstract

Aims: We investigate long-term radial velocity (RV) variability in the K-dwarf star GJ 1137 (HD 93083, HIP52521), a known Saturn-mass exoplanet host, and assess the role of stellar activity in shaping the observed signals. Methods: We analyse 13 years of archival high-precision spectroscopic observations obtained with the High Accuracy Radial velocity Planet Searcher spectrograph (HARPS). We performed an extensive spectroscopic analysis of the stellar activity indicators and applied an RV modelling approach, incorporating Keplerian fits, Gaussian process regression as a proxy for stellar activity, and other stellar activity diagnostics. Furthermore, we refined the orbital parameters and the minimum mass of the known exoplanet GJ 1137 b and searched for additional planetary candidates in the system. Results: We detect a long-period RV signal that, if interpreted as planetary, would suggest the presence of a Jovian analogue companion. However, our spectroscopic activity analysis provides strong evidence that this variability is induced by the star's long-term magnetic cycle ( Pcyc = 5870+(480)-(350) days) rather than by an orbiting planet. The signal is detected in both full width at half maximum (FWHM) of the crosscorrelation function and the chromospheric activity index log R'Hk. We measure the stellar rotation period to Prot = 32.3+(1.2)-(1.3) d and identify a significant short-period RV signal, which we attribute to a Super Earth with a period of 9.6412+(12)-(11) d and a minimum mass of 5.12+(0.70)-(0.69) Earth masses, making GJ 1137 a multiple-planet system.

Figures (18)

• Figure 1: SED fit for GJ 1137. The best-fitting model, BT-Settl, is displayed in black, with normalised residuals shown below. Blue points represent flux values from photometry, while purple diamonds show the flux from synthetic photometry in the same passband.
• Figure 2: HARPS RV time series for GJ 1137. The data are divided into two subsets due to known instrumental RV offsets introduced by the HARPS optical fibre upgrade in May 2015. Purple circles represent RVs obtained before the fibre change, while green squares correspond to data taken after the upgrade. The observed RV variations are primarily driven by the signal of the known Saturn-mass exoplanet GJ 1137 b, with a period of $P = 144.7$ d. In addition, the RV data reveals a second, long-term signal from a potential Jovian planet with a minimum mass of approximately $1.3\,[]{\jupitermass}$ and a period $P = 5640\pm240$ d. Subsequent analysis attributes this long-term signal to stellar magnetic activity.
• Figure 3: Raw time series of mean-subtracted: (a) RV, (c) FWHM, (e) R$'_\text{HK}$. Measurements are marked depending on the data source: blue circles for HARPS-pre, and green squares for HARPS-post. (b,d,f) Associated wide-period GLSPs of RV and activity indicators. Three false alarm probability (FAP) levels: 10%, 1%, and 0.1%, split GLSP ordinates in bands of different colour. We highlight the three most prominent peaks in each GLSP.
• Figure 4: Time series (markers) against our LTF model (black), over the whole dataset baseline in: (a) RV, (c) FWHM, (e) R$'_\text{HK}$. We implicitly correct for the HARPS-pre FWHM drift before visualising. (b,d,f) Associated GLSPs of the residual time series, accounting for the model jitter .
• Figure 5: Bayesian evidence comparison between different planetary configurations and stellar-activity kernels. Planetary configurations include circular orbit (C) and Keplerian (K) components. For stellar activity, we utilised the MEP kernel and the ESP kernel with 2, 3, and 4 harmonics. The model that we further elected for analysis assumed a model with a Keplerian signal, a circular signal, and an MEP kernel (red border; $\ln Z=-807.3$). We give the Bayesian factor $\Delta\ln Z$ of remaining models relative to this model.