Identifying Exoplanets with Deep Learning. IV. Removing Stellar Activity Signals from Radial Velocity Measurements Using Neural Networks

Beurs, Zoë L. de; Vanderburg, Andrew; Shallue, Christopher J.; Dumusque, X.; Cameron, A. Collier; Buchhave, Lars A.; Cosentino, R.; Ghedina, A.; Haywood, R. D.; Langellier, Nicholas; Latham, David W.; López‐Morales, Mercedes; Mayor, M.; Micela, G.; Milbourne, Timothy; Mortier, A.; Molinari, E.; Pepe, F.; Phillips, David F.; Pinamonti, M.; Piotto, G.; Rice, Ken; Sasselov, Dimitar; Sozzetti, A.; Udry, S.; Watson, Christopher A.

doi:10.3847/1538-3881/ac738e

Cited by 27 publications

(19 citation statements)

References 101 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, we have explored solving the problem of stellar variability by simply getting lots of observations with current accuracy. There are complementary efforts to try to measure velocities in ways that are less contaminated by stellar variability that may provide additional improvements (e.g., Davis et al 2017;Dumusque 2018;Wise et al 2018;Collier Cameron et al 2021;de Beurs et al 2022;Cretignier et al 2021;Holzer et al 2021).…”

Section: Statistical Methodology For Detecting Planets In Presence Of...mentioning

confidence: 99%

“…Additional improvements (relative to our results) may be possible if combined with simultaneous activity time-series, which have not been accounted for in these calculations. Further improvements may be also possible if we are able to correct stellar activity in the wavelength domain (e.g., Davis et al 2017;Dumusque 2018;Wise et al 2018;Collier Cameron et al 2021;de Beurs et al 2022;Cretignier et al 2021;Holzer et al 2021). With these approaches, larger apertures may still be valuable or even critical, if our ability to correct for or characterize stellar variability depends on S/N and/or resolution.…”

Section: Active Regions and Rotationally Linked Variabilitymentioning

confidence: 99%

See 1 more Smart Citation

Impact of Correlated Noise on the Mass Precision of Earth-analog Planets in Radial Velocity Surveys

Luhn

Guo

Gilbertson

et al. 2023

View full text Add to dashboard Cite

Characterizing the masses and orbits of near-Earth-mass planets is crucial for interpreting observations from future direct imaging missions (e.g., HabEx, LUVOIR). Therefore, the Exoplanet Science Strategy report recommended further research so future extremely precise radial velocity surveys could contribute to the discovery and/or characterization of near-Earth-mass planets in the habitable zones of nearby stars prior to the launch of these future imaging missions. Newman et al. (2023) simulated such 10 yr surveys under various telescope architectures, demonstrating they can precisely measure the masses of potentially habitable Earth-mass planets in the absence of stellar variability. Here, we investigate the effect of stellar variability on the signal-to-noise ratio (S/N) of the planet mass measurements in these simulations. We find that correlated noise due to active regions has the largest effect on the observed mass S/N, reducing the S/N by a factor of ∼5.5 relative to the no-variability scenario; granulation reduces by a factor of ∼3, while p-mode oscillations has little impact on the proposed survey strategies. We show that in the presence of correlated noise, 5 cm s−1 instrumental precision offers little improvement over 10 cm s−1 precision, highlighting the need to mitigate astrophysical variability. With our noise models, extending the survey to 15 yr doubles the number of Earth-analogs with mass S/N > 10, and reaching this threshold for any Earth-analog orbiting a star >0.76 M ⊙ in a 10 yr survey would require an increase in the number of observations per star from that in Newman et al. (2023).

show abstract

Section: Statistical Methodology For Detecting Planets In Presence Of...mentioning

confidence: 99%

Section: Active Regions and Rotationally Linked Variabilitymentioning

confidence: 99%

Impact of Correlated Noise on the Mass Precision of Earth-analog Planets in Radial Velocity Surveys

Luhn

Guo

Gilbertson

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Maldonado et al 2019;Milbourne et al 2019Milbourne et al , 2021 and data analysis to mitigate stellar activity in RV measurements (e.g. Collier Cameron et al 2021;de Beurs et al 2022;Langellier et al 2021).…”

Section: Solar Observations With Harps and Harps-nmentioning

confidence: 99%

“…The HARPS-N solar telescope and the goal of the project is described in Dumusque et al (2015) and Phillips et al (2016). Various important scientific results have already come out from the obtained data, in terms of data reduction (e.g., Collier Cameron et al 2019;Dumusque et al 2021), understanding of solar activity (e.g., Maldonado et al 2019;Milbourne et al 2019Milbourne et al , 2021, and data analysis to mitigate stellar activity in RV measurements (e.g., Collier Cameron et al 2021;Langellier et al 2021;de Beurs et al 2022).…”

Section: Solar Observations With Harps and Harps-nmentioning

confidence: 99%

Stellar signal components seen in HARPS and HARPS-N solar radial velocities

Moulla

Dumusque

Figueira

et al. 2023

A&A

View full text Add to dashboard Cite

Context. Radial velocity (RV) measurements induced by the presence of planets around late-type stars are contaminated by stellar signals that are of the order of a few meters per second in amplitude, even for the quietest stars. Those signals are induced by acoustic oscillations, convective granulation patterns, active regions co-rotating with the stellar surface, and magnetic activity cycles. Aims. This study investigates the properties of all coherent stellar signals seen on the Sun on timescales up to its sidereal rotational period. By combining HARPS and HARPS-N solar data spanning several years, we are able to clearly resolve signals on timescales from minutes to several months. Methods. We use a Markov Chain Monte Carlo (MCMC) mixture model to determine the quality of the solar data based on the expected airmass-magnitude extinction law. We then fit the velocity power spectrum of the cleaned and heliocentric RVs with all known variability sources, to recreate the RV contribution of each component. Results. After rejecting variations caused by poor weather conditions, we are able to improve the average intra-day root mean square (RMS) value by a factor of ∼1.8. On sub-rotational timescales, we are able to fully recreate the observed RMS of the RV variations. In order to also include rotational components and their strong alias peaks introduced by nightly sampling gaps, the alias powers are accounted for by being redistributed to the central frequencies of the rotational harmonics. Conclusions. In order to enable a better understanding and mitigation of stellar activity sources, their respective impact on the total RV must be well-measured and characterized. We are able to recreate RV components up to rotational timescales, which can be further used to analyse the impact of each individual source of stellar signals on the detectability of exoplanets orbiting very quiet solar-type stars and test the observational strategies of RV surveys.

show abstract

“…NNs can connect physical parameters and observational measurements through a statistical model, without designating a specific physical model. Recently, various machine learning techniques including NNs have been utilized in many astronomical fields e.g., to classify observations (Wu et al 2019;Walmsley et al 2021;Whitmore et al 2021), to identify structures or exoplanets (Abraham et al 2018;de Beurs et al 2022), and to predict physical parameters (Fabbro et al 2018;Ksoll et al 2020;Olney et al 2020;Sharma et al 2020;Shen et al 2022). In this study, we adopt the supervised learning approach, a type of machine learning that trains the network using labelled data sets, as we want to estimate specific physical parameters we want from quantities we can measure from observed star-forming regions.…”

Section: Introductionmentioning

confidence: 99%

Noise-Net: Determining physical properties of HII regions reflecting observational uncertainties

Kang¹,

Klessen²,

Ksoll³

et al. 2023

Preprint

View full text Add to dashboard Cite

Stellar feedback, the energetic interaction between young stars and their birthplace, plays an important role in the star formation history of the universe and the evolution of the interstellar medium (ISM). Correctly interpreting the observations of star-forming regions is essential to understand stellar feedback, but it is a non-trivial task due to the complexity of the feedback processes and degeneracy in observations. In our recent paper, we introduced a conditional invertible neural network (cINN) that predicts seven physical properties of star-forming regions from the luminosity of 12 optical emission lines as a novel method to analyze degenerate observations. We demonstrated that our network, trained on synthetic star-forming region models produced by the WARPFIELD-Emission predictor (WARPFIELD-EMP), could predict physical properties accurately and precisely. In this paper, we present a new updated version of the cINN that takes into account the observational uncertainties during network training. Our new network named Noise-Net reflects the influence of the uncertainty on the parameter prediction by using both emission-line luminosity and corresponding uncertainties as the necessary input information of the network. We examine the performance of the Noise-Net as a function of the uncertainty and compare it with the previous version of the cINN, which does not learn uncertainties during the training. We confirm that the Noise-Net outperforms the previous network for the typical observational uncertainty range and maintains high accuracy even when subject to large uncertainties.

show abstract

Identifying Exoplanets with Deep Learning. IV. Removing Stellar Activity Signals from Radial Velocity Measurements Using Neural Networks

Cited by 27 publications

References 101 publications

Impact of Correlated Noise on the Mass Precision of Earth-analog Planets in Radial Velocity Surveys

Impact of Correlated Noise on the Mass Precision of Earth-analog Planets in Radial Velocity Surveys

Stellar signal components seen in HARPS and HARPS-N solar radial velocities

Noise-Net: Determining physical properties of HII regions reflecting observational uncertainties

Contact Info

Product

Resources

About