Identifying Exoplanets with Deep Learning. IV. Removing Stellar Activity Signals from Radial Velocity Measurements Using Neural Networks
Massachusetts Institute of Technology（麻省理工学院）；University of Texas at Austin（德克萨斯大学奥斯汀分校）；University of Wisconsin-Madison（威斯康星大学麦迪逊分校）；Université de Genève（日内瓦大学）；Observatoire de Genève（日内瓦天文台）；University of St Andrews（圣安德鲁斯大学）；Facebook（脸书）；Technical University of Denmark（丹麦技术大学）；DTU Space, National Space Institute（丹麦技术大学国家空间研究所）
Zoe L. de Beurs, Andrew Vanderburg, Christopher J. Shallue, Xavier Dumusque, Andrew Collier Cameron, Christopher Leet, Lars A. Buchhave, Rosario Cosentino, Adri… 展开作者
Zoe L. de Beurs, Andrew Vanderburg, Christopher J. Shallue, Xavier Dumusque, Andrew Collier Cameron, Christopher Leet, Lars A. Buchhave, Rosario Cosentino, Adriano Ghedina, Raphaëlle D. Haywood, Nicholas Langellier, David W. Latham, Mercedes López-Morales, Michel Mayor, Giusi Micela, Timothy W. Milbourne, Annelies Mortier, Emilio Molinari, Francesco Pepe, David F. Phillips, Matteo Pinamonti, Giampaolo Piotto, Ken Rice, Dimitar Sasselov, Alessandro Sozzetti, Stéphane Udry, Christopher A. Watson
备注：
28 pages, 14 figures, Accepted for publication in the Astronomical Journal
Exoplanet detection with precise radial velocity (RV) observations is currently limited by spurious RV signals introduced by stellar activity. We show that machine learning techniques such as linear regression and neural networks can effectively remove the activity signals (due to starspots/faculae) from RV observations. Previous efforts focused on carefully filtering out activity signals in time using modeling techniques like Gaussian Process regression (e.g. Haywood et al. 2014). Instead, we systematically remove activity signals using only changes to the average shape of spectral lines, and no information about when the observations were collected. We trained our machine learning models on both simulated data (generated with the SOAP 2.0 software; Dumusque et al. 2014) and observations of the Sun from the HARPS-N Solar Telescope (Dumusque et al. 2015; Phillips et al. 2016; Collier Cameron et al. 2019). We find that these techniques can predict and remove stellar activity from both simulated data (improving RV scatter from 82 cm/s to 3 cm/s) and from more than 600 real observations taken nearly daily over three years with the HARPS-N Solar Telescope (improving the RV scatter from 1.753 m/s to 1.039 m/s, a factor of ~ 1.7 improvement). In the future, these or similar techniques could remove activity signals from observations of stars outside our solar system and eventually help detect habitable-zone Earth-mass exoplanets around Sun-like stars.
Max Planck Institute for Intelligent Systems（马克斯·普朗克智能系统研究所）；New York University（纽约大学）
We describe a method for removing the effect of confounders in order to reconstruct a latent quantity of interest. The method, referred to as half-sibling regression, is inspired by recent work in causal inference using additive noise models. We provide a theoretical justification and illustrate the potential of the method in a challenging astronomy application.
For years, scientists have used data from NASA's Kepler Space Telescope to look for and discover thousands of transiting exoplanets. In its extended K2 mission, Kepler observed stars in various regions of sky all across the ecliptic plane, and therefore in different galactic environments. Astronomers want to learn how the population of exoplanets are different in these different environments. However, this requires an automatic and unbiased way to identify the exoplanets in these regions and rule out false positive signals that mimic transiting planet signals. We present a method for classifyi…
The discovery of habitable exoplanets has long been a heated topic in astronomy. Traditional methods for exoplanet identification include the wobble method, direct imaging, gravitational microlensing, etc., which not only require a considerable investment of manpower, time, and money, but also are limited by the performance of astronomical telescopes. In this study, we proposed the idea of using machine learning methods to identify exoplanets. We used the Kepler dataset collected by NASA from the Kepler Space Observatory to conduct supervised learning, which predicts the existence of exoplanet…
X-ray observations of star-planet systems are important to grow our understanding of exoplanets; these observation allow for studies of photoevaporation of the exoplanetary atmosphere, and in some cases even estimations of the size of the outer planetary atmosphere. The German-Russian eROSITA instrument onboard the SRG (Spectrum Roentgen Gamma) mission is performing the first all-sky X-ray survey since the 1990s, and provides X-ray fluxes and spectra of exoplanet host stars over a much larger volume than was accessible before. Using new eROSITA data as well as archival data from XMM-Newton, Ch…
The Kepler mission and subsequent ground-based follow-up observations have revealed a number of exoplanet host stars with nearby stellar companions. This study presents speckle observations of 57 Kepler objects of interest (KOIs) that are also double stars, each observed over a 3 to 8 year period, which has allowed us to track their relative motions with high precision. Measuring the position angle and separation of the companion with respect to the primary can help determine if the pair exhibits common proper motion, indicating it is likely to be a bound binary system. We report on the motion…
Limb-darkening is fundamental in determining transit lightcurve shapes, and is typically modeled by a variety of laws that parametrize the intensity profile of the star that is being transited. Confronted with a transit lightcurve, some authors fix the parameters of these laws, the so-called limb-darkening coefficients (LDCs), while others prefer to let them float in the lightcurve fitting procedure. Which of these is the best strategy, however, is still unclear, as well as how and by how much each of these can bias the retrieved transit parameters. In this work we attempt to clarify those poi…
The next generation of ground and space-based telescopes will image habitable planets around nearby stars. A growing literature describes how to characterize such planets with spectroscopy, but less consideration has been given to the usefulness of planet colors. Here, we investigate whether potentially Earth-like exoplanets could be identified using UV-visible-to-NIR wavelength broadband photometry (350-1000 nm). Specifically, we calculate optimal photometric bins for identifying an exo-Earth and distinguishing it from uninhabitable planets including both Solar System objects and model exopla…
Identifying Exoplanets with Deep Learning VI. Enhancing neural network mitigation of stellar activity RV signals with additional metrics
Imperial College London（帝国理工学院）；Massachusetts Institute of Technology（麻省理工学院）；University of Birmingham（伯明翰大学）；Technical University of Denmark（丹麦技术大学）；University of St Andrews（圣安德鲁斯大学）；Fundación Galileo Galilei—INAF（伽利略基金会-INAF）；Observatoire de Genève, Université de Genève（日内瓦天文台，日内瓦大学）
Comparing telluric removal methods in their capability to recover injected exoplanet atmosphere signals with high resolution emission spectroscopy
University of Warwick（华威大学）；Dipartimento di Fisica Università degli Studi di Torino（都灵大学物理系）；INAF-Osservatorio Astrofisico di Torino（意大利国家天体物理研究所都灵天文台）
This paper presents an analysis that searched for systematic effects within the CoRoT exoplanet field light curves. The analysis identified a systematic effect that modified the zero point of most CoRoT exposures as a function of stellar magnitude. We could find this effect only after preparing a set of learning light curves that were relatively free of stellar and instrumental noise. Correcting for this effect, rejecting outliers that appear in almost every exposure, and applying SysRem, reduced the stellar RMS by about 20 %, without attenuating transit signals.
[Abridged] Recently, there have been a series of detections of molecules in the atmospheres of extrasolar planets using high spectral resolution (R~100,000) observations, mostly using the CRyogenic high-resolution InfraRed Echelle Spectrograph (CRIRES) on the Very Large Telescope. These measurements are able to resolve molecular bands into individual absorption lines. Observing many lines simultaneously as their Doppler shift changes with time allows the detection of specific molecules in the atmosphere of the exoplanet. We performed simulations of high-resolution CRIRES observations of a plan…
Massachusetts Institute of Technology（麻省理工学院）；Google（谷歌）；University of Southern Queensland（南昆士兰大学）；Princeton University（普林斯顿大学）
The TESS mission produces a large amount of time series data, only a small fraction of which contain detectable exoplanetary transit signals. Deep learning techniques such as neural networks have proved effective at differentiating promising astrophysical eclipsing candidates from other phenomena such as stellar variability and systematic instrumental effects in an efficient, unbiased and sustainable manner. This paper presents a high quality dataset containing light curves from the Primary Mission and 1st Extended Mission full frame images and periodic signals detected via Box Least Squares (…
NASA's Kepler Space Telescope was designed to determine the frequency of Earth-sized planets orbiting Sun-like stars, but these planets are on the very edge of the mission's detection sensitivity. Accurately determining the occurrence rate of these planets will require automatically and accurately assessing the likelihood that individual candidates are indeed planets, even at low signal-to-noise ratios. We present a method for classifying potential planet signals using deep learning, a class of machine learning algorithms that have recently become state-of-the-art in a wide variety of tasks. W…
The strongest remotely detectable signature of life on our planet today is the photosynthetically produced oxygen (O2) in our atmosphere. However, recent studies of Earth's geochemical proxy record suggest that for all but the last ~500 million years, atmospheric O2 would have been undetectable to a remote observer--and thus a potential false negative for life. During an extended period in Earth's middle history (2.0 - 0.7 billion years ago, Ga), O2 was likely present but in low concentrations, with pO2 estimates of ~0.1 - 1% of present day levels. Although O2 has a weak spectral impact at the…
Identifying deformed exoplanets with JWST and Ariel: A target selection metric for phase curve observations
Leiden University（莱顿大学）；SRON, Space Research Organisation Netherlands（荷兰空间研究组织）；Observatoire astronomique de l’Université de Genève（日内瓦大学天文台）；University of New South Wales（新南威尔士大学）；Universidad de Antioquia（安蒂奥基亚大学）
Given the forthcoming launch of the James Webb Space Telescope (JWST) which will allow observ- ing exoplanet atmospheres with unprecedented signal-over-noise ratio, spectral coverage and spatial resolution, the uncertainties in the atmosphere modelling used to interpret the data need to be as- sessed. As the first step, we compare three independent 1D radiative-convective models: ATMO, Exo-REM and petitCODE. We identify differences in physical and chemical processes taken into ac- count thanks to a benchmark protocol we developed. We study the impact of these differences on the analysis of obs…
In antiquity, all of the enduring celestial bodies that were seen to move relative to the background sky of stars were considered planets. During the Copernican revolution, this definition was altered to objects orbiting around the Sun, removing the Sun and Moon but adding the Earth to the list of known planets. The concept of planet is thus not simply a question of nature, origin, composition, mass or size, but historically a concept related to the motion of one body around the other, in a hierarchical configuration. After discussion within the IAU Commission F2 "Exoplanets and the Solar Syst…
The upcoming launch of the James Webb Space Telescope (JWST) means that we will soon have the capability to characterize the atmospheres of rocky exoplanets. However, it is still unknown whether such planets orbiting close to M dwarf stars can retain their atmospheres, or whether high-energy irradiation from the star will strip the gaseous envelopes from these objects. We present a new method to detect an atmosphere on a synchronously rotating rocky exoplanet around a K/M dwarf, by using thermal emission during secondary eclipse to infer a high dayside albedo that could only be explained by br…
NASA's Transiting Exoplanet Survey Satellite (TESS) presents us with an unprecedented volume of space-based photometric observations that must be analyzed in an efficient and unbiased manner. With at least $\sim1,000,000$ new light curves generated every month from full frame images alone, automated planet candidate identification has become an attractive alternative to human vetting. Here we present a deep learning model capable of performing triage and vetting on TESS candidates. Our model is modified from an existing neural network designed to automatically classify Kepler candidates, and i…
We discuss the feasibility of direct multipixel imaging of exoplanets with the solar gravitational lens (SGL) in the context of a realistic deep space mission. For this, we consider an optical telescope, placed in the image plane that forms in the strong interference region of the SGL. We consider an Earth-like exoplanet located in our immediate stellar neighborhood and model its characteristics using our own Earth. We estimate photon fluxes from such a compact, extended, resolved exoplanet. This light appears in the form of an Einstein ring around the Sun, seen through the solar corona. The s…
NASA Goddard Space Flight Center（美国国家航空航天局戈达德太空飞行中心）；Universities Space Research Association（大学空间研究协会）；The Catholic University of America（美国天主教大学）；Universita di Napoli Federico II（那不勒斯费德里科二世大学）；Science Systems and Applications, Inc.（科学系统与应用公司）；University of Maryland, Baltimore County（马里兰大学巴尔的摩县分校）
The Transiting Exoplanet Survey Satellite (TESS) mission measured light from stars in ~75% of the sky throughout its two year primary mission, resulting in millions of TESS 30-minute cadence light curves to analyze in the search for transiting exoplanets. To search this vast data trove for transit signals, we aim to provide an approach that is both computationally efficient and produces highly performant predictions. This approach minimizes the required human search effort. We present a convolutional neural network, which we train to identify planetary transit signals and dismiss false positiv…
Stellar flares with Ariel: detection prospects and impacts on exoplanet spectroscopy
Konkoly Observatory（孔科利天文台）；HUN-REN Research Centre for Astronomy and Earth Sciences（匈牙利研究网络地球与天文学研究中心）；HUN-REN CSFK MTA Centre of Excellence（匈牙利研究网络 CSFK 马塔科学院卓越中心）；SRON Space Research Organisation Netherlands（荷兰空间研究组织 SRON）
Context. Spectroscopy of exoplanet atmospheres at high resolving powers is rapidly gaining popularity to measure the presence of atomic and molecular species. While this technique is robust against contaminant absorption in the Earth's atmosphere, the non stationary stellar spectrum creates a non-negligible source of noise that can alter or even prevent detection. Aims. We aim at using three-dimensional stellar simulations to directly remove the signature of the star from observations, and prior to cross correlation with templates for the planet's atmosphere Methods. We compute synthetic spect…
Terrestrial planets orbiting M dwarfs may soon be observed with the James Webb Space Telescope (JWST) to characterize their atmospheric composition and search for signs of habitability or life. These planets may undergo significant atmospheric and ocean loss due to the superluminous pre-main-sequence phase of their host stars, which may leave behind abiotically-generated oxygen, a false positive for the detection of life. Determining if ocean loss has occurred will help assess potential habitability and whether or not any O2 detected is biogenic. In the solar system, differences in isotopic ab…
Transit and radial velocity searches are two techniques for identifying nearby extrasolar planets to Earth that transit bright stars. Identifying a robust sample of these exoplanets around bright stars for detailed atmospheric characterization is a major observational undertaking. In this study we describe a framework that answers the question of whether a transit or radial velocity survey is more efficient at finding transiting exoplanets given the same amount of observing time. Within the framework we show that a transit survey's window function can be approximated using the hypergeometric p…
SIMTERFERE: An optical interferometry simulator for quantifying the coherent flux stability of VLTI/GRAVITY+. Reaching per mill stability: Application to exoplanet spectroscopy
Max-Planck-Institut für Astronomie（马克斯·普朗克天文学研究所）；Fakultät für Physik und Astronomie, Universität Heidelberg（海德堡大学物理与天文学学院）；Max-Planck-Institut für extraterrestrische Physik（马克斯·普朗克地外物理学研究所）；I. Physikalisches Institut, Universität zu Köln（科隆大学第一物理研究所）
Exoplanetary atmospheric observations require an exquisite precision in the measurement of the relative flux among wavelengths. In this paper, we aim to provide a new adaptive method to treat light curves before fitting transit parameters in order to minimize systematic effects that affect, for instance, ground-based observations of exo-atmospheres. We propose a neural-network-based method that uses a reference built from the data itself with parameters that are chosen in an unsupervised fashion. To improve the performance of proposed method, K-means clustering and Silhouette criteria are used…
We examine the repeatability, reliability, and accuracy of differential exoplanet eclipse depth measurements made using the InfraRed Array Camera (IRAC) on the Spitzer Space Telescope during the post-cryogenic mission. We have re-analyzed an existing 4.5 μm data set, consisting of 10 observations of the XO-3b system during secondary eclipse, using seven different techniques for removing correlated noise. We find that, on average, for a given technique, the eclipse depth estimate is repeatable from epoch to epoch to within 156 parts per million (ppm). Most techniques derive eclipse depths that …
University of Oxford（牛津大学）；University of Central Florida（中佛罗里达大学）；King’s College London（伦敦国王学院）；Luxembourg Institute of Science and Technology（卢森堡科学技术研究院）；University of Pittsburgh（匹兹堡大学）；NASA Goddard Space Flight Center（美国国家航空航天局戈达德太空飞行中心）；University of Bern（伯尔尼大学）；Blue Marble Space Institute of Science（蓝大理石空间科学研究所）
Machine learning is now used in many areas of astrophysics, from detecting exoplanets in Kepler transit signals to removing telescope systematics. Recent work demonstrated the potential of using machine learning algorithms for atmospheric retrieval by implementing a random forest to perform retrievals in seconds that are consistent with the traditional, computationally-expensive nested-sampling retrieval method. We expand upon their approach by presenting a new machine learning model, \texttt{plan-net}, based on an ensemble of Bayesian neural networks that yields more accurate inferences than …
High-resolution Doppler spectroscopy is a powerful tool for identifying molecular species in the atmospheres of both transiting and non-transiting exoplanets. Currently, such data is analysed using cross-correlation techniques to detect the Doppler shifting signal from the orbiting planet. In this paper we demonstrate that, compared to cross-correlation methods currently used, the technique of Doppler tomography has improved sensitivity in detecting the subtle signatures expected from exoplanet atmospheres. This is partly due to the use of a regularizing statistic, which acts to suppress noise…
Current exoplanet surveys using the radial velocity (RV) technique are targeting M dwarfs because any habitable zone terrestrial-mass planets will induce a high RV and orbit on shorter periods than for more massive stars. One of the main caveats is that M dwarfs show a wide range of activity levels from inactive to very active, which can induce an asymmetry in the line profiles and, consequently, a spurious RV measurement. We aim to benchmark the impact of stellar activity on high-precision RV measurements using regular-cadence CARMENES visible and near-infrared observations of the active M3.5…
We compare limb darkening laws derived from 3D hydrodynamical model atmospheres and 1D hydrostatic MARCS models for the host stars of the two transiting exoplanet systems HD 209458 and HD 189733. The surface brightness distribution of the stellar disks is calculated using 3D LTE spectrum formation and opacity sampling. We test our predictions using least-squares fits of model light curves to primary eclipses that were observed with the Hubble Space Telescope (HST). The limb darkening law derived from the 3D model of HD 209458 between 2900 A and 5700 A produces significantly better fits to the …
Since the start of the Wide Angle Search for Planets (WASP) program, more than 160 transiting exoplanets have been discovered in the WASP data. In the past, possible transit-like events identified by the WASP pipeline have been vetted by human inspection to eliminate false alarms and obvious false positives. The goal of the present paper is to assess the effectiveness of machine learning as a fast, automated, and reliable means of performing the same functions on ground-based wide-field transit-survey data without human intervention. To this end, we have created training and test datasets made…
Spinspotter is a robust and automated algorithm designed to extract stellar rotation periods from large photometric datasets with minimal supervision. Our approach uses the autocorrelation function (ACF) to identify stellar rotation periods up to one-third the observational baseline of the data. Our algorithm also provides a suite of diagnostics that describe the features in the ACF, which allows the user to fine-tune the tolerance with which to accept a period detection. We apply it to approximately 130,000 main-sequence stars observed by the Transiting Exoplanet Survey Satellite (TESS) at 2-…
Spectral retrieval techniques are currently our best tool to interpret the observed exoplanet atmospheric data. Said techniques retrieve the optimal atmospheric components and parameters by identifying the best fit to an observed transmission/emission spectrum. Over the past decade, our understanding of remote worlds in our galaxy has flourished thanks to the use of increasingly sophisticated spectral retrieval techniques and the collective effort of the community working on exoplanet atmospheric models. A new generation of instruments in space and from the ground is expected to deliver higher…
Detecting Atmospheric CO2 Trends as Population-Level Signatures for Long-Term Stable Water Oceans and Biotic Activity on Temperate Terrestrial Exoplanets
Diagnostic potential of the spectral signatures of Raman scattering imprinted in planetary albedo spectra at short optical wavelengths has been demonstrated in research on Solar System planets and has recently been proposed as a probe of exoplanet atmospheres, complementary to albedo studies at longer wavelengths. Spectral features caused by Raman scattering offer insight into the properties of planetary atmospheres, such as the atmospheric depth, composition, and temperature, as well as the possibility of detecting and spectroscopically identifying spectrally inactive species, such as H$_2$ a…
Space-based photometric surveys have discovered large numbers of planets transiting other stars, but these observe in a single band-pass and yield only the planet radius, orbital period, and transit duration. Information on the masses, compositions, and any atmospheres of these planets requires additional observations from the ground or space. The Transiting Exoplanet Survey Satellite (TESS) will yield thousands of planets around bright stars suitable for such follow-up. In the absence of spectroscopy or spectrophotometry from space, observations through the different pass-bands of multiple sp…
