Uncovering Hidden Systematics in Extreme-Precision Radial Velocity Measurements SciPy library ( 36 ) , NumPy ( 23 ; 35 ) , Astropy ( 1 ; 28 )

Lily L. Zhao, Debra A. Fischer, Andrew E. Szymkowiak, John M. Brewer, Joe Llama · 2026

Read Uncovering Hidden Systematics in Extreme-Precision Radial Velocity Measurements SciPy library ( 36 ) , NumPy ( 23 ; 35 ) , Astropy ( 1 ; 28 ) by Lily L. Zhao, Debra A. Fischer, Andrew E. Szymkowiak, John M. Brewer, Joe Llama (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-02296.txt.

Introduction

Lily L. Zhao and four co authors hunt for hidden instrument drift in seven years of data from EXPRES, a spectrograph built to measure star motions with extreme precision. Small drifts can look exactly like the wobble caused by an orbiting planet. The worst drift appeared around January 2022, when the instrument temperature varied more than usual, and it reached a swing large enough to hide or fake a planet signal.

To correct it, they track shifts, stretching, and rotation in the recorded light patterns, plus line shape measures from a laser frequency comb. Combining those traces with instrument readings in one regression cuts the solar drift by about two thirds and reduces scatter for twelve quiet stars. A spurious planet signal around the star Rho Coronae Borealis disappears once the correction is applied.

Catalog record

CREATORLily L. Zhao, Debra A. Fischer, Andrew E. Szymkowiak, John M. Brewer, Joe Llama
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-02296

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Uncovering Hidden Systematics in Extreme-Precision Radial Velocity Measurements SciPy library ( 36 ) , NumPy ( 23 ; 35 ) , Astropy ( 1 ; 28 ) Lily L. Zhao, Debra A. Fischer, Andrew E. Szymkowiak, John M. Brewer, Joe Llama Abstract We identify and correct for small but coherent instrumental drifts in seven years of radial velocity data from the EXtreme PREcision Spectrograph ( EXPRES ). The systematics are most notable for the six months before and after 2022 January, when EXPRES experienced larger temperature variations, and we see a systematic trough-to-peak amplitude of 2.8 m s -1 in the radial velocities. This is large enough to mimic or obscure planetary signatures. To isolate and correct these effects, we develop a suite of diagnostics that track two-dimensional échellogram shifts, scalings, and rotation, as well as line bisector spans (LBS) derived from laser frequency comb (LFC) lines. By combining these empirical tracers with instrument telemetry in a multi-dimensional regression, we reduce the EXPRES instrument trend traced with solar RVs from an RMS of 1.32 m s -1 to 0.43 m s -1 , a 67% improvement, and the aggregate of twelve chromospherically quiet stars show a 26% reduction in velocity scatter. Our injection–recovery simulations further demonstrate a doubling in sensitivity to low-amplitude planetary signals after correction. When applied to the stellar time series of ρ \rho Coronae Borealis ( ρ \rho CrB), the correction removes a spurious planet d signal, restoring the integrity of the data. These results highlight the need for long-term monitoring and multi-dimensional calibration diagnostics on the path toward true centimeter-per-second precision in next-generation EPRV instruments. I Introduction Planet-hunting spectrographs are now enabling sub-meter-per-second radial velocity (RV) measurements ( 25 ; 6 ; 24 ; 33 ; 16 ; 34 ; 10 ) . This level of precision allows for the detection of smaller-amplitude signals and lower-mass planets ( 8 ; 11 ; 2 , e.g.) . The latest generation of instruments are all variants of ultra-stabilized échelle spectrographs that yield very high-fidelity spectra. Échelle spectrographs disperse light in two perpendicular directions, forming a two-dimensional projection of the spectrum on the detector known as an échellogram . The resulting horizontal “bands,” appear like the rungs of a ladder; indeed, échelle is French for “ladder.” The horizontal bands are referred to as échelle orders. The bands are distinct from one another, and so the échelle order number and pixel position across the order is often used to uniquely define a position within the échellogram . The precision achieved by a spectrograph depends not only on its hardware but also on the extraction pipeline, which converts the two-dimensional échellogram into a one-dimensional spectrum. This process involves tracing each order across the detector and extracting its corresponding flux profile. A central step in this process is wavelength calibration, the mapping between detector pixels and physical wavelengths. For radial-velocity (RV) planet searches, where precision at the 10 cm s -1 level is required, the fidelity of this calibration is paramount. Emission-line calibration sources give rise to reference wavelengths that can be fitted by low-order polynomials. More recently, non-parametric models have been enabled by laser frequency combs and other dense calibrators ( 29 ; 39 , e.g.,) . Current approaches implicitly assume that calibration sources perfectly trace all instrumental variations and that these variations affect emission and absorption lines in the same way. In reality, any wavelength-dependent deviations from these assumptions remain uncorrected and can propagate directly as scatter in stellar RV measurements. Moreover, asymmetric instrument profiles (IPs) can bias the measured positions of calibration lines. For example, recent work with ESPRESSO calibration data showed that explicitly modeling IP asymmetries led to a fivefold improvement in wavelength calibration ( 32 ) . Such uncorrected effects contribute to instrument drift , a shift in the spectrograph’s RV zero point relative to a reference calibration. Historically represented as a single scalar offset, instrumental drift is now understood to vary across the detector, motivating two-dimensional drift maps ( 7 ) . These maps reveal spatial structure in residual calibration errors that can degrade long-term RV stability. In this work, we identify a coherent instrumental trend in long-term EXPRES observations that is not corrected by the standard pipeline calibration. The trend became apparent only after several years of data accumulation and careful temporal binning. In § II we introduce the EXPRES data set; in § III we describe binning choices and diagnostic tracers of systematics; in § IV we develop a simple correction model; and in § V we present the corrected results. We conclude with discussion and implications in § VI and summarize in § VII . II Data We make use of data taken by EXPRES , a stabilized, fiber-fed échelle spectrograph ( 16 ) commissioned on the 4.3 m Lowell Discovery Telescope ( 17 , LDT ;) near Flagstaff, Arizona, in 2019 January. As part of the 100 Earths Survey ( 4 ) , EXPRES has acquired observations spanning seven years at the time of this writing. When considering calibration sources alone, EXPRES exhibits an instrumental stability of 4–7 cm s -1 ( 3 ) , while on-sky observations achieve a demonstrated precision of 25-30 cm s -1 for SNR > 200 ( 26 ) . The spectrograph delivers a median resolving power of R\approx 137{,}000 , aided by a rectangular optical fiber that approximates the behavior of a slit. During the initial alignment of EXPRES , care was taken to align the projection of this pseudo-slit with the columns of the detector. For wavelength calibration, EXPRES utilizes a Menlo Systems laser frequency comb (LFC) ( 37 ; 21 ; 29 ; 30 ; 20 ) , which produces evenly spaced lines across the \sim 490–730 nm range. LFC exposures are obtained every 20–45 minutes during both nighttime and daytime operations, with three consecutive LFC frames included in the beginning- and end-of-night calibration sequences. Because of EXPRES ’s rectangular fiber, individual LFC lines are modeled using a super-Gaussian function, I(x)=A\exp{\!\left[-\left(\frac{(x-\mu)^{2}}{2\sigma^{2}}\right)^{p}\right]}, (1) where \mu represents the measured line center. Environmental monitoring is provided by an array of commercial platinum resistive temperature detectors (RTDs). A Lakeshore sensor records the temperature of the Invar optical bench within the vacuum enclosure, while an Omega sensor measures the exterior temperature of the vacuum vessel. A third Omega sensor is mounted on the concrete slab that supports EXPRES , which is seismically isolated from the surrounding telescope structure. These will hereafter be referred to as the optical-bench, vacuum-enclosure, and slab temperatures, respectively. Each sensor records measurements approximately every 2 minutes and 24 seconds with milli-kelvin precision. Hardware issues temporarily halted the reading of temperature measurements for the beginning of 2025. II.1 Solar Data EXPRES has also been collecting disk-integrated solar observations (i.e., Sun-as-a-star observations) during the daytime since 2020 November. With solar data, it is possible to remove the known Doppler shifts from orbiting Solar System planets, leaving only instrumental systematics and RV variations due to solar surface variations. Light from a dedicated solar telescope is fed into the front-end module of EXPRES , after which the light proceeds through the same optical path as nighttime observations ( 19 ) . The diagnostic power of solar observations has motivated

Read the complete text