pith. sign in

arxiv: 2605.28635 · v1 · pith:N2YEOKBAnew · submitted 2026-05-27 · 🌌 astro-ph.EP · astro-ph.IM

Comparing telluric removal methods in their capability to recover injected exoplanet atmosphere signals with high resolution emission spectroscopy

Marcelo Aron Fetzner Keniger , Matteo Brogi , David Armstrong , Siddharth Gandhi This is my paper

Pith reviewed 2026-06-29 09:46 UTC · model grok-4.3

classification 🌌 astro-ph.EP astro-ph.IM
keywords telluric removalexoplanet atmosphereshigh-resolution spectroscopyPCAinjection recoveryτ Bootis bCARMENESwater detection
0
0 comments X

The pith

PCA can produce higher SNR in exoplanet spectra but degrades the planetary signal more than fitting-based telluric removal.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests how well three methods remove Earth's atmospheric lines from ground-based spectra to reveal exoplanet atmosphere signals. Using real CARMENES data of τ Bootis b with added fake water signals, they show that principal component analysis sometimes boosts the detection statistic but removes more of the true signal than do explicit modeling approaches such as Molecfit or Astroclimes. The tests also reveal that signals from slower-orbiting planets are harder to recover and can appear shifted in velocity space, an effect strongest with PCA. Re-analysis of the same planet yields no water detection with any method.

Core claim

Through systematic injection and recovery of H2O signals into near-infrared observations of the dayside of τ Bootis b, the study demonstrates that PCA detrending achieves higher signal-to-noise ratios at the cost of greater planetary signal attenuation compared to model-based methods. All three techniques—PCA, Molecfit, and Astroclimes—exhibit increased difficulty recovering signals at lower orbital velocities, distorting both amplitude and velocity location, with the distortion most pronounced under PCA. No method recovers a previously reported water feature from the planet.

What carries the argument

Injection-recovery tests that insert synthetic exoplanet emission spectra into real observations before applying each detrending method and then measure the recovered cross-correlation signal in velocity space.

If this is right

  • Characterizations of exoplanet atmospheres may carry systematic biases in velocity and abundance if PCA is used without correction.
  • Model-based telluric removal is less likely to erase real signals but may require more accurate line lists.
  • Claims of atmospheric detections should be re-checked with multiple detrending approaches.
  • Low-velocity planets require extra caution in analysis pipelines.
  • Non-detections in re-analyses can arise from the choice of detrending rather than absence of the signal.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Hybrid detrending that combines PCA for initial cleaning with model fitting for final steps might reduce the trade-off.
  • Velocity biases could be corrected post-facto by adjusting the template grid after detrending.
  • Similar tests on other instruments or wavelengths would test whether the PCA degradation is universal.
  • The results suggest that some literature non-detections of atmospheres may be method-dependent.

Load-bearing premise

The synthetic injected signals interact with the observed spectra and the detrending algorithms in exactly the same way that genuine exoplanet signals would.

What would settle it

A high-SNR detection of an exoplanet atmosphere signal recovered at the expected velocity and amplitude after PCA detrending, matching independent model-based results on the same data.

Figures

Figures reproduced from arXiv: 2605.28635 by David Armstrong, Marcelo Aron Fetzner Keniger, Matteo Brogi, Siddharth Gandhi.

Figure 1
Figure 1. Figure 1: From top to bottom, evolution of the airmass, humidity, precipitable water vapour (PWV) and measured instrumental SNR for the observations of 𝜏 Bootis taken on the nights of March 26th 2018 (night 1), May 11th 2018 (night 2), March 12th 2019 (night 3), March 15th 2019 (night 4) and April 11th 2019 (night 5). The crosses correspond to the points that were removed from the analysis based on the airmass and S… view at source ↗
Figure 2
Figure 2. Figure 2: Top: All (good) observations of 𝜏 Bootis from the night of March 26th 2018, with their respective Astroclimes best-fit telluric models and the PHOENIX spectrum. Bottom: residuals obtained by subtracting the combined best-fit telluric plus PHOENIX models from the observational spectra. (3) is accompanied by a +0.5 term to account for the 180◦ symmetry between the planet and star orbits. After being shifted … view at source ↗
Figure 3
Figure 3. Figure 3: Removal process of telluric and stellar lines for Astroclimes for CARMENES NIR order 12. Each row on the 𝑦-axis corresponds to an individual observation, and the colour bars on the right side correspond to the flux level of each panel. The white vertical lines correspond to the points that were masked out due to being too deep or NaNs. (a) is the observational spectra, (b) is the observational spectra divi… view at source ↗
Figure 4
Figure 4. Figure 4: Cross-correlation map in phase-RV space obtained from the residu￾als when detrending with Astroclimes (a), Molecfit (b) and PCA with 𝑁𝐶 = 3 (c) for an injected planetary signal 10x stronger than the original one. These results are from the injection and recovery test analysis for night 1 with 𝐾p = 110 km/s. The planetary signal corresponds to the slanted line on the left side of each panel. The discontinui… view at source ↗
Figure 5
Figure 5. Figure 5: Left: Velocity map obtained from the residuals when detrending with Astroclimes (a), Molecfit (b) and PCA with 𝑁𝐶 = 3 for an injected planetary signal that is 10x stronger than the original one. These results are from the injection and recovery test analysis for night 1 with 𝐾p = 110 km/s. The values were converted to SNR as explained in the text. The black dashed lines, defined by visual inspection as 𝑦 =… view at source ↗
Figure 6
Figure 6. Figure 6: SNR calculated from the velocity map (a), scaling factor 𝛿 (b), 𝑉sys (c) and 𝐾p (d) as a function of the injected signal strength multiplier for Astroclimes, Molecfit and PCA with 2, 3, 4, 5, 6 and 10 components. These results are for night 1 with 𝐾p = 110 km/s. The horizontal dashed line on panel (a) corresponds to the SNR detection threshold (SNR = 4). For panels (b)-(d), only the results from the MCMC r… view at source ↗
Figure 7
Figure 7. Figure 7: Same as [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Same as [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Same as [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: (a) Injected planetary signal which has a strength of 10x the original signal. (b) Median subtracted residuals difference when removing telluric and stellar lines using Astroclimes with and without the injected signal. (c) Same as (b), but for Molecfit. (d) Same as (b), but for PCA with 𝑁𝐶 = 3. Panel (e) shows how panels (a), (b), (c) and (d) look for an individual exposure taken at phase 0.569. The wavel… view at source ↗
Figure 11
Figure 11. Figure 11: Combined SNR for nights 1-5 from the residuals when detrending with Astroclimes (a) and PCA (b). The optimal number of PCA components for each night was 3, 14, 5, 5 and 5, respectively, as determined by the framework from Cheverall et al. (2023). The dashed rectangle delimits the “signal” and “noise” regions used to calculate the SNR, and its boundaries are 𝐾p = 80, 140 km/s and 𝑉rest = −30, 30 km/s. The … view at source ↗
read the original abstract

For ground-based high-resolution spectroscopic exoplanet atmosphere studies, removing the contamination from telluric and stellar lines is a crucial step in the analysis process. Despite that, there is no consensus in the literature on the most appropriate way to carry this out. Typically, tellurics are either directly modelled to a percent-level precision, or blindly detrended via Principal Component Analysis (PCA) algorithms, the latter particularly common at infrared wavelengths. Here, we compare three different detrending methods, PCA, Molecfit and our own fitting algorithm Astroclimes, measuring their performance in the context of removing telluric and stellar lines to detect exoplanetary atmospheric signals. We specifically look for H2O detections, which are particularly affected by residual, time-correlated variability of the telluric spectrum. We use near-infrared CARMENES observations of the day-side of $\tau$ Bootis b to carry out injection and recovery tests. We find that while PCA can sometimes achieve higher SNR, it comes at the expense of stronger signal degradation. All methods struggle more for injected signals with lower orbital velocities, not just affecting the signal's magnitude but its location in velocity space as well. This behaviour is more prominent for PCA than for Astroclimes and Molecfit. These results highlight the importance of understanding the effects of different detrending methods on exoplanetary signals, which can lead to biases when characterising real detections. Finally, we report that our attempts to detect a previously claimed water signal from $\tau$ Bootis b all resulted in non-detections.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The paper compares three telluric and stellar line removal methods (PCA, Molecfit, and Astroclimes) for high-resolution near-IR emission spectroscopy using CARMENES observations of τ Bootis b. Through injection-and-recovery experiments of H2O signals, it reports that PCA can yield higher SNR at the cost of greater signal degradation, that all methods exhibit worse performance (in both amplitude and velocity location) at lower orbital velocities with the effect strongest for PCA, and that attempts to recover a previously claimed water signal result in non-detections.

Significance. If the injection tests are representative of real signal–telluric–detrending interactions, the results provide a useful empirical demonstration that common detrending choices can bias both detection significance and retrieved orbital parameters, which is relevant for the growing sample of high-resolution exoplanet atmosphere studies.

major comments (1)
  1. [Abstract / Methods (injection tests)] The central performance rankings and non-detection conclusion rest entirely on injection-recovery tests (abstract and associated methods/results sections). The manuscript does not report additional validation (e.g., comparison against synthetic datasets with known extra systematics such as airmass-dependent telluric depth variations or residual stellar-line correlations) that would confirm the injected signals experience the same linear and non-linear interactions as genuine planetary emission; this assumption is load-bearing for the claim that PCA degrades signals more severely and for the velocity-shift results.
minor comments (2)
  1. [Abstract] The abstract states that 'all methods struggle more for injected signals with lower orbital velocities' but does not quantify the velocity bias (e.g., mean offset in km/s or dependence on K_p); adding a table or figure panel with these metrics would strengthen the presentation.
  2. [Methods] Notation for the three methods is introduced without a dedicated methods subsection summarizing their algorithmic differences (e.g., number of PCA components, fitting parameters in Astroclimes); a short comparison table would improve clarity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and constructive feedback on our manuscript. We respond to the major comment below.

read point-by-point responses

  1. Referee: [Abstract / Methods (injection tests)] The central performance rankings and non-detection conclusion rest entirely on injection-recovery tests (abstract and associated methods/results sections). The manuscript does not report additional validation (e.g., comparison against synthetic datasets with known extra systematics such as airmass-dependent telluric depth variations or residual stellar-line correlations) that would confirm the injected signals experience the same linear and non-linear interactions as genuine planetary emission; this assumption is load-bearing for the claim that PCA degrades signals more severely and for the velocity-shift results.

    Authors: We thank the referee for raising this point. Our injection-recovery experiments inject synthetic planetary emission signals directly into the real CARMENES spectra prior to detrending. This design ensures the injected signals encounter the actual time-varying telluric absorption, airmass trends, and any residual stellar-line correlations present in the observations. We maintain that this is the most direct test of how each method interacts with genuine data properties, consistent with standard practice in high-resolution exoplanet spectroscopy. That said, we acknowledge the referee's concern that additional cross-checks against controlled synthetic datasets could further test the robustness of the observed trends. We will therefore revise the Methods and Discussion sections to expand on the assumptions of the injection approach, explicitly note its limitations relative to purely synthetic tests, and compare our methodology to similar studies in the literature. revision: partial

Circularity Check

0 steps flagged

No circularity: purely empirical method comparison via injection tests

full rationale

The paper contains no mathematical derivation chain, no fitted parameters renamed as predictions, and no self-citation load-bearing steps. All claims (SNR comparisons, velocity shifts, non-detections) are direct outputs of the described injection-and-recovery experiments on CARMENES data. The study is self-contained against external benchmarks with no reduction of results to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on the domain assumption that injection-recovery tests provide a fair proxy for real-signal recovery and that the three methods were applied without unstated differences in tuning.

axioms (1)
  • domain assumption Injected signals accurately represent real exoplanet atmospheric signals without introducing additional biases or interactions with the detrending algorithms
    Performance is evaluated solely via recovery of these injected signals (abstract section on injection and recovery tests).

pith-pipeline@v0.9.1-grok · 5832 in / 1283 out tokens · 30711 ms · 2026-06-29T09:46:20.911651+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

3 extracted references · 1 canonical work pages

  1. [1]

    J., et al., 2019, A&A, 621, A74 Bailey J., Simpson A., Crisp D., 2007, PASP, 119, 228 Basilicata M., et al., 2024, A&A, 686, A127 Bello-Arufe A., Buchhave L

    Ahrer E.-M., et al., 2023, Nature, 614, 653 Alderson L., et al., 2023, Nature, 614, 664 Allart R., Lovis C., Pino L., Wyttenbach A., Ehrenreich D., Pepe F., 2017, A&A, 606, A144 Allart R., et al., 2019, A&A, 623, A58 Alonso-Floriano F. J., et al., 2019, A&A, 621, A74 Bailey J., Simpson A., Crisp D., 2007, PASP, 119, 228 Basilicata M., et al., 2024, A&A, 6...

    work page doi:10.1117/12.551480 2023

  2. [2]

    Each row on the𝑦-axis corresponds to an individual observation, represented by the orbital phase, and the colour maps on the right side correspond to the flux level of each panel. For panels (a) and (b), the colour bar limits are the minimum and maximum values of the quantity being plotted, whereas for panels (c)-(h) the limits were taken to be the median...

    2026

  3. [3]

    signal” and “noise

    Each row on the𝑦-axis corresponds to an individual observation, represented by the orbital phase, and the colour maps on the right side correspond to the flux level of each panel. The white vertical lines correspond to the points that were masked out due to being too deep or NaNs. (a) is the observationalspectra,(b)istheobservationalspectradivideditsmedia...

    2026