{"id":"350d04df-8604-438f-b0eb-3f3bdeaee46b","arxiv_id":"2602.22830","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Simulated ELT/ANDES HRCCS observations predict H2O, H2S, and CO detections in V1298 Tau b within about 10 hours and C/O differentiation for TOI-451 c in about 17-18 hours, but only when out-of-transit exposures are included and detrending is applied to templates.","lead":"This paper uses simulated ELT/ANDES observations to test whether high-resolution spectroscopy can detect molecules in the atmospheres of two young, long-period sub-Neptunes, V1298 Tau b and TOI-451 c. It finds that including out-of-transit exposures and reprocessing the model templates through the detrending pipeline is essential, and predicts about 10 hours of observing time for molecule detections in V1298 Tau b.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unmodeled stellar activity of V1298 Tau could break the 10-hour detection budget; the static-PHOENIX host-star assumption is the most load-bearing weakness.","rationale":"The central claim is an injection-recovery forecast, and the paper's internal logic is coherent: the methodological message that template reprocessing and out-of-transit exposures improve recovery is supported by their comparisons. The most plausible failure mode is not pipeline self-consistency but the realism of the host-star input. The authors themselves flagged the PHOENIX approximation in Section 3.1.1, so this is an explicit limitation, not a hidden error. A rotating, spotted stellar model with the same injection, plus a no-planet control, would settle whether the ~10 h budgets and KP recovery survive. This concern does not disprove the paper; it strengthens the conditional framing already chosen by the reader. I see no basis to reject or accept outright, so the reader's CONDITIONAL verdict should stand. The reader's weakest assumption (static host star) is the same one I identify; the ANDES/CARMENES layout approximation is less threatening because the authors argue, reasonably, that finer ANDES coverage should only improve sensitivity. The abstract discrepancy on the TOI-451 c solar case is real and should be corrected, but it is secondary to the stellar-activity question.","tokens_in":27357,"tokens_out":6655,"duration_ms":75255,"concrete_test":"Rerun the V1298 Tau b Nights 2+3 simulations with a time-variable stellar spectrum: PHOENIX plus a two-temperature spot/facula map rotating with P_rot=2.91 d, with spot filling factor and photometric amplitude matched to V1298 Tau's observed variability, injecting the same planetary transmission model. Run the Section 2.4 SVD+MLR pipeline with the same k and masks. If H2O/H2S/CO significances drop below 4sigma, or the recovered KP shifts outside the injected 2sigma contour—or if a null-injection control (spots, no planet) produces a >4sigma artifact at the expected KP—the stated detection budgets do not transfer to the real star.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The V1298 Tau b forecast requires the host star to be spectroscopically inert during the simulated nights, but Section 3.1.1 concedes that a single PHOENIX spectrum 'might not be the best fit' for this pre-main-sequence star. V1298 Tau is a <30 Myr magnetically active K-type T-Tauri star with P_rot=2.91 d (Table 1). Its spots, faculae, and rotation generate time-dependent stellar line-profile variations that are absent from the simulated flux cuboids. The atmospheric transmission features in Fig. 4 are only ~100-200 ppm above the continuum; stellar activity can produce comparable or larger line-profile residuals, especially during spot-crossing events in the 6.42 h transit. Time-domain SVD (Section 2.4.2) removes only the strongest common-mode or slowly varying components, so activity may either be absorbed into the SVD modes (eroding the planetary signal) or survive as Doppler-shifting residuals that the CCF-to-log(L) step could misattribute. Template reprocessing propagates only the SVD modes learned from a static-star dataset, so it cannot repair this missing variability. The reported >4sigma H2O/H2S/CO budgets and KP recovery are therefore conditional on a spectroscopically inert host star; if V1298 Tau exhibits activity at typical T-Tauri levels, the central forecast may not transfer. A secondary inconsistency: the front abstract's 'sub-solar and solar from super-solar' C/O statement is contradicted by Fig. 11, where the solar case remains non-differentiable even with four nights.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents simulated ELT/ANDES high-resolution cross-correlation spectroscopy (HRCCS) observations of two young, long-period sub-Neptunes, V1298 Tau b and TOI-451 c, using the Ratri observation simulator and the Upamana SVD+MLR detrending and CCF-to-likelihood analysis framework. Atmospheric structures and disequilibrium chemistry are computed with petitCODE/VULCAN, and high-resolution transmission spectra are generated with petitRADTRANS and injected into synthetic YJH-band nights. The central methodological claim is that including out-of-transit exposures improves detectability only if the detrending operation is also applied to the template models before cross-correlation; without this reprocessing, recovered orbital parameters can be biased or the signal can be lost. For V1298 Tau b, the authors report possible >4–5σ detections of H2O, H2S, and CO in ~10 hours (two nights) in a cloud-free scenario, with H2O alone surviving in the cloudy case. For TOI-451 c, they find that one 4.5-hour night can detect sub-solar and super-solar C/O scenarios, but that differentiating these cases requires roughly 18 hours (four nights), while the solar C/O case remains non-differentiable. The paper frames these as feasibility forecasts, not empirical detections.","tokens_in":27798,"tokens_out":6555,"duration_ms":66531,"significance":"If the forecast holds, the paper gives concrete guidance for ELT/ANDES observing programs: out-of-transit exposures and template reprocessing are necessary for long-period sub-Neptunes, and a few nights of YJH spectroscopy could measure molecular abundances in objects that are otherwise accessible only to JWST. The pipeline is internally consistent, the comparison between reprocessed and non-reprocessed templates is clean, and the use of HST/Spitzer/JWST-informed models for V1298 Tau b is a strength. The result is nevertheless a self-recovery injection test: the injected signals are generated with the same forward models used as templates, so the quoted significances are sensitivity forecasts conditional on the atmospheric and stellar assumptions. The most serious limitations are the static PHOENIX host star for an active T-Tauri target, the ad hoc SVD truncation, and the un-simulated extrapolation from one night to four/five nights for the C/O differentiation claim. These issues make the headline numbers less robust than the text suggests.","major_comments":[{"comment":"The V1298 Tau b detection budgets assume a spectroscopically inert host star. Ratri uses a single PHOENIX spectrum, and §3.1.1 concedes that “this host star is a pre-main sequence star, so the PHOENIX spectrum might not be the best fit.” V1298 Tau is a <30 Myr active T-Tauri star with rotation period 2.91 d (Table 1). Spots, faculae, and rotational modulation produce time-dependent stellar line-profile variability that is absent from the simulated flux cuboids, while the planetary transmission features in Fig. 4 are only ~100–200 ppm. The time-domain SVD detrending (§2.4.2) can either absorb such variability into the removed modes (eroding the planetary signal) or leave Doppler-shifting residuals that the v_rest–K_P search could misattribute. The >4–5σ budgets and recovered K_P values are therefore conditional on a spectroscopically inert host; a spot/faculae simulation or activity-scali","section":"§3.1.1, Fig. 4, Table 1"},{"comment":"The abstract states that “distinguishing sub-solar and solar from super-solar C/O requires ~17 hours,” but the paper’s own model-selection test does not support inclusion of the solar case. Fig. 11 (right) shows that the solar C/O model remains non-differentiable even after scaling to four/five nights, and §3.2.3 explicitly says “the solar case cannot be differentiated at all.” The Conclusions correctly restrict the 18-hour statement to sub-solar versus super-solar. The abstract therefore overstates the result. In addition, the four/five-night projection is obtained by scaling one night’s log(L) matrix (§3.2.3: “scaled up the log(L) matrix by the number of nights”), not by simulating independent nights; this assumes identical noise, airmass, telluric, and barycentric realizations for each night. Simulated multi-night co-adds are needed before the 17/18-hour claim can be considered robust","section":"§3.2.3, Fig. 11, abstract, Conclusions"},{"comment":"The SVD truncation rank k is a free parameter, and the choice is explicitly ad hoc: k = 4 for V1298 Tau b is said to be “not based on any injection-retrieval studies,” and k = 8 is used for TOI-451 c with no stated justification. The paper reports partial robustness for V1298 Tau b (“detections can be found ... until at least k = 10”) but does not scan k for TOI-451 c. Since the entire study is about how SVD detrending affects signal recovery, the significance levels in Figs. 7–11 are conditional on this choice. A k-sensitivity analysis or an objective selection criterion should be provided, otherwise the reported detection significances and the comparison across nights/scenarios are not fully trustworthy.","section":"§2.4.2, §3.1.2, §3.2.3"},{"comment":"The ANDES simulations are performed using the CARMENES CCD layout as a proxy: 28 spectral orders, 4096 pixels per order, a CARMENES read-out time of 34 s, and H-band coverage only to 1.7 µm instead of the ANDES 1.8 µm. The paper states that this makes the results conservative, but that claim is asserted rather than demonstrated. ANDES is expected to have roughly twice the number of orders, and denser order coverage could change the CCF line statistics and the SVD detrending behavior. Given that the title and abstract make quantitative claims about ELT/ANDES, the instrument approximation is load-bearing for the quoted significances. At least a targeted discussion, or a mock-order sensitivity test, is required to justify the “conservative” terminology.","section":"§2.3"}],"minor_comments":[{"comment":"The abstract in the full text states “>5σ” detections for V1298 Tau b, while the arXiv abstract supplied with the manuscript states “>4σ.” The number should be standardized, and the final significance for CO (described in §3.1.2 as ~4–5σ) should be reported consistently.","section":"Abstract"},{"comment":"There is a numerical inconsistency: the text says “co-adding 5 similar nights,” the Fig. 11 caption says “4 nights,” and the Conclusions say “at least 18 hours i.e. 4 nights.” If Night 1 is 4.5 hours, five nights would be 22.5 hours, not 18 hours. Please align the text, caption, and conclusions.","section":"Fig. 11, §3.2.3, Conclusions"},{"comment":"The treatment of aerosols for V1298 Tau b is presented inconsistently. §2.1 states that a cloud-free atmosphere is supported by HST/WFC3 showing “no detectable aerosol opacity,” while §2.2 introduces a gray cloud deck at 0.01 bar “indicated from JWST observations in Barat et al. (2025).” It should be clarified whether the cloudy case is observationally motivated or is an idealized test scenario.","section":"§2.1 and §2.2"},{"comment":"Minor typographical issues: “metallciity” should be “metallicity”; “SCO” in Fig. 3 and §2.2 should be “OCS” (carbonyl sulfide); “read-out noise (ROD)” should be “read-out noise (RON).” Also, the reference to the ANDES ETC in the caption of Fig. A.1 uses Palle et al. 2025b while the text cites Palle et al. 2025a; please check which publication contains the ETC description.","section":"§2.1, §2.2, §2.3"}],"recommendation":"major_revision","confidential_remarks":"The paper makes a useful and timely forecast, but the abstract and conclusions currently overstate the robustness of the central numbers. In particular, the 17/18-hour C/O-differentiation claim depends on a scaling of a single night's log-likelihood and is not supported by the authors' own Fig. 11 for the solar case. The static-host assumption for V1298 Tau is also a real physical risk given the target's known activity. These issues can be addressed with additional simulations or by reframing the claims, so I see this as a major-revision rather than a reject."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a careful injection-recovery study, not an empirical detection; treat the hour budgets as sensitivity forecasts conditional on the forward models. Second, the abstract's 'sub-solar and solar from super-solar' C/O statement is not supported by their own Figure 11, where the solar case never differentiates even after four nights.\n\nWhat's genuinely new: hour-budget forecasts for V1298 Tau b and TOI-451 c with ELT/ANDES, and a clean quantitative demonstration that out-of-transit exposures help only when the detrending (SVD+MLR) is propagated to the template. Without reprocessing, orbital parameters drift and significances drop. That is a useful, practical message for observers. The atmospheric modeling chain (petitCODE+VULCAN+petitRADTRANS) is standard but sensibly tied to JWST constraints for V1298 Tau b, and the comparative chemistry between the two planets is a nice touch.\n\nSoft spots.\n\n- Unmodeled stellar activity. V1298 Tau is a young active T-Tauri star, and the simulation uses a single static PHOENIX spectrum. The paper admits this in Section 3.1.1. Spots and rotation can produce line-profile residuals comparable to the ~100-200 ppm planetary signal, and SVD detrending may either absorb or mimic the signal. This is the strongest reason to doubt the 10-hour budget transferring to real data. It is not fatal to the study's logic, but it should be front and center.\n\n- C/O differentiation overstatement. The text says 'distinguishing sub-solar and solar from super-solar' but the solar case is not differentiable. The conclusion more carefully says only sub-solar vs super-solar. The abstract should be corrected.\n\n- ANDES approximation. Using the CARMENES CCD order layout and 4096 pixels is a practical choice, and the authors note ANDES has finer coverage, making their predictions conservative. That caveat is probably right, but no code or data are released, so the reproducibility claim has to be taken on trust.\n\n- Self-recovery circularity. Injection-recovery uses the same forward models for both injection and template, so the numbers measure pipeline sensitivity, not reality. That is standard in observability forecasts, but it means the C/O differentiation statement is about the model grid, not the planets.\n\nWho this is for: people planning ELT/ANDES HRCCS programs, especially on young, low-mass, long-period planets. They will actually use these hour budgets. The paper deserves peer review; with moderate revision it's publishable. I would not desk-reject.","headline":"Solid injection-recovery forecast for ELT/ANDES on two young sub-Neptunes, but the abstract overstates the C/O claim and the static host-star assumption should worry anyone planning time.","tokens_in":28257,"tokens_out":2552,"would_cite":true,"duration_ms":26145,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This simulation study forecasts that ELT/ANDES high-resolution spectroscopy can detect H2O, H2S, and CO in the young sub-Neptune V1298 Tau b in under ten hours, and can separate sub-solar from super-solar C/O in TOI-451 c in about seventeen","keywords":["exoplanet atmospheres","high-resolution cross-correlation spectroscopy","ELT/ANDES","sub-Neptune progenitors","V1298 Tau b","TOI-451 c","transmission spectroscopy","detrending algorithms"],"falsifier":"Take one real night of high-resolution spectra of an active young star like V1298 Tau, inject a synthetic planetary transmission signal with a known rest velocity and Kp, and run the same SVD detrending-and-reprocessing pipeline with and without out-of-transit exposures: if the injected signal is not recovered at more than 4-sigma with correct orbital parameters whenever out-of-transit exposures are present, the central forecast fails. A cheaper pre-ANDES version would use archival CARMENES or ESPRESSO observations of a young active star with an injected Doppler-shifting template.","tokens_in":27289,"feed_emoji":"🔭","tokens_out":7369,"duration_ms":68412,"temperature":0.7,"pith_summary":"The paper tries to establish that high-resolution cross-correlation spectroscopy with the future ELT/ANDES instrument can reach young, long-period sub-Neptunes, not just hot Jupiters. Using fully simulated nights of YJH-band observations and injection-recovery tests, it claims that for V1298 Tau b, water, hydrogen sulfide, and carbon monoxide would be detected at more than 4-sigma in less than about ten hours (two nights), and that for TOI-451 c, roughly four nights would separate sub-solar and super-solar carbon-to-oxygen ratios from each other. The methodological claim is that including out-of-transit exposures and propagating the detrending operation onto the template spectrum before cross-correlation is necessary: without that step, orbital parameters shift or the signal disappears. A sympathetic reader would care because these are formative-stage planets whose atmospheres still record their formation and mass-loss history, and because the result defines the observing strategy needed to characterize them from the ground.","feed_headline":"Ten hours on ELT/ANDES can reveal a young sub-Neptune's atmosphere","feed_subtitle":"Simulations show water, hydrogen sulfide, and CO become detectable if observers also take out-of-transit exposures.","key_machinery":"The mechanism that carries the argument is the SVD+MLR detrending step inside the paper's pipeline, together with a reprocessing step: the saved detrending matrix is applied to the Doppler-shifted model template before cross-correlation, so the template suffers the same line attenuation and distortion as the injected signal. In long-period planets the planetary Doppler drift is slow, so the signal is easily absorbed into the first few singular vectors; out-of-transit exposures produce artifacts that, once reproduced in the template, actually increase the likelihood contrast at the correct orbital velocity. The observation simulator supplies the synthetic ANDES data, using PHOENIX stellar spe","core_discovery":"On its own terms, the paper's central claim is a sensitivity forecast: if the injected atmospheric models are right, the combination of ELT/ANDES and an SVD-based detrending pipeline will recover molecular signals in young long-period sub-Neptunes. For V1298 Tau b, whose model is anchored to HST, Spitzer, and JWST spectra, the forecast is more than 4-sigma detections of H2O, H2S, and CO within at most ten hours (two nights) in the cloud-free case, dropping to H2O alone when a 0.01-bar cloud deck is added. For TOI-451 c, a single 4.5-hour night makes sub-solar and super-solar C/O models detectable but not distinguishable; roughly 17-18 hours (four nights) are needed to tell them apart, while","pith_inferences":["My inference: if stellar activity on V1298 Tau produces Doppler-shifting line distortions, the predicted two-night budget may be optimistic; a testable extension is to repeat the injection-recovery with a time-variable stellar spectrum instead of a static PHOENIX model.","My inference: the same out-of-transit and reprocessing logic should apply to emission-phase observations during occultation, so the result could generalize to non-transiting long-period planets observed in thermal emission.","My inference: the difficulty with the solar C/O case hints that cross-correlation detection significance alone is not a reliable proxy for chemical abundance constraints; a full Bayesian retrieval on the same simulated nights might separate the scenarios earlier or reveal degeneracies.","My inference: because the paper uses a CARMENES CCD layout with coarser wavelength coverage, the true ANDES detections could be stronger than forecast; a direct test is to rerun the pipeline once the ANDES detector design and order format are finalized."],"forward_implications":["Ground-based high-resolution spectroscopy can plausibly be extended from hot Jupiters to young sub-Neptunes on 10-20 day orbits once ELT/ANDES is online.","Observing proposals for long-period transiting planets should budget for out-of-transit exposures; a transit-only night can erase the signal completely.","Any future HRCCS analysis that uses SVD/SYSREM-style detrending should reprocess model templates through the detrending matrix, or reported orbital parameters and detection significances will be misleading.","Cloud decks change the detectable molecule set: with a 0.01-bar cloud deck, H2O remains detectable in V1298 Tau b but H2S and CO weaken below detection in the same time budget.","For TOI-451 c, sub-solar and super-solar C/O become distinguishable after about four nights, but a solar C/O atmosphere is not cleanly differentiated by this method alone."],"fun_headline_variants":["ELT/ANDES can spot water and CO on young sub-Neptune in 10 hrs","Out-of-transit data vital for ELT/ANDES detecting young atmospheres","Young sub-Neptune atmospheres: ELT/ANDES simulations show detections","ELT/ANDES forecasts detect H2O, H2S, CO on V1298 Tau b in ~10h","Distinguishing C/O on TOI-451 c requires ~17 hours with ELT/ANDES"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The forecast rests on treating the host star's spectrum as static and smooth and the telluric absorption as time-independent; V1298 Tau is actually a magnetically active pre-main-sequence star, so unmodeled starspots and rotation could shift stellar lines in ways that either mask the planetary signal or mimic it, and the paper itself notes the PHOENIX spectrum 'might not be the best fit.'","fun_headline_variants_meta":{"raw":{"variants":["ELT/ANDES can spot water and CO on young sub-Neptune in 10 hrs","Out-of-transit data vital for ELT/ANDES detecting young atmospheres","Young sub-Neptune atmospheres: ELT/ANDES simulations show detections","ELT/ANDES forecasts detect H2O, H2S, CO on V1298 Tau b in ~10h","Distinguishing C/O on TOI-451 c requires ~17 hours with ELT/ANDES"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001478,"raw_usage":{"total_tokens":5884,"prompt_tokens":959,"completion_tokens":4925,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":703,"completion_tokens_details":{"reasoning_tokens":4799}},"tokens_in":703,"tokens_out":4925,"duration_ms":32489,"temperature":1.0,"reasoning_tokens":4799,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T20:34:23.566930+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take one real night of high-resolution spectra of an active young star like V1298 Tau, inject a synthetic planetary transmission signal with a known rest velocity and Kp, and run the same SVD detrending-and-reprocessing pipeline with and without out-of-transit exposures: if the injected signal is not recovered at more than 4-sigma with correct orbital parameters whenever out-of-transit exposures are present, the central forecast fails. A cheaper pre-ANDES version would use archival CARMENES or ESPRESSO observations of a young active star with an injected Doppler-shifting template.","supporting_citations":[],"review_version":1}