{"id":"d196e107-cb48-41f1-baef-7ca4e1817346","arxiv_id":"1908.10695","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Simulated retrievals show CRIRES+ can recover hot Jupiter temperature profiles from about 1 bar down to 1e-4 or 1e-6 bar, but molecular mixing ratios need combined wavelength regions and S/N above 10.","lead":"This paper simulates what the CRIRES+ spectrograph at the Very Large Telescope will be able to learn about hot Jupiter atmospheres from high-resolution infrared light. It finds that temperature profiles can be recovered from emission spectra at low signal-to-noise, while molecular abundances require two wavelength regions and stronger signals.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'assumption-free' temperature retrieval is built on an untested constant-VMR assumption; the synthetic validation shares the assumption, so composition gradients could bias the retrieved T-P profile beyond the quoted errors.","rationale":"The paper is a transparent simulation study; it explicitly states the constant-VMR assumption and the absence of telluric and instrumental effects. The reader's verdict of CONDITIONAL is appropriate. I focused on the constant-VMR assumption because it is the one place where the paper's own language overreaches: 'assumption-free temperature profiles' is not literally true, and the untested interaction with composition gradients is a concrete, internally testable risk. The same-forward-model issue is real but harder to settle without external data; the constant-VMR test can be run with the paper's own code and would either confirm or falsify the temperature-retrieval claim in the idealized case. The reader's weakest_assumption overlaps with mine (same forward model, same physical assumptions) but did not specifically flag the constant-VMR assumption, hence partial agreement. I do not see a reason to change the verdict: the feasibility claim is plausible, but it should be conditional on running the proposed sensitivity test along with the reader's requested independent-model checks and telluric-instrument systematic estimates.","tokens_in":31508,"tokens_out":4902,"duration_ms":52678,"concrete_test":"Generate synthetic emission spectra with tauREx using a vertically varying H2O/CO profile (e.g., a smooth factor-of-10 gradient between 10^-4 and 1 bar, consistent with thermochemical equilibrium), then run the paper's OE retrieval under its exact setup (1.6 and 2.3 micron bands, S/N=10, R=100,000, lcorr=1.5, constant VMR). If the retrieved T-P profile departs from the true profile by more than the quoted ~90 K in the 1 bar to 10^-4 bar range, the central claim fails even before telluric and instrumental effects are considered. A stronger variant is to generate the mock data with an independent forward model (e.g., petitRADTRANS or different line lists) to simultaneously test opacity errors.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that high-resolution CRIRES+ spectra allow 'assumption-free' retrieval of the vertical temperature profile (Abstract, Sections 4.1 and 6) rests on a hidden assumption: the retrieval fixes molecular mixing ratios to be constant with altitude (Section 2.2: 'we assumed constant mixing ratios of molecular species'). The synthetic observations are generated with the same constant-VMR assumption, so the validation cannot detect biases caused by real composition gradients (e.g., thermochemical layering of CH4/CO, photochemical H2O variations). The paper itself shows a strong temperature-abundance degeneracy at low S/N (Figure 4), and this degeneracy is depth-dependent once VMRs vary with pressure. A vertically varying abundance can therefore be absorbed into a biased T-P profile while still fitting the data, and the quoted 70-120 K errors (from the OE posterior covariance, Equation 3) do not include this systematic. The claim of assumption-free temperature sounding is thus overstated unless sensitivity to non-constant composition is demonstrated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a simulation-based feasibility study for retrieving the thermal structure and molecular abundances of hot-Jupiter atmospheres from high-resolution near-infrared emission spectroscopy with CRIRES+. The authors generate synthetic day-side spectra of HD 189733b with the tauREx forward model at R=100,000 and R=50,000 in five spectral windows, add Gaussian noise at S/N = 5-50, and invert them with an optimal-estimation retrieval that fits temperatures at 50 pressure levels, four constant volume mixing ratios, and a continuum scaling per band. They validate their retrieval code against HST/Spitzer photometry of HD 189733b, reproduce the Lee et al. (2012) temperature and water abundance, and then report that temperature can be recovered to roughly 70-120 K between about 1 bar and 10^-4 bar (and up to 10^-6 bar with the 2.3 + 4.9 micron combination) while molecular abundances require S/N>10 and two bands. They close with observational strategies and an exposure-time target list for CRIRES+.","tokens_in":31750,"tokens_out":7364,"duration_ms":71534,"significance":"If the quantitative claims survive more realistic error injection, this is a useful and timely feasibility study: it maps which CRIRES+ settings are most informative, shows that two-band combinations help break abundance degeneracies, and provides concrete integration-time estimates for a small target list. Strengths of the paper include the validation against real HST/Spitzer photometry (Fig. 1), the explicit use of averaging kernels to define the sounded pressure range, the robustness checks over multiple initial guesses (Fig. 2), and the candid statement of what is not simulated. The main limitation is that the headline accuracy numbers are derived from a closed self-consistency loop: the same tauREx forward model, with the same constant-VMR assumption, generates and inverts the spectra, so the quoted errors are formal precisions under an idealized, telluric-free model rather than demonstrated accuracies for real CRIRES+ data. The paper's independent photometric validation is an important partial counterweight, but it operates at low spectral resolution and does not exercise the R=100,000 opacities and line shapes that the high-resolution claims depend on.","major_comments":[{"comment":"The paper's central claim of 'assumption-free' retrieval of the vertical temperature profile is not supported as stated, because the forward model assumes constant mixing ratios of all molecular species with altitude (§2.2: 'we assumed constant mixing ratios of molecular species') and the synthetic observed spectra are generated with that same tauREx forward model. The retrieval therefore cannot be surprised by a vertically varying abundance, and any real composition gradient (e.g., thermochemical CO/CH4 layering, photochemical H2O variations) can be absorbed into a biased T-P profile while still fitting the data; this systematic is absent from both the recovery test and from the posterior covariance in Eq. (3). I request a test in which the input spectra are generated with a pressure-dependent VMR (e.g., a chemical-equilibrium profile) and retrieved with the constant-VMR model, quantifying the resulting T-P bias against the quoted 70-120 K errors.","section":"§2.2, §4.1, Abstract"},{"comment":"The claim that ground-based CRIRES+ observations can probe temperatures up to 10^-6 bar rests on the (2.28-2.38)+(4.80-5.00) micron combination, yet the 4.8-5.0 micron window is exactly the region the authors describe as having 'many more telluric lines' with noticeably worse old-CRIRES performance (§5), and all simulated observations omit telluric absorption, flat-fielding, order merging, and wavelength calibration (§2.2 and §6). As the paper itself states, the impact of weak telluric lines on retrievals 'should be tested with real observations.' Until such effects are added or the claims are relabeled as idealized no-telluric limits, the abstract's 10^-6 bar statement overstates what is demonstrated for the actual instrument.","section":"§4.2, §5, §6"},{"comment":"The abstract states that 'the temperature can already be derived accurately with the lowest value that we considered (S/N=5),' but the paper's own Fig. 4 shows that for the 2.28-2.38 micron region at S/N=5, an initial VMR guess 0.5 dex above the truth drives the retrieval to a different temperature solution with the same normalized cost function phi=0.73. The text acknowledges that for S/N<=10 'the final values may deviate from the true ones (but are still within the 3 sigma error bars) depending on the initial guess.' The headline S/N=5 temperature-accuracy claim therefore does not hold for all allowed starting points, and the quoted errors from Eq. (3) do not include this initialization-dependent scatter. Please report the spread of retrieved temperature profiles over the full grid of initial guesses at each S/N, or restrict the accuracy claim to S/N>=10.","section":"§4.1, Fig. 4, Abstract"}],"minor_comments":[{"comment":"In the first paragraph, 'Photometric observations are the more efficient than spectroscopic observations' should read 'are more efficient than'.","section":"§1"},{"comment":"The text 'HD 198733 b' appears to be a typo for HD 189733 b.","section":"§3"},{"comment":"The text refers to 'the EO method' in one place; this should be 'OE method' for consistency.","section":"§4.1"},{"comment":"The caption states 'randomly color-coded for clarity?' with a stray question mark, and the color coding of the red crosses versus red dashed lines is not fully explained in the caption.","section":"Fig. 3 caption"},{"comment":"The column labeled 'd' with unit 'a.u.' is described as 'distance to the parent star' but is actually the orbital semi-major axis; please rename the column and footnote accordingly.","section":"Table 1"},{"comment":"The sentence 'a simultaneous observations in two separate infrared regions ... helps to obtain' has a subject-verb agreement error; it should be 'simultaneous observations ... help'.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The paper is within A&A scope and I see no reason to doubt the authors' good faith. My main editorial concern is the mismatch between the abstract's strong wording ('assumption-free', 'accurate temperature distribution' at S/N=5, probing to 10^-6 bar) and the paper's own caveats in Sections 2.2 and 6. The requested additions (non-constant-VMR test, telluric/systematics sensitivity test, initialization-scatter reporting) are feasible extensions rather than fatal flaws, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a solid and useful feasibility study: it gives quantitative expectations for what CRIRES+ can retrieve from hot Jupiter emission spectra, and its main recommendations — two bands at 1.6 and 2.3 microns, S/N>10 for mixing ratios, S/N=5 for temperature — are concrete and testable. Second, the headline claim of 'assumption-free' temperature retrieval is overstated, because the model assumes constant-with-altitude molecular mixing ratios, and the synthetic observations share that assumption. The stress-test note lands.\n\nThe genuinely new piece is the CRIRES+-specific, instrument-matched retrieval study with a free temperature at each of 50 layers, multiple band combinations, S/N levels, and exposure-time estimates. Earlier work by Lee et al. (2012) and Brogi & Line (2019) is related, but this paper goes further on observational strategy. The validation against real HST/Spitzer photometry in Fig. 1 is a real check, and the multiple-initial-guess tests show the OE solution is stable in the cases considered.\n\nThe circularity is real: tauREx generates the synthetic observations and tauREx retrieves them, so recovering the input truth confirms internal consistency, not accuracy against reality. The authors admit this, and the photometric validation helps, but the quoted 70–120 K errors are conditional on the forward model being right. The constant-VMR assumption is more serious because it is easy to miss: Section 2.2 says it in one sentence, while 'assumption-free' appears in the abstract and introduction. Real hot Jupiters will have vertical composition gradients from thermochemistry and photochemistry, and those can be absorbed into a biased T-P profile. Figure 4 already shows a temperature-abundance degeneracy at S/N=5; that degeneracy becomes depth-dependent when VMRs vary with pressure. So the pressure-range claim should be read as 'under constant composition.' Telluric and instrumental systematics are also absent; the authors say this and note weak telluric lines could matter at the 1e-3 level, but they do not quantify it.\n\nMinor: the correlation length lcorr=1.5 and the a priori temperature uncertainty of 200 K are hand-chosen to keep retrievals smooth, and results could depend on them. Also the 4.8–5.0 micron region is mostly unhelpful except for high-altitude pressure; that is a nuance, not a flaw.\n\nThis paper is for people planning ground-based high-resolution observations and for retrieval code developers. It deserves a serious referee. I would send it out, but require the authors to either qualify or rename the assumption-free claim, and ideally add a test with an independent forward model or with non-constant VMRs. If that is done, this will be a useful reference.","headline":"A useful, honest feasibility study of CRIRES+ retrievals, but the 'assumption-free temperature' claim is overstated because the model assumes constant molecular mixing ratios and validates synthetic data with its own forward model.","tokens_in":32267,"tokens_out":2284,"would_cite":true,"duration_ms":24768,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Ground-based high-resolution near-infrared spectra can recover a hot Jupiter's vertical temperature profile from 1 bar down to 10^-6 bar, and two infrared bands can pin down molecular abundances.","keywords":["exoplanetary atmospheres","hot Jupiters","high-resolution spectroscopy","atmospheric retrieval","optimal estimation","temperature profile","molecular abundances","near-infrared spectroscopy"],"falsifier":"Add a 0.5% telluric absorption residual and a 1% continuum-normalization error to the simulated spectra and re-run the retrieval; if the recovered temperature profile shifts by more than the quoted 70–120 K at any pressure between 1 bar and $10^{-4}$ bar, the paper's feasibility claim is falsified for real data.","tokens_in":31302,"feed_emoji":"🪐","tokens_out":10926,"duration_ms":94828,"temperature":0.7,"pith_summary":"This paper establishes, by simulation, that ground-based high-resolution near-infrared spectroscopy—the kind the CRIRES+ instrument will deliver—can move exoplanet atmosphere studies from detecting molecules to measuring structure. Using the hot Jupiter HD 189733b as a test case, the authors generate synthetic emission spectra, add realistic noise, and run an optimal-estimation retrieval that treats the temperature in each of 50 atmospheric layers as a free parameter. They find that the vertical temperature distribution can be recovered from about 1 bar down to $10^{-4}$ bar with a single spectral band, and down to $10^{-6}$ bar when certain bands are combined, with temperature errors of 70–120 K depending on signal-to-noise. Molecular mixing ratios are harder: accurate values for H2O, CO, and CO2 require S/N above 10 and simultaneous observations in two bands, ideally near 1.6 and 2.3 micron. If correct, this means existing and near-future ground-based spectrographs can characterize hot-Jupiter atmospheres in detail, not just detect their molecules.","feed_headline":"Ground-based IR spectra can map a hot Jupiter's temperature","feed_subtitle":"Simulations show CRIRES+ recovers the vertical temperature profile to 10^-6 bar and molecular abundances with two near-IR bands.","key_machinery":"The machinery is an optimal-estimation retrieval (the standard Bayesian maximum-a-posteriori method of atmospheric remote sensing) wrapped around a line-by-line radiative-transfer forward model. The forward model computes the emitted spectrum from a 1D, cloud-free, LTE atmosphere using high-resolution molecular opacities; the retrieval adjusts 50 layer temperatures, four molecular mixing ratios, and continuum scaling factors to fit the simulated spectrum, with a correlation length of 1.5 pressure scale heights smoothing the temperature profile. Averaging kernels tell which pressure levels each spectral region actually constrains, and it is this sensitivity map that lets the authors claim a wide retrievable pressure range. The key enabler is the large simultaneous wavelength coverage of CRIRES+, which puts thousands of molecular lines of different strengths into a single spectrum, so different line depths probe different atmospheric depths.","core_discovery":"The central claim is that assumption-free retrieval of the vertical temperature structure of a hot Jupiter is feasible from high-resolution (R = 100,000) near-infrared emission spectra: the temperature in each atmospheric layer is a free parameter, with no parameterized T-P profile imposed. In simulated CRIRES+ observations of HD 189733b, the retrieved temperatures match the true profile to within about 120 K at S/N=5 and 70–90 K at higher S/N, over pressures from 1 bar to $10^{-4}$ bar (single band) and up to $10^{-6}$ bar with the right band combination. The authors also find that accurate molecular number densities for H2O, CO, and CO2 require S/N > 10 and at least two spectral regions, with the 1.50–1.70 micron and 2.28–2.38 micron combination preferred; CH4 cannot be constrained at HD 189733b's assumed abundance but becomes retrievable in cooler or carbon-rich atmospheres where it is more abundant.","pith_inferences":["The paper does not explore it, but its S/N thresholds imply a concrete survey design: rank known and future hot Jupiters by host-star brightness and planet-to-star contrast, and allocate two-band (1.6 and 2.3 micron) observations only to targets that can reach S/N=10 in the available nights.","Because the simulated and retrieved spectra share the same forward model, the 70–120 K temperature errors are lower bounds; a mismatched-parameter test (e.g., retrieving with an older opacity table or a cloudy atmosphere) would quantify how much real-systematic error adds.","The paper's constant-mixing-ratio assumption means the retrieved abundances are vertical averages; extending the method to retrieve altitude-dependent mixing ratios, which the optimal-estimation framework can in principle handle, would be the natural next step and would test chemistry models.","If real CRIRES+ data on 51 Peg b reproduce the retrieved profile shape, the same two-band strategy could be applied to cooler or carbon-rich hot Jupiters, where the paper predicts methane becomes accessible; those targets would be a direct test of equilibrium-chemistry predictions."],"forward_implications":["With a single CRIRES+ setting at 2.28–2.38 micron, even S/N=5 is enough to recover the temperature profile down to about $10^{-4}$ bar, though molecular mixing ratios at that noise level can be biased unless the initial guess is close.","Observing two bands, 1.50–1.70 and 2.28–2.38 micron, at S/N=10 constrains the mixing ratios of CO and CO2 about twice as tightly as either band alone and brings temperature errors down to 70–90 K.","Lowering the spectral resolution from R=100,000 to R=50,000 degrades abundance retrievals but barely affects temperature retrievals, so observers can trade resolution for S/N and integration time.","Only two known hot Jupiters, 51 Peg b and tau Boo b, can reach the required S/N with affordable integration times from the southern hemisphere, with 51 Peg b needing about 48 hours for the preferred two-band S/N=10 configuration.","Methane becomes retrievable in cooler planets or carbon-rich (C/O>1) atmospheres, where its mixing ratio is orders of magnitude higher than in HD 189733b."],"supporting_citations":[{"why":"Supplies the HD 189733b temperature-pressure profile and molecular mixing ratios used both as the truth for simulated spectra and as the a priori for validation retrievals.","marker":"Lee et al. (2012)"},{"why":"Describes the tauREx forward model that the authors modified to compute simulated high-resolution emission spectra.","marker":"Waldmann et al. (2015b)"},{"why":"Provides the earlier tauREx retrieval of HD 189733b whose differing temperature profile motivates the discussion of opacity tables.","marker":"Waldmann et al. (2015a)"},{"why":"Introduces the cross-correlation-to-log-likelihood retrieval method that the paper compares with optimal estimation as the main alternative.","marker":"Brogi & Line (2019)"},{"why":"Source of the optimal-estimation implementation details, including the 1.5-scale-height correlation length used to smooth the retrieved temperature profile.","marker":"Irwin et al. (2008)"},{"why":"Defines the optimal-estimation formalism, the cost function, and the averaging-kernel diagnostics that the retrievals rely on.","marker":"Rodgers (2000)"},{"why":"Describes the CRIRES+ instrument upgrade, its near-infrared bandpass, and its resolving power, which set the simulated observational setup.","marker":"Dorn et al. (2014)"},{"why":"Supplies the ExoMol molecular line lists that tauREx uses for the opacities of atmospheric molecules.","marker":"Tennyson & Yurchenko (2012)"},{"why":"Supplies the HITEMP line lists used for the molecular opacities in the forward-model cross sections.","marker":"Rothman et al. (2010)"}],"fun_headline_variants":["High-res IR spectra reveal hot Jupiter's atmospheric layers","Two infrared bands needed for exoplanet molecule mapping","CRIRES+ can map exoplanet temperatures to 10^-6 bar","Hot Jupiter temperature profiles recovered from ground-based IR","Molecule detection in exoplanets requires high S/N and two bands"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The feasibility result rests on the assumption that real CRIRES+ spectra of a hot Jupiter differ from the model only by added Gaussian noise: same molecular opacities, same one-dimensional cloud-free LTE atmosphere, and no telluric residuals, instrumental artifacts, or data-reduction errors; if any of these fail, the quoted 70–120 K temperature errors are lower bounds and the retrieved abundances could be biased.","fun_headline_variants_meta":{"raw":{"variants":["High-res IR spectra reveal hot Jupiter's atmospheric layers","Two infrared bands needed for exoplanet molecule mapping","CRIRES+ can map exoplanet temperatures to 10^-6 bar","Hot Jupiter temperature profiles recovered from ground-based IR","Molecule detection in exoplanets requires high S/N and two bands"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000316,"raw_usage":{"total_tokens":1843,"prompt_tokens":1055,"completion_tokens":788,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":671,"completion_tokens_details":{"reasoning_tokens":703}},"tokens_in":671,"tokens_out":788,"duration_ms":7128,"temperature":1.0,"reasoning_tokens":703,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:09:24.023574+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Add a 0.5% telluric absorption residual and a 1% continuum-normalization error to the simulated spectra and re-run the retrieval; if the recovered temperature profile shifts by more than the quoted 70–120 K at any pressure between 1 bar and $10^{-4}$ bar, the paper's feasibility claim is falsified for real data.","supporting_citations":[{"cited_title":"N., & Irwin , P","cited_arxiv_id":null,"evidence_quote":"Supplies the HD 189733b temperature-pressure profile and molecular mixing ratios used both as the truth for simulated spectra and as the a priori for validation retrievals."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Source of the optimal-estimation implementation details, including the 1.5-scale-height correlation length used to smooth the retrieved temperature profile."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the optimal-estimation formalism, the cost function, and the averaging-kernel diagnostics that the retrievals rely on."},{"cited_title":"& Yurchenko , S","cited_arxiv_id":null,"evidence_quote":"Supplies the ExoMol molecular line lists that tauREx uses for the opacities of atmospheric molecules."}],"review_version":1}