{"id":"8e41e9cc-843d-42ba-9e8d-fe9af3c83d33","arxiv_id":"2502.00553","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Chromospheric and coronal light from the host star changes the NIR transmission spectrum of HAT-P-18 b and raises its retrieved atmospheric temperature from about 536 K to above 700 K.","lead":"This paper adds a model of stellar chromospheres and coronae to exoplanet transit fits and reanalyzes JWST observations of hot Jupiter HAT-P-18 b. It finds that ignoring these hot outer layers biases retrieved atmospheric temperatures and molecular abundances.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed temperature increase from 536 K to 736 K (abstract) is not statistically significant: body reports 720.8+376.4−188.0 K and the 1σ intervals overlap. The central quantitative claim is unsupported.","rationale":"The central claim has two pillars: (i) the TACHELES model detects chromospheric/coronal emission in the light curves, and (ii) correcting for this emission changes the retrieved atmospheric properties, most notably increasing the temperature from 536 K to 736 K. The reader's weakest_assumption targets pillar (i), specifically the exponential emission law. That is a legitimate modeling concern, especially because the SDO validation is at 17.1 nm rather than NIR and the chromospheric lines used (Hα, He I, Ca II) are not all optically thin. However, the more decisive issue is in pillar (ii): the reported uncertainties on the two temperatures overlap substantially, making the headline increase statistically insignificant. This is an internal consistency problem, not a matter of external consensus. It can be settled directly from the published posterior samples. If the temperature shift is not significant, the paper's conclusion that ignoring these layers biases retrieved temperatures is unsupported, even if the model itself is valid. The Bayes factors for 24 bins are suggestive but do not by themselves establish the magnitude of the atmospheric bias. Therefore I focus on the temperature significance as the single most load-bearing concern. My recommendation remains CONDITIONAL, as the paper presents a novel model and valuable data; the reported evidence simply does not yet support the quantitative claim as stated.","tokens_in":14694,"tokens_out":8867,"duration_ms":87032,"concrete_test":"Using the posterior samples from the Zenodo record (https://zenodo.org/records/13361162), compute the posterior distribution of ΔT = Tref(TACHELES, one het.) − Tref(photosphere, one het.). If the 95% credible interval of ΔT includes 0, or if the posterior mass for ΔT > 0 is below 95%, the claimed temperature increase is not statistically significant. Also verify whether the best-fit Tref is 720.8 K, 721 K, or 736 K, and reconcile the abstract/body discrepancy.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's most important quantitative result—that accounting for the chromosphere/corona raises the retrieved atmospheric temperature from 536 K to 736 K, closer to the 852 K equilibrium—is not supported by the reported uncertainties. Table 3 gives the one-heterogeneity POSEIDON retrievals as Tref = 536.0+188.9−100.8 K (photospheric) and Tref = 720.8+376.4−188.0 K (TACHELES). The point-estimate increase is 184.8 K, which is smaller than the lower 1σ error on the TACHELES value (188 K) and comparable to the upper error on the photospheric value. Combining the asymmetric errors in quadrature gives a 1σ uncertainty of roughly 266 K, so the shift is about 0.7σ. The abstract quotes 736 K, while §6 quotes 721 K; neither matches Table 3 exactly. Because the temperature shift is the central claimed consequence and is not statistically significant, the paper's conclusion that ignoring these layers biases atmospheric properties is not established by this dataset. The Bayes factors in §3 show wavelength-dependent preference for TACHELES in some bins, but the atmospheric retrieval does not demonstrate a significant temperature change.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces TACHELES, a transit model that adds an exponentially decaying, optically thin emitting shell to a standard Mandel-Agol photospheric light curve, parameterized by a brightness ratio Br and a scale height H. The model is fit to JWST NIRISS/SOSS observations of HAT-P-18 b; wavelength bins with Bayes factors favoring TACHELES are used to construct a 'chromospheric/coronal spectrum' and to identify emission lines using CHIANTI. The corrected transmission spectrum is then passed to POSEIDON atmospheric retrievals. The authors report that including the stellar outer layers raises the retrieved temperature from 536 K to about 721-736 K, closer to the equilibrium temperature, decreases the retrieved H2O and CO2 abundances, and opens a new window on stellar outer layers.","tokens_in":14993,"tokens_out":8307,"duration_ms":81698,"significance":"If the claims were robust, TACHELES would address an important systematic in transmission spectroscopy: stellar chromospheres and coronae are usually neglected, and for active stars they could bias retrieved temperatures and abundances. The model derivation is transparent, the code and derived data are publicly available, and the synthetic retrieval tests in Appendix B and the solar validation in Appendix A are useful contributions. However, the headline quantitative claim is not supported by the reported uncertainties, and the NIR detection lacks independent validation. The concept is promising and worth publishing after major revision that tempers the claims and adds supporting tests.","major_comments":[{"comment":"The central claim that accounting for the chromosphere/corona increases the retrieved temperature from 536 K to about 736 K is not statistically supported. Table 3 reports Tref = 536.0+188.9-100.8 K for the photospheric one-heterogeneity model and Tref = 720.8+376.4-188.0 K for the TACHELES one-heterogeneity model. The 1-sigma intervals overlap; the point-estimate shift of 184.8 K is smaller than the lower 1-sigma error on the TACHELES value and comparable to the upper error on the photospheric value, corresponding to roughly 0.7 sigma if the asymmetric errors are combined in quadrature. The abstract's value of 736 K also does not match the body's 720.8 K. The conclusion that ignoring the stellar outer layers biases retrieved atmospheric temperatures is therefore not established by this dataset and should be reframed as tentative or suggestive.","section":"§4, Table 3; Abstract"},{"comment":"The exponential, optically thin emission law I0 e^{-r/H} integrated along the line of sight (Eq. 2) is the foundation of the TACHELES detection, but its applicability at 0.85-2.8 um is not independently validated. Appendix A validates the exponential form only with SDO EUV images at 17.1 nm, a very different spectral regime, and the fitted Br and H values are not checked against alternative explanations such as unocculted spots, faculae, or wavelength-dependent systematics. Because the derived chromospheric/coronal spectrum and the line identifications are constructed from these same fitted parameters, the CHIANTI matching in §5 is a consistency check of the fit rather than an independent confirmation. The manuscript needs an explicit test (e.g., injection-recovery with spot/faculae models, or a comparison of the NIR line ratios with a solar NIR spectrum) before claiming a detection.","section":"§2, Eq. (2); Appendix A"},{"comment":"The statistical evidence for TACHELES is evaluated bin-by-bin without addressing the multiplicity of wavelength bins: the paper reports 24 bins with Bayes factor >=3 and 6 with Bayes factor >=10, but it never states the total number of bins, so the false-positive rate cannot be assessed. The same issue affects the line identifications: 9 emission peaks at >=5 sigma are matched to CHIANTI lines, but no estimate is given for the expected number of chance matches given the line density in the CHIANTI list and the NIRISS spectral resolution. A binomial or false-discovery-rate calculation, or a null test on the photospheric residuals, is needed to support the per-bin detections and the line identifications.","section":"§3, §5"},{"comment":"The abstract states that the analysis 'decreases the best-fit abundance of CO2 by almost an order of magnitude,' but §4 describes the CO2 decrease as tentative and based on a singular feature at the edge of the spectrum, and it notes that CO2 is hard to constrain with NIRISS/SOSS data. Similarly, the H2O decrease is only about 2 sigma. The abstract and conclusions should carry the same caveats as the body, otherwise readers will take an unsecured abundance shift as a principal result.","section":"Abstract; §4"}],"minor_comments":[{"comment":"The temperature value in the abstract (736 K) differs from the value quoted in §6 (721 K) and in Table 3 (720.8 K); please harmonize these numbers.","section":"Abstract; §6; Table 3"},{"comment":"'Jeffrey's scale' should be 'Jeffreys scale'.","section":"§3"},{"comment":"The column header 'of f' appears to be a typo for the baseline offset parameter.","section":"Table 4"},{"comment":"The black and red line identifications in the figure are hard to distinguish; a legend or distinct marker styles would improve clarity.","section":"Fig. 6"},{"comment":"The statement that the H-alpha peak is 34 A blueward of line center deserves a quantitative discussion, since such a large offset may undermine the line-identification procedure.","section":"§5"},{"comment":"The azimuthally averaged SDO brightness profile is shown only over a limited radial range (roughly 1-1.2 R*), whereas the fitted scale heights in §3 range from 0.07 to 0.31 R*; please comment on the extrapolation of the solar validation to the fitted regime.","section":"Appendix A"}],"recommendation":"major_revision","confidential_remarks":"The TACHELES model and its public code are worthwhile contributions, and the paper is generally well organized. The main problem is that the abstract and conclusions overstate the temperature result: the 536 K to ~721 K shift is not significant given the quoted 1-sigma uncertainties, and the abstract's 736 K does not match the body. The authors should either strengthen the evidence (for example, by a joint fit of all wavelength bins or additional observations) or substantially weaken the headline claims. The CO2 abundance decrease should also be presented with the same 'tentative' language used in the body. With these revisions, the paper would be a solid contribution on a novel systematic effect."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is TACHELES: a transit model that adds an optically thin, exponentially decaying shell to the usual Mandel and Agol photosphere and applies it to JWST NIRISS/SOSS data. That is a real idea, and the geometry is close to the X-ray model of King et al. (2024), which the paper openly cites as independent parallel work. The SDO validation of the exponential emission law is a nice touch, the code is public, and the paper is transparent about its limitations, including the tentative CO2 result and the lack of UV/X-ray data. For those reasons, the paper deserves a serious referee.\n\nThe problem is the central quantitative claim. The abstract says the retrieved temperature increases from 536 K to 736 K, making it close to the 852 K equilibrium. The body and Table 3 give 720.8 +376.4/-188.0 K for the TACHELES one-heterogeneity retrieval, compared with 536.0 +188.9/-100.8 K for the photospheric fit. Those 1-sigma intervals overlap comfortably; the shift is about 0.7 sigma. So the paper does not demonstrate that ignoring the chromosphere biases the retrieved temperature for this target. The abstract's 736 K is also a different number from the body's 720.8 K, which does not inspire confidence in the reporting.\n\nThere are other soft spots, in proportion. The chromospheric/coronal spectrum used for line identification is constructed from the same fitted Br and H values that produced the transit depths, so matching CHIANTI lines is a consistency check on the fit, not an independent detection. The exponential shell is validated at 17.1 nm with SDO images, but the NIR wavelengths here are a different regime; the paper itself notes that the retrieved scale heights and brightness ratios are not separately validated. The CO2 abundance decrease is explicitly labeled tentative and rests on a single feature. None of these kill the model, but they should be stated more carefully.\n\nThe strongest part remains the model and its public implementation. The retrieval application to HAT-P-18 b is a proof of concept, not a confirmed bias. The paper is for exoplanet atmosphere modelers and stellar activity folks, and it is worth their time. I would not cite it for the temperature-shift claim, but I would cite it as a method. A serious referee should engage with it, and the authors should be pushed to fix the abstract and to be honest that the temperature difference is not currently significant.\n\nBottom line: send to peer review, but with the expectation that the central claims get reined in.","headline":"New transit model (TACHELES) worth knowing, but the headline temperature claim is about 0.7 sigma and the abstract overstates it.","tokens_in":694,"tokens_out":1024,"would_cite":true,"duration_ms":27616,"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":"Stellar chromospheres and coronae bias exoplanet transmission spectra, and including them raises HAT-P-18 b's retrieved temperature from 536 K to about 736 K.","keywords":["exoplanet atmospheres","transmission spectroscopy","stellar chromosphere","stellar corona","HAT-P-18 b","JWST NIRISS/SOSS","transit light curve modeling","stellar activity"],"falsifier":"Take a JWST transmission-spectroscopy target whose coronal and chromospheric emission is independently measured in X-ray and ultraviolet at the same epoch; if TACHELES-fitted brightness ratios and scale heights do not track the independently measured emission strengths and their wavelength pattern, the exponential-shell interpretation is wrong. A cheaper check is to measure the Sun's off-limb brightness at 0.85 to 2.8 microns to test whether the exponential line-of-sight model holds at the near-infrared wavelengths where the paper applies it.","tokens_in":14513,"feed_emoji":"🌞","tokens_out":5752,"duration_ms":51440,"temperature":0.7,"pith_summary":"Transmission spectroscopy measures the starlight filtered through a planet's atmosphere during transit, and standard models assume the star's light comes from a sharp photospheric disk. This paper argues that main-sequence stars also possess optically thin chromospheres and coronae that emit at near-infrared wavelengths, and that these layers make transits shallower and wider in a wavelength-dependent way. Using JWST observations of the hot Jupiter HAT-P-18 b, the authors show that adding such a layer to the transit model changes the derived transmission spectrum enough to raise the best-fit atmospheric temperature from 536 K to about 736 K, close to the predicted equilibrium temperature of 852 K, and to lower the CO2 abundance by nearly an order of magnitude. A sympathetic reader would care because most JWST transmission-spectroscopy targets orbit stars at least as active as HAT-P-18, so this neglected effect may be biasing many published atmospheric retrievals.","feed_headline":"A star's corona shifts exoplanet temperatures by 200 K","feed_subtitle":"Adding the host star's chromosphere to transit models raises HAT-P-18 b's fitted temperature from 536 K to 736 K.","key_machinery":"The carrying object is TACHELES, a transit model that adds an optically thin, spherically symmetric, exponentially decaying emitting shell with brightness profile $I_0 e^{-r/H}$ around a standard photospheric Mandel-Agol limb-darkened disk. The shell brightness is integrated along the line of sight (Eq. 2) to give an on-sky radial profile, and the planet occults both photosphere and shell. Two new free parameters, the brightness ratio $\\mathrm{Br}$ and scale height $H$, capture the wavelength-dependent line emission of the chromosphere and corona; at near-IR wavelengths this emission is strong enough in some bins to alter the transit shape and depth. The model is validated against solar EUV images showing an exponential line-of-sight integrated profile, and its detection limits are characterized with simulated light curves.","core_discovery":"The paper's central claim is that the chromosphere and corona of the host star, not just its photosphere, must be included in transit models before the planetary transmission spectrum is interpreted. For HAT-P-18 b, the authors find that 24 wavelength bins show substantial and 6 show strong statistical evidence for the extended-layer model over a purely photospheric one, with fitted brightness ratios of 0.04 to 0.25 and scale heights of 0.07 to 0.31 stellar radii. Once these layers are accounted for, the best-fit atmospheric temperature rises from 536 K to about 721 to 736 K, much closer to the 852 K equilibrium temperature, and the CO2 mixing ratio drops by almost an order of magnitude. The retrieved chromospheric and coronal spectrum contains nine lines above 5 sigma, all matched to CHIANTI plasma-model lines, including coronal Fe X emission. If this is right, the usual assumption of a purely photospheric transit source is a systematic error source for active host stars.","pith_inferences":["A natural extension, not tested here, is to fold chromospheric and coronal emission into retrieval codes as a standard stellar nuisance parameter alongside spots and faculae; this would let archival JWST datasets be re-fit without new observations.","The paper's line detections suggest that high-SNR transmission spectroscopy could be used to build differential emission measure models of other stars, effectively extending solar coronal physics to exoplanet hosts.","If the exponential shell is confirmed at near-infrared wavelengths on the Sun, the same model could be applied to smaller planets or grazing transits, where the finite planet size does not smear out the chromospheric geometry, to map the layer structure."],"forward_implications":["Atmospheric temperatures retrieved from transmission spectra of active-host exoplanets are biased low when chromospheric and coronal emission is ignored; the effect can be hundreds of kelvin.","For HAT-P-18 b, including the outer stellar layers raises the best-fit temperature to near the equilibrium temperature and reduces the CO2 abundance by almost an order of magnitude.","Wavelength-dependent residuals attributed to spots or third-light dilution may partly be chromospheric and coronal line emission, so published spectra of active stars may need reanalysis.","TACHELES turns transmission spectroscopy into a probe of stellar chromospheres and coronae, with detected lines such as Fe X, Mg II, and He I in a K dwarf other than the Sun.","Since most current and scheduled JWST targets orbit stars more active than HAT-P-18, the bias should be larger, not smaller, for the wider sample."],"supporting_citations":[{"why":"Supplies the reduced JWST NIRISS/SOSS light curves of HAT-P-18 b and the spot-influenced transmission spectrum and retrieval setup that this paper reanalyzes.","marker":"Fournier-Tondreau et al., 2024"},{"why":"The photospheric transit model that TACHELES extends with the chromospheric and coronal shell.","marker":"Mandel & Agol 2002"},{"why":"Source of the exponential decay description of chromospheric emission used in Eq. 2.","marker":"Linsky 1980"},{"why":"Establishes that chromospheric and coronal line emission is wavelength-dependent, motivating the wavelength-resolved analysis.","marker":"Linsky 2019"},{"why":"Independent recent X-ray transit model with the same optically thin shell geometry, used as methodological comparison.","marker":"King et al. 2024"},{"why":"The POSEIDON retrieval framework used to compare photospheric and TACHELES-derived transmission spectra.","marker":"MacDonald & Madhusudhan 2017"},{"why":"Provides PHOENIX stellar models used for limb-darkening priors and for converting TACHELES brightness ratios into an emission spectrum.","marker":"Husser 2013"},{"why":"Supplies the log R'HK activity value for HAT-P-18 that places the host in the moderately active regime.","marker":"Claudi et al. 2024"},{"why":"Source of the HAT-P-18 b orbital period and system parameters fixed in the light-curve fits.","marker":"Hartman et al. 2011"}],"fun_headline_variants":["Corona fix lifts exoplanet temperature to 736 K","Star's chromosphere shifts exoplanet spectra by 200 K","Corona cuts CO2 10x in hot Jupiter atmosphere","Transit spectra need star's corona, not just photosphere","Including star's corona warms hot Jupiter to 736 K"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on the assumption that the chromosphere and corona of HAT-P-18 can be described as a single optically thin exponential shell whose brightness falls off as $e^{-r/H}$ at near-infrared wavelengths, an assumption validated only by solar EUV images at 17.1 nm and not by any near-infrared measurement of the target star.","fun_headline_variants_meta":{"raw":{"variants":["Corona fix lifts exoplanet temperature to 736 K","Star's chromosphere shifts exoplanet spectra by 200 K","Corona cuts CO2 10x in hot Jupiter atmosphere","Transit spectra need star's corona, not just photosphere","Including star's corona warms hot Jupiter to 736 K"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000855,"raw_usage":{"total_tokens":3718,"prompt_tokens":951,"completion_tokens":2767,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":567,"completion_tokens_details":{"reasoning_tokens":2679}},"tokens_in":567,"tokens_out":2767,"duration_ms":18754,"temperature":1.0,"reasoning_tokens":2679,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T18:33:39.484662+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a JWST transmission-spectroscopy target whose coronal and chromospheric emission is independently measured in X-ray and ultraviolet at the same epoch; if TACHELES-fitted brightness ratios and scale heights do not track the independently measured emission strengths and their wavelength pattern, the exponential-shell interpretation is wrong. A cheaper check is to measure the Sun's off-limb brightness at 0.85 to 2.8 microns to test whether the exponential line-of-sight model holds at the near-infrared wavelengths where the paper applies it.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Source of the exponential decay description of chromospheric emission used in Eq. 2."},{"cited_title":"2019, Host Stars and their Effects on Exoplanet Atmospheres : An Introductory Overview, 1st edn., Lecture Notes in Physics, 955 (Cham: Springer International Publishing)","cited_arxiv_id":null,"evidence_quote":"Establishes that chromospheric and coronal line emission is wavelength-dependent, motivating the wavelength-resolved analysis."},{"cited_title":"Generating X-ray transit profiles with batman","cited_arxiv_id":"2410.01559","evidence_quote":"Independent recent X-ray transit model with the same optically thin shell geometry, used as methodological comparison."},{"cited_title":"2024, , 682, A136, 10.1051/0004-6361/202347079","cited_arxiv_id":null,"evidence_quote":"Supplies the log R'HK activity value for HAT-P-18 that places the host in the moderately active regime."}],"review_version":1}