{"id":"9ab928b7-a6ff-4ab5-a828-c56a82ee6764","arxiv_id":"2607.06491","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":7,"one_line_summary":"A quench-pressure parameterisation with two free parameters recovers unbiased C/O and [M/H] from synthetic JWST spectra with vertical mixing, and tentatively detects quenching and photochemical H2S depletion in HD 189733b NIRCam data.","lead":"This paper introduces a way to account for disequilibrium chemistry in exoplanet atmosphere retrievals by adding quench-pressure parameters for carbon and nitrogen species. It shows the method recovers correct atmospheric properties on synthetic JWST data and finds tentative evidence for quenching and photochemical activity in real JWST observations of HD 189733b.","discovery_kind":"unclear","skeptic_critique":{"model":"glm-5.2","headline":"The headline Bayes factor of 7.3 (DisEq vs ChemEq) is computed between models that both omit the H2S parameterisation; when H2S is included — as the paper itself argues is necessary — the disequilibrium advantage vanishes (ΔlnZ ≈ 0.1).","rationale":"The reader correctly identified that the H2S parameterisation is ad hoc and that the disequilibrium and equilibrium+H2S models are statistically indistinguishable. However, the reader did not fully pinpoint the most load-bearing issue: the headline Bayes factor of 7.3 is computed from a comparison (DisEq vs ChemEq, both without H2S) that the paper's own logic renders invalid. The paper argues H2S modelling is necessary for this dataset, yet the headline statistic comes from models that omit it. When H2S is included, the disequilibrium preference disappears entirely (ΔlnZ ≈ −0.1). This is not merely a limitation the paper acknowledges — it undermines the specific quantitative claim (log Bayes factor 7.3) used to support 'tentative evidence for quenching.' The synthetic tests (Section 3) are methodologically sound and the quench-pressure parameterisation is a reasonable and useful contribution. The concern is narrowly about the real-data application: the strongest statistical claim uses an incomplete model comparison. The verdict remains CONDITIONAL because the method itself is defensible, but the observational claims should be reframed around the fair comparison (DisEq+H2S vs ChemEq+H2S), which yields no evidence for disequilibrium. The paper would need either NIRISS/SOSS data (as the authors themselves suggest) or a different statistical approach to break the degeneracy before claiming evidence for quenching in HD 189733b.","tokens_in":17974,"tokens_out":3193,"duration_ms":240544,"concrete_test":"Compute the Bayes factor between DisEq+H2S and ChemEq+H2S directly (the paper already reports lnZ=569.4 vs 569.5, so ΔlnZ≈−0.1). Then examine the best-fit spectrum of DisEq (no H2S) in the H2S feature region (~3.7–4.1 μm): if it misfits the H2S spectral feature while DisEq+H2S fits it properly, the 7.3 Bayes factor is driven by misattribution of the H2S signal to other species rather than genuine evidence for quenching. Report the residuals in this wavelength region for both models.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central real-data claim is a log Bayes factor of 7.3 favouring disequilibrium over equilibrium (Section 4.1, Table 2). However, this comparison is between DisEq (lnZ=571.8) and ChemEq (lnZ=564.5), both of which omit the H2S parameterisation. The paper itself develops an H2S parameterisation because the DisEq framework 'does not take into account sulfur chemistry' (Section 4.1), and Fu et al. (2024) detect H2S at 4.5σ. When H2S is included in both models, the evidence becomes: DisEq+H2S (lnZ=569.4) vs ChemEq+H2S (lnZ=569.5), yielding ΔlnZ ≈ −0.1 — i.e., no preference for disequilibrium, if anything slightly favouring equilibrium. The paper acknowledges the two models are 'indistinguishable within the error of the observations' but still leads with the 7.3 Bayes factor from the incomplete comparison. The deeper issue is why DisEq (no H2S) achieves higher evidence than DisEq+H2S (571.8 vs 569.4): the no-H2S model appears to be absorbing the H2S spectral signal into other free parameters (e.g., elevated CO, CO2, H2O VMRs via deep quenching, as shown in Figure 8), producing a spuriously good fit by misattributing a detected molecule's feature. The headline claim of 'tentative evidence for quenching' in HD 189733b therefore rests on a model comparison that excludes a component the paper identifies as essential, while the fair comparison yields a null result.","agreement_with_reader":"partial"},"referee_report":{"model":"glm-5.2","summary":"This paper presents a parameterised disequilibrium retrieval framework in which the quench pressures of carbon- and nitrogen-bearing species are treated as free parameters, with abundances truncated at the quench pressure following Eq. (3). The method is validated on synthetic JWST observations generated with VULCAN (forward model) and retrieved with NemesisPy/FastChem (retrieval model), testing multiple instrument configurations, mixing strengths, and TP profile parameterisations. The framework is then applied to published JWST/NIRCam transmission observations of HD 189733b, where the authors also introduce an ad hoc H2S vertical profile parameterisation to account for photochemical depletion. The synthetic tests demonstrate that the disequilibrium framework recovers bulk C/O and [M/H] more accurately than chemical equilibrium when vertical mixing is present, and the real-data application yields a carbon quench pressure of log P_q,C ~ 1.7 bar.","tokens_in":19025,"tokens_out":1727,"duration_ms":295899,"significance":"The quench-pressure parameterisation is a sensible and computationally efficient middle ground between free-chemistry and full kinetic-grid retrievals, and the synthetic retrieval tests are well designed: the forward (VULCAN) and retrieval (NemesisPy/FastChem) models are genuinely independent, priors are stated, and multiple instrument configurations and TP profiles are tested. The application to real JWST/NIRCam data and the introduction of an H2S vertical-gradient parameterisation are timely. However, the central real-data claim of 'tentative evidence for quenching' rests on a model comparison that is more nuanced than presented, as detailed below.","major_comments":[{"comment":"Section 4.1, Table 2: The headline log Bayes factor of 7.3 favouring disequilibrium (DisEq lnZ=571.8 vs ChemEq lnZ=564.5) is computed between models that both omit the H2S parameterisation. The paper itself argues that the H2S parameterisation is necessary because the DisEq framework 'does not take into account sulfur chemistry' and Fu et al. (2024) detect H2S at 4.5 sigma. When H2S is included in both models, the evidence becomes DisEq+H2S (lnZ=569.4) vs ChemEq+H2S (lnZ=569.5), yielding Delta lnZ ~ -0.1 — i.e., no preference for disequilibrium. The paper acknowledges the two models are 'indistinguishable within the error of the observations' but still leads with the 7.3 Bayes factor from the incomplete comparison in the abstract, Section 4.1, and conclusions. The fair comparison (both models including H2S) should be the primary basis for the real-data claim, and the abstract and结论should","section":null},{"comment":"Section 4.1, Table 2 and Figure 8: The DisEq (no H2S) model achieves higher evidence (lnZ=571.8) than DisEq+H2S (lnZ=569.4), which is counterintuitive — adding a physically motivated component that matches a 4.5-sigma detection should not decrease evidence. The text (Section 4.1, final paragraph) explains that the DisEq model without H2S elevates CO, CO2, and H2O VMRs via deep quenching, with CO being 'incompatible with the free retrieval.' This suggests the no-H2S model is absorbing the H2S spectral feature into other free parameters, producing a spuriously good fit by misattributing a detected molecule's feature. The paper should explicitly discuss this as a risk of the quenching framework — that deep quenching can inflate CO/CO2/H2O abundances to compensate for missing opacity sources — and clarify why the DisEq+H2S model, which includes the correct absorber, is the more physically可靠","section":null},{"comment":"Section 3.3, Eq. (4): The 'true' quench pressures used for validation are derived from the VULCAN models using an ad hoc threshold (a=0.25 for carbon, a=0.1 for nitrogen) applied to adjacent-layer VMR differences. The paper notes that 'the carbon quenching pressure is recovered at higher pressures in the atmosphere compared to the true values.' This systematic offset may be a direct consequence of the threshold definition rather than a retrieval failure: Eq. (4) identifies quenching when the relative VMR difference between adjacent layers falls below 'a', but this is not the same as the chemical-timescale vs mixing-timescale criterion that defines quenching in the kinetic model. The sensitivity of the 'true' quench pressure to the choice of 'a' should be quantified (e.g., by varying a by a factor of 2) to determine whether the systematic offset is a validation failure or an artifact of","section":null},{"comment":"Section 4.1, H2S parameterisation: The functional form — a deep well-mixed abundance plus a break pressure and power-law decline — is ad hoc, with no derivation from photochemical kinetics. It is applied to real data where only a single spectral feature constrains H2S (as shown in Figure 1, right panel), and the break pressure is then interpreted as the boundary of the photochemically active region. The paper itself notes the disequilibrium+H2S and equilibrium+H2S models are 'indistinguishable within the error of the observations.' The physical interpretation of P_break as the photochemical boundary should therefore be stated as conditional on the assumed functional form, and the claim of 'first tentative evidence of the constraint on the photochemically active region' should be moderated accordingly.","section":null}],"minor_comments":[{"comment":"Abstract: 'this is the first time this has been constrained in a hot Jupiter atmosphere' — the H2S constraint is tentative and model-dependent; consider softening.","section":null},{"comment":"Section 2.1: The footnote 'For the quenching retrievals, no sulfur chemistry is used' is important context that should be stated more prominently, as it affects the real-data application.","section":null},{"comment":"Table 2: The ordering of columns (DisEq+H2S, DisEq, ChemEq+H2S, ChemEq) makes cross-comparison awkward; consider reordering to group with/without H2S.","section":null},{"comment":"Figure 7: The bottom panel shows the difference between DisEq and ChemEq+H2S models, but the y-axis range and label are unclear; a clearer annotation of the 50 ppm deviation region would help.","section":null},{"comment":"Section 4.1: The statement 'we can begin to constrain the photochemically active regions of exoplanet atmospheres with JWST' is strong given the model degeneracy; should be framed as such.","section":null},{"comment":"Table 1 (free chemistry): This table is labelled Table 1 but conflicts with the prior table also labelled Table 1; should be renumbered.","section":null},{"comment":"Section 3.3: The statement that nitrogen quenches at deeper levels than carbon is consistent with theory, but the fact that nitrogen quenching hits the 100 bar boundary for high Kzz should be noted as a limitation.","section":null},{"comment":"Figure 8: The y-axis labels for VMR panels show inconsistent formatting (e.g., '10 5' vs '10^{-5}'); ensure consistent formatting.","section":null}],"recommendation":"major_revision","confidential_remarks":"The core methodological contribution — the quench-pressure parameterisation and its synthetic validation — is sound and useful. The real-data application, however, has a significant framing problem: the headline Bayes factor excludes a component the paper itself identifies as essential, and the fair comparison yields a null result. This is fixable by reframing the real-data claims around the DisEq+H2S vs ChemEq+H2S comparison and being transparent about the degeneracy. The H2S parameterisation is novel but ad hoc; it should be presented as a proof-of-concept rather than a robust constraint. I would encourage the authors to explicitly address why the no-H2S DisEq model achieves higher evidence than the H2S-included version, as this has implications for the reliability of the quenching framework when opacity sources are missing."},"author_rebuttal":{"model":"glm-5.2","summary":"We thank the referee for a thorough and constructive report. The referee correctly identifies that the fair model comparison for the HD 189733b data (both models including H2S) yields no preference for disequilibrium, and we will revise the abstract, Section 4.1, and conclusions to lead with this comparison rather than the incomplete comparison. We also agree that the counterintuitive evidence behaviour of the no-H2S DisEq model should be explicitly discussed as a risk of the quenching framework, that the sensitivity of the 'true' quench pressure to the threshold parameter 'a' should be quantified, and that the H2S parameterisation claims should be moderated. All four major comments will be addressed in revision.","responses":[{"response":"The referee is correct. The fair comparison is the one in which both models include the H2S parameterisation, and this yields Delta lnZ ~ -0.1, i.e., no statistical preference for disequilibrium over equilibrium. We will revise the abstract, Section 4.1, and the conclusions to lead with this fair comparison. The lnZ=7.3 Bayes factor from the incomplete comparison (both models omitting H2S) will be retained in the text for completeness but explicitly flagged as an incomplete comparison that should not be used as the basis for the quenching claim. The abstract will be revised to state that the disequilibrium and equilibrium models (both including H2S) are statistically indistinguishable for the NIRCam data, and that the tentative evidence for quenching is therefore conditional and would require additional data (e.g., NIRISS/SOSS) to break the degeneracy.","revision_made":"yes","referee_comment":"Section 4.1, Table 2: The headline log Bayes factor of 7.3 favouring disequilibrium is computed between models that both omit the H2S parameterisation. When H2S is included in both models, the evidence becomes DisEq+H2S (lnZ=569.4) vs ChemEq+H2S (lnZ=569.5), yielding Delta lnZ ~ -0.1 — i.e., no preference for disequilibrium. The fair comparison should be the primary basis for the real-data claim, and the abstract and conclusions should be revised accordingly."},{"response":"We agree with the referee's interpretation. The evidence pattern is consistent with the no-H2S DisEq model compensating for the missing H2S opacity by elevating CO, CO2, and H2O VMRs through deep quenching, as we note in the final paragraph of Section 4.1. We will add an explicit discussion of this as a known risk of the quenching framework: when an opacity source is missing, deep quenching can inflate the abundances of other species to compensate, producing a spuriously high evidence. This reinforces why the DisEq+H2S model, which includes the correct absorber, is the more physically reliable model and should be the basis for comparison. We will state this clearly in the revised text.","revision_made":"yes","referee_comment":"Section 4.1, Table 2 and Figure 8: The DisEq (no H2S) model achieves higher evidence than DisEq+H2S, which is counterintuitive. The no-H2S model may be absorbing the H2S spectral feature into other free parameters (CO, CO2, H2O) via deep quenching, producing a spuriously good fit. The paper should explicitly discuss this as a risk of the quenching framework and clarify why DisEq+H2S is the more physically reliable model."},{"response":"This is a valid concern. The threshold-based definition of the 'true' quench pressure in Eq. (4) is not equivalent to the chemical-timescale vs mixing-timescale criterion used in the kinetic model, and the systematic offset we report could partly reflect this mismatch. We will quantify the sensitivity of the 'true' quench pressures to the choice of 'a' by varying a by a factor of 2 in each direction (a=0.125 and 0.5 for carbon; a=0.05 and 0.2 for nitrogen) and will add a figure or table showing how the 'true' quench pressures shift. This will allow the reader to assess whether the retrieved offset is comparable to the uncertainty introduced by the threshold definition. We will also add a discussion noting that the threshold-based definition is an approximation and that a timescale-based comparison would be a useful future refinement.","revision_made":"yes","referee_comment":"Section 3.3, Eq. (4): The 'true' quench pressures used for validation are derived using an ad hoc threshold (a=0.25 for carbon, a=0.1 for nitrogen) applied to adjacent-layer VMR differences. The systematic offset between retrieved and 'true' quench pressures may be an artifact of the threshold definition rather than a retrieval failure. The sensitivity of the 'true' quench pressure to the choice of 'a' should be quantified (e.g., by varying a by a factor of 2)."},{"response":"We agree that the H2S parameterisation is ad hoc and that the interpretation of P_break as the photochemical boundary is conditional on the assumed functional form. We will revise the text to state this conditionality explicitly. The claim of 'first tentative evidence of the constraint on the photochemically active region' will be moderated to reflect that this constraint is conditional on the parameterisation and that, given the statistical indistinguishability of the DisEq+H2S and ChemEq+H2S models, it should be regarded as a tentative demonstration of the approach rather than a definitive measurement. The abstract will be revised accordingly.","revision_made":"yes","referee_comment":"Section 4.1, H2S parameterisation: The functional form is ad hoc, with no derivation from photochemical kinetics. The physical interpretation of P_break as the photochemical boundary should be stated as conditional on the assumed functional form, and the claim of 'first tentative evidence of the constraint on the photochemically active region' should be moderated accordingly."}],"tokens_in":18308,"tokens_out":1294,"duration_ms":171521,"standing_objections":[]},"desk_editor":{"model":"glm-5.2","letter":"The main thing to know: the synthetic tests are well-designed and the quench-pressure parameterisation is a useful, practical tool. But the real-data application has a framing problem that undercuts its headline claim, and the paper needs to confront it directly before publication. The stress-test note lands squarely. Table 2 shows that the advertised log Bayes factor of 7.3 (DisEq vs ChemEq) is computed between models that both omit H2S. When H2S is included in both — as the paper itself argues is necessary, given Fu et al.'s 4.5σ detection — the evidence difference collapses to ΔlnZ ≈ −0.1. The paper acknowledges this in the body text (Section 4.1: the two models are 'indistinguishable within the error of the observations'), but the abstract and conclusions still lead with 'tentative evidence for quenching.' That framing is not supported by the fair comparison. There's also a tell in the numbers: DisEq without H2S (lnZ=571.8) scores higher than DisEq with H2S (lnZ=569.4), which is backwards — adding a physically motivated component should help, not hurt. This suggests the no-H2S DisEq model is absorbing the H2S spectral feature into elevated CO, CO2, and H2O via deep quenching, as Figure 8 shows. That's a misattribution, not a detection of quenching. What the paper does well: the synthetic retrieval suite is genuinely independent (VULCAN forward, NemesisPy/FastChem retrieval), tests multiple instrument configurations and two TP profiles, and demonstrates clearly that equilibrium retrievals produce biased C/O and [M/H] when vertical mixing is present. The separate C/N quench pressures are a real improvement over single-quench-pressure approaches. The elemental vs molecular quenching comparison is a useful diagnostic. The carbon quench pressure is systematically recovered at higher pressure than truth, and the threshold values (a=0.25, a=0.1) in Eq. 4 are hand-set — these are minor issues that the paper is mostly honest about. The H2S parameterisation is ad hoc — three free parameters with no derivation from kinetics, applied to a single spectral feature. The interpretation of P_break as the photochemically active boundary is physically motivated but rests on a model preference that is statistically null. This paper is for retrieval practitioners who need a lightweight disequilibrium framework. The method deserves to be in the literature. But the real-data claims need honest reframing: the fair comparison yields a null result, and the paper should say so in the abstract, not just in Section 4.1. I'd recommend a serious referee who can push on this framing issue and on the H2S parameterisation's validation. The synthetic work is publishable; the observational claims need to be scaled back to what the evidence supports.","headline":"The synthetic retrieval tests are solid and the method is a genuine contribution, but the headline real-data claim doesn't survive the fair model comparison the paper itself reports.","tokens_in":18831,"tokens_out":1352,"would_cite":false,"duration_ms":104352,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"glm-5.2","headline":"Two free parameters fix disequilibrium biases in JWST exoplanet retrievals","keywords":["exoplanet atmospheres","atmospheric retrieval","disequilibrium chemistry","quench pressure","vertical mixing","JWST","HD 189733b","transmission spectroscopy"],"falsifier":"If a retrieval assuming chemical equilibrium on a planet with known strong vertical mixing recovers the correct C/O and [M/H] without quench parameters, or if the quench pressure framework fails to recover known input values on synthetic spectra with mixing strengths between the four tested Kzz values, the parameterisation's utility would be undermined.","tokens_in":18059,"feed_emoji":"🪐","tokens_out":1136,"duration_ms":108880,"temperature":0.7,"pith_summary":"This paper argues that the standard assumption of chemical equilibrium in exoplanet atmospheric retrievals produces systematically wrong estimates of carbon-to-oxygen ratio and metallicity whenever vertical mixing is present, and that the fix is cheap: add two free parameters representing the quench pressures for carbon-bearing and nitrogen-bearing species. At pressures above the quench point, abundances are frozen at their deep-atmosphere equilibrium values rather than following equilibrium throughout. Using synthetic JWST spectra of the hot Jupiter HD 189733b generated with a full kinetic chemistry model, the authors show that equilibrium retrievals drift further from truth as mixing strength increases, while the two-parameter quenching framework recovers the correct bulk composition in most cases. Applied to real JWST/NIRCam transmission data of HD 189733b, the disequilibrium model is favoured over equilibrium with a log Bayes factor of 7.3, with the carbon quench pressure retrieved at roughly 50 bar, implying strong vertical mixing. The paper also introduces a three-parameter parameterisation for the vertical profile of hydrogen sulfide (H2S), consisting of a deep well-mixed abundance, a break pressure, and a power-law decline, which it applies to the same data to tentatively locate the boundary of the photochemically active region at around 0.6 millibar.","feed_headline":"Two free parameters fix disequilibrium biases in JWST exoplanet retrievals","feed_subtitle":"Adding quench pressures for carbon and nitrogen species to standard equilibrium retrievals recovers correct atmospheric composition where垂直混","key_machinery":"quench_pressure_parameterisation","core_discovery":"The central object is the quench pressure parameterisation (Eq. 3): for each chemical family (carbon or nitrogen), a single pressure level is retrieved above which the volume mixing ratio is held constant at its equilibrium value at that pressure. This two-parameter extension to standard equilibrium retrievals is shown to recover unbiased C/O and [M/H] from synthetic spectra where vertical mixing is present, while standard equilibrium retrievals do not. On real JWST/NIRCam data of HD 189733b, the framework yields a log Bayes factor of 7.3 over equilibrium and places the carbon quench pressure at approximately 1.7 bar (or deeper, at roughly 50 bar, in the variant without the H2S parameteriser","pith_inferences":["If quench pressures are reliably retrievable from JWST spectra, a sample of hot Jupiters with measured quench pressures could be used to calibrate eddy diffusion prescriptions in general circulation models, providing an observational anchor for a parameter that is currently set by theory alone.","The fact that nitrogen and carbon quench at different pressures, and that both are independently retrieved, suggests that multi-instrument wavelength coverage spanning both CH4 and NH3 spectral features is necessary to break degeneracies between the two quench parameters.","The H2S break pressure interpretation as the photochemical boundary could be tested by comparing retrieved break pressures across a temperature gradient of hot Jupiters, since photochemical timescales scale with temperature and the boundary should shift predictably."],"forward_implications":["Future exoplanet retrieval studies that assume chemical equilibrium should routinely include quench pressure parameters for carbon and nitrogen species, as the paper demonstrates measurable bias when they are omitted.","Retrieved quench pressures can serve as proxies for vertical mixing strength, linking transmission spectroscopy to deep atmospheric dynamics that are otherwise unobservable.","The H2S vertical profile parameterisation, if validated on planets with both H2S and SO2 spectral features, could map the transition between the quenched region and the photochemically active region in exoplanet atmospheres.","The degeneracy between the disequilibrium model and the equilibrium-plus-H2S model on HD 189733b can be broken with NIRISS/SOSS observations, where the two models diverge by over 50 ppm."],"fun_headline_variants":["Parameterized quenching fixes JWST retrieval biases on HD 189733b","Two-parameter quenching model recovers unbiased JWST exoplanet spectra","Disequilibrium retrievals correct C/O biases in JWST hot Jupiter data","First H2S constraint in a hot Jupiter from JWST disequilibrium retrievals","Adding quench pressures corrects JWST exoplanet atmospheric biases"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The H2S parameterisation is an ad hoc functional form with three free parameters and no derivation from photochemical kinetics, applied to real data where only a single spectral feature constrains H2S. The paper itself notes that the disequilibrium model and the equilibrium-plus-H2S model are statistically indistinguishable given current observations, so the physical interpretation of the break pressure as the photochemical boundary rests on a model preference that is not yet","fun_headline_variants_meta":{"raw":{"variants":["Parameterized quenching fixes JWST retrieval biases on HD 189733b","Two-parameter quenching model recovers unbiased JWST exoplanet spectra","Disequilibrium retrievals correct C/O biases in JWST hot Jupiter data","First H2S constraint in a hot Jupiter from JWST disequilibrium retrievals","Adding quench pressures corrects JWST exoplanet atmospheric biases","New quench pressure framework removes biases from JWST retrievals"]},"model":"glm-5.2","effort":"high","cost_usd":0.0,"raw_usage":{"total_tokens":1560,"prompt_tokens":636,"completion_tokens":924,"prompt_tokens_details":null},"tokens_in":636,"tokens_out":924,"duration_ms":43169,"temperature":1.0,"reasoning_tokens":758,"cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-08T03:59:46.883989+00:00","model_set":{"reader":"glm-5.2"},"falsifier":"If a retrieval assuming chemical equilibrium on a planet with known strong vertical mixing recovers the correct C/O and [M/H] without quench parameters, or if the quench pressure framework fails to recover known input values on synthetic spectra with mixing strengths between the four tested Kzz values, the parameterisation's utility would be undermined.","supporting_citations":[],"review_version":1}