{"id":"156fe80c-574b-4317-91f1-52a3822785c1","arxiv_id":"1908.07757","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Using a photoionization model of the NGC 5548 broad line region, the authors show that diffuse continuum emission can add several light-days to measured inter-band continuum delays and provide a scaling recipe for correcting them.","lead":"This paper models how diffuse continuum light from the broad line region of an active galaxy contaminates measured time delays between continuum bands. It finds the contamination can reach roughly 40 percent of the light near the Balmer jump and offers a recipe to correct disk reverberation measurements.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quantitative correction in Equation 3 assumes the underlying disk is lag-less at all wavelengths, but the NGC 5548 scale of observed disk lags is comparable to the predicted DC delays, so the measured inter-band delay cannot be cleanly separated and the DC contamination amplitude is not…","rationale":"The paper is a careful forward-modeling study, and the central qualitative claim, that a BLR diffuse continuum significant enough to affect continuum inter-band delays is a natural consequence of models calibrated to the strongest UV lines, is well supported. The photoionization calculations use a realistic SED, the line-strength calibration is explicit, and the model's predictions of a strong Balmer/Paschen continuum are consistent with the long-standing spectral decomposition evidence for a Balmer excess in AGN. The main numerical results, however, are obtained under a specific separation ansatz: the underlying disk continuum is a scaled, lag-less replica of the driver. This assumption is not a minor technical detail because the paper's own Figure 10 predicts diluted DC delays of only a few days, and the observed disk inter-band lags in the same source are also a few days. In that regime the measured delay is not an interpolation between zero and tau_DC; it is a nonlinear combination of at least three time-variable components with comparable lags. Equation 3, therefore, cannot be used to correct the measured delays unless the disk lag spectrum is known independently, which is exactly the quantity disk reverberation experiments are trying to measure. This concern does not overturn the conclusion that DC contamination exists and is important; it does mean the quantitative correction recipe and the exact day-scale of contamination are conditional on an assumption that is not valid for the target observations. The reader's weakest-assumption statement identified the same issue, including the acknowledged limitation in Section 3.3 and the driver-proxy caveat in Section 2.4.2, so the conditional verdict remains appropriate.","tokens_in":26786,"tokens_out":6783,"duration_ms":77353,"concrete_test":"Recompute Figure 10 with the lambda-1157-Angstrom driver unchanged but replacing the lag-less underlying band light curves with a disk-reprocessing model whose wavelength-dependent lags are set to the observed NGC 5548 inter-band delays (or to a lamppost model with tau(lambda) proportional to lambda^(4/3) normalized to those delays), then fit Equation 3 to the resulting summed light curves. If the input DC fraction cannot be recovered to within about 0.05 in x at lambda-3641 and lambda-5200 Angstrom, the recipe fails exactly in the regime it was designed for.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2.5 fixes the non-DC component in every band as a scaled, lag-less copy of the lambda-1157-Angstrom driver ('we here assume to be lag-less with respect to the driver'), and Section 3.3 explicitly limits the recipe to cases where underlying disk lags are 'much shorter' than BLR DC delays. This is not satisfied by the data the paper targets. For NGC 5548 the observed UV-optical inter-band lags relative to a UV reference are of order 1-3 days, while the diluted DC delays in Figure 10 are of the same order (about 0.5-3 days) and the un-diluted DC delays in Figure 6 are larger (about 2-10 days). With two comparable delay scales, the CCF centroid of the summed light is a nonlinear mixture of the disk lag spectrum and the BLR DC lag spectrum. Equation 3 and Figure 11, which interpolate between tau = 0 and tau = tau_DC at fixed DC fraction, cannot by construction retrieve the DC fraction or the clean disk lag. Thus the quantitative 'several-day' contamination amplitude and the proposed correction are not uniquely determined from this model, even though the existence of a significant BLR DC contribution is independently supported by the calibrated Ly-alpha/C IV match and the Balmer-excess phenomenology. The related driver-proxy uncertainty is also acknowledged in the Section 2.4.2 footnote, which notes that the UV-optical continuum may be a poor proxy for the driving EUV continuum during the anomalous 2014 state; this further shifts the STORM-derived numbers.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses Cloudy v17.00 photoionization models in a LOC framework to compute the broad-line-region diffuse continuum (DC) flux and delay spectra for NGC 5548, spanning 1000–10000 Å. The model is calibrated by matching the observed Lyα and C IV equivalent widths and line ratios, and the predicted line lags are compared against the 1989 IUE, 1993 HST, and 2014 AGN STORM campaigns. The authors find that models matching the strong UV lines also produce a significant DC component, reaching roughly 40% of the total light near the Balmer jump, and that this component contributes wavelength-dependent delays of order a few days to the measured continuum inter-band delays. They explore how the DC luminosity and delays depend on gas density, column density, driving-continuum amplitude, and variability timescale, and they provide a recipe (Eq. 3) for estimating and correcting the DC contribution in disk reverberation mapping experiments.","tokens_in":27156,"tokens_out":5165,"duration_ms":49599,"significance":"If the central result holds, the paper establishes that BLR diffuse continuum emission is a non-negligible contaminant in UV-optical disk reverberation mapping, potentially biasing measured disk sizes and time lags. The strengths of the paper are its use of a public, well-documented photoionization code; the explicit calibration against observed line strengths and line lags from three campaigns (Table 2); and the falsifiable, quantitative predictions for the wavelength dependence of DC delays and flux fractions. The finding that the DC delay spectrum is much shorter than the Hβ lag, and that Hβ is a poor proxy for Balmer-continuum delays, is a useful and nontrivial result. The paper also makes a clear case that the DC contribution can partly explain the 'bluer-when-brighter' behavior. However, the quantitative correction recipe and the reported diluted-delay amplitudes depend on an assumption about the underlying disk lags that is not satisfied for the target source, as detailed in the major comments.","major_comments":[{"comment":"The quantitative analysis that produces the diluted delays and the correction recipe assumes the underlying disk continuum in each band is a scaled, lag-less replica of the λ1157 Å driver ('we here assume to be lag-less with respect to the driver'). This assumption is not satisfied for NGC 5548, the source to which the model is applied: observed UV-optical inter-band lags relative to a UV reference are of order 1–3 days, while the diluted DC delays in Fig. 10 are ~0.5–3 days (and the undiluted DC delays in Fig. 6 are ~2–10 days). With two comparable delay scales, the CCF centroid of the summed light is a nonlinear mixture of the disk lag spectrum and the BLR DC lag spectrum, so Eq. (3), which interpolates between τ=0 and τ=τ_DC at fixed DC fraction, cannot by construction recover either the DC fraction or the clean disk lag. Section 3.3 limits the recipe to cases where disk lags are 'much shorter' than the DC delays, which excludes the NGC 5548 case for which the numbers are computed. The quantitative contamination amplitudes in Fig. 10 and the proposed correction are therefore not uniquely determined from this model. I recommend the authors either forward-model a wavelength-dependent disk lag spectrum and recompute the dilution, or explicitly reframe the recipe and the quoted amplitudes as upper limits applicable only when the disk lag spectrum is known to be sub-dominant.","section":"§2.5, Fig. 10, Eq. (3)"},{"comment":"The driving continuum proxy is the λ1157 Å band, and the paper acknowledges in a footnote that the UV-optical continuum may be a poor proxy for the driving EUV continuum during the anomalous 2014 state. Because all predicted delays in Figs. 6, 10, and 11 are correlations with this proxy, a mismatch between the proxy and the true ionizing continuum would shift the DC delay spectra and also change the fractional DC contributions used in the dilution simulations. The paper's sensitivity analysis in §2.7.1 scales the amplitude of the driver, but the full simulations in §2.4–2.6 are not rerun with the EUV-scaled driver, so the impact on the reported delay spectra and on Eq. (3) is not quantified. Please provide a quantitative sensitivity test (e.g., rerun the dilution simulations with the α=1.5 driver and report the resulting changes in Figs. 10 and 11) or state a conservative uncertainty on the absolute delays due to the driver-proxy ambiguity.","section":"§2.4.2, footnote 6, and §2.7"}],"minor_comments":[{"comment":"The text says the extended light curve includes a significant continuum event 'starting ≈200 days prior to the start of the HST campaign,' while the lower-panel caption of Fig. 5 says the event is '≈140 days prior.' Please reconcile the numbers.","section":"§2.4.1, Fig. 5"},{"comment":"The observed line lags are quoted without uncertainties, which makes it difficult to judge the quality of the model match. Adding the published measurement uncertainties (e.g., from De Rosa et al. 2015 and Pei et al. 2017) would strengthen the calibration claim.","section":"Table 2"},{"comment":"There is a typo in 'NGGC 5548' (extra G). Also, the abundance description '0.5× solar metallicity, except solar values in C/H and N/H' would benefit from a brief reminder that this is the same abundance set as KG00, to avoid forcing the reader to look up the reference.","section":"§2.2"},{"comment":"The discussion of open vs. closed geometries and the missing Lyman continuum is interesting but is not directly connected to the quantitative results in §2.5–2.6; consider condensing it or adding a sentence that states its implications for the DC delay predictions.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is a careful and well-calibrated modeling study, and the qualitative conclusion that BLR diffuse continuum is a significant contaminant in disk reverberation mapping is well supported. The main issue is that the quantitative correction recipe and the quoted dilution amplitudes rely on a lag-less underlying disk, which is not satisfied for NGC 5548; this needs to be addressed with either a forward model including disk lags or an explicit re-framing as upper limits. The self-citations to KG00/KG01 are appropriate given that the paper explicitly extends that framework. The manuscript fits the journal scope well."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is the paper that turns KG01's warning about BLR diffuse continuum contamination into a quantitative, testable model for NGC 5548. The central qualitative result survives scrutiny: a BLR model that reproduces the observed Lyα and C IV strengths and lags also produces a diffuse continuum that is a significant fraction of the UV-optical light, and that fraction would imprint a wavelength-dependent, multi-day signature on measured inter-band delays. That alone matters for disk reverberation mapping.\n\nWhat is genuinely new is the package: a modern SED, three observed driving light curves, wavelength-dependent DC delay spectra, a non-linear dilution relation (Eq. 3), and a geometry robustness check. The F(r) ~ r^-2 result and the point that Hβ lags are poor proxies for Balmer continuum lags are useful. Appendix A's census of NLR and torus continuum contributions is a nice bonus. The forward-model strategy—constraining the BLR with line strengths and lags before predicting the DC—keeps the circularity burden low. Self-citations to KG00/KG01 are appropriate; this is an extension, not a repackaging.\n\nThe soft spot is real, and it is in the quantitative recipe. Equation 3 and Figure 11 assume the underlying disk continuum in each band is a lag-less, scaled copy of the 1157 Å driver. The paper itself says the recipe works only when underlying disk lags are 'much shorter' than the BLR DC delays. For NGC 5548, the observed UV-optical inter-band disk lags are ~1–3 days, and the diluted DC delays in Figure 10 are ~0.5–3 days. Those scales are not separated. When the two components have comparable delays, the CCF centroid of their sum is a nonlinear mixture, and interpolating between τ=0 and τ_DC at fixed DC fraction cannot uniquely recover either the DC fraction or the clean disk lag. So the paper convincingly establishes that BLR DC contamination is significant, but the specific amplitude and the correction recipe are not uniquely determined by this model. The footnote about the 2014 UV-optical continuum being a possible poor proxy for the EUV driver adds another shift to the STORM-derived numbers.\n\nThis is not a fatal flaw. The qualitative claim is robust, and the paper is honest about the limitation in Section 3.3. I would send it to a serious referee. The right referee request is to either demonstrate the recipe on a case where disk lags are genuinely negligible, or explicitly restrict the recipe's domain and discuss how the comparable-lag case changes the inference. I would not desk-reject; I would cite it for the qualitative contamination result, with the caveat.","headline":"BLR diffuse continuum contamination of inter-band delays is real and quantitatively modeled; the correction recipe assumes a lag-less disk, which fails for the source it targets.","tokens_in":27670,"tokens_out":4534,"would_cite":true,"duration_ms":43975,"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":"Models that reproduce NGC 5548's strong UV lines also produce a diffuse continuum reaching about 40 percent of the light in some continuum bands, contaminating measured inter-band delays with a several-day, wavelength-dependent signature.","keywords":["active galactic nuclei","broad line region","diffuse continuum","reverberation mapping","inter-band continuum delays","accretion disk","NGC 5548","photoionization modeling"],"falsifier":"Compute the diffuse-continuum fraction $x$ in each continuum band from a high signal-to-noise spectrum of an AGN with a measured inter-band delay spectrum. Equation 3 predicts the delay should approach zero as $x\\to0$ and follow $(1-A)x/(1-Ax)$ at intermediate $x$; observing multi-day, wavelength-dependent lags in the lowest-$x$ bands, or a delay-versus-$x$ curve that cannot be fitted with a single $A$, would falsify the lag-less disk assumption and demonstrate that the disk itself contributes wavelength-dependent delays.","tokens_in":26552,"feed_emoji":"🔭","tokens_out":13693,"duration_ms":184707,"temperature":0.7,"pith_summary":"Disk reverberation mapping measures time delays between continuum bands to infer the structure of the accretion disk, but the disk is not the only variable continuum source. The paper argues that the same broad-line-region gas that emits AGN spectral lines also emits a diffuse continuum, and that in a BLR model reproducing the strong UV lines of NGC 5548 this diffuse component can reach about 40 percent of the total light in a band near the Balmer jump. Because the diffuse continuum arises over a large volume and responds on time scales of days, it adds a wavelength-dependent delay signal that contaminates measured inter-band delays. The authors show that the contamination is diluted by the underlying delay-free continuum, derive a nonlinear correction recipe, and demonstrate that the size of the effect depends on the amplitude and variability timescale of the driving continuum. Getting this correction right matters because failing to remove it biases inferred disk sizes and mass accretion rates.","feed_headline":"Diffuse gas glow contaminates AGN time-delay maps","feed_subtitle":"At the Balmer jump the diffuse glow reaches about 40 percent of a band's light, enough to bias disk-size estimates.","key_machinery":"The central machinery is a locally optimally emitting cloud (LOC) model: a spherical ensemble of photoionized clouds spanning a range of hydrogen densities and column densities, whose summed emission is forced to match the observed Lyα and C IV strengths and lags in NGC 5548. Within this model, the diffuse-continuum bands have radial surface emissivity distributions $F(r)$ approximately proportional to $r^{-2}$, i.e. a responsivity $\\eta(r)\\approx1$, which makes their response to driving continuum variations simple and relatively insensitive to BLR geometry. The second load-bearing element is the dilution formula, Equation 3, $\\tau_\\lambda\\approx\\tau_\\lambda^{\\rm DC}(1-A)x/(1-Ax)$, which connects the measured delay in a band to the diffuse-continuum fraction $x$ and to $A$, a constant encoding how the cross-correlation centroid responds to a mixture of lag-less disk light and delayed diffuse-continuum light.","core_discovery":"Using a spherical ensemble of photoionized clouds that reproduces the observed Lyα and C IV luminosities and time delays in NGC 5548, the paper shows that the same gas emits a diffuse continuum made of free-bound, free-free, and scattered light. In the steady-state model, the diffuse continuum reaches roughly 60 percent of the incident continuum at the Balmer jump (about 40 percent of total light), with emissivity-weighted radii of about 20 to 40 light-days across 1000 to 10000 Å. Driven by the 2014 monitoring light curve, the diffuse-continuum-only cross-correlation delays average about $6.5 \\pm 1.7$ days over that range, with a strong wavelength dependence and abrupt changes at the Balmer and Paschen jumps. When a scaled, lag-less version of the 1157 Å driver is added to represent the underlying disk, the measured delay is a nonlinear function of the diffuse-continuum fraction $x$, approximately $\\tau_\\lambda \\approx \\tau_\\lambda^{\\rm DC}(1-A)x/(1-Ax)$ with $A\\approx0.65$ to $0.76$ for the 2014 campaign. The paper provides a recipe for scaling these predictions to other AGN luminosities and for correcting observed delay spectra.","pith_inferences":["If the 1157 Å reference band itself contains diffuse continuum, all measured delays are relative to a contaminated zero point; switching the driver to a band closer to the ionizing continuum or to a high-ionization line should reveal a uniform shift in the recovered delay spectrum.","Because the diffuse-continuum surface emissivity is nearly $r^{-2}$, its transfer function is a smoothed, almost linear mirror of the driver; lag spectra could be fit with a two-component model, a compact disk response plus an $r^{-2}$ BLR response, to estimate both the disk temperature profile and the BLR radius from the same data.","Since AGN are bluer when brighter, the diffuse-continuum fraction $x$ changes with continuum state; Equation 3 then predicts that measured lags in a given band should vary as the source brightens, a testable prediction with simultaneous spectral and lag monitoring.","Sources with stronger emission lines relative to the underlying continuum should show systematically larger diffuse-continuum fractions; the Lyα equivalent width could serve as a cheap predictor of the expected contamination level."],"forward_implications":["Uncorrected inter-band delays fitted with $\\tau(\\lambda)\\propto\\lambda^{4/3}$ will overestimate disk sizes and mass accretion rates, because each measured delay is a mixture of disk response and BLR diffuse-continuum response.","The enhanced delays observed around the Balmer continuum can be reproduced by BLR diffuse continuum without requiring an anomalous disk temperature profile.","Corrections that linearly scale measured lags by the diffuse-continuum fraction will overestimate the contamination for large $x$; Equation 3 with a fitted $A$ is needed.","The Hβ lag is a poor proxy for the Balmer-continuum delay: the diffuse continuum responds from radii about a factor of two smaller, so its lag and variability amplitude differ from Hβ.","The contamination, and hence the required correction, depends on the amplitude and characteristic timescale of the driving continuum, so campaign-specific light-curve properties enter the delay calibration."],"supporting_citations":[{"why":"Supplies the spherical BLR cloud-distribution model whose emission-line strengths and lags calibrate the diffuse-continuum predictions.","marker":"KG00"},{"why":"Earlier estimate of a linear DC contribution to inter-band delays, the baseline this paper revises.","marker":"KG01"},{"why":"Prior modeling of BLR diffuse-continuum contamination that this work extends across 1000–10000 Å with new constraints.","marker":"Lawther et al. (2018)"},{"why":"The photoionization code used to compute the cloud emission grids.","marker":"Ferland et al. (2017)"},{"why":"Provides the adopted NGC 5548 spectral energy distribution that fixes the underlying continuum level.","marker":"Magdziarz et al. (1998)"},{"why":"Supplies the 2014 continuum light curves and rms variability amplitudes used to drive and normalize the simulations.","marker":"Fausnaugh et al. (2016)"},{"why":"Reports observed UV-optical inter-band delays that define the observational target.","marker":"Edelson et al. (2015)"},{"why":"Supplies measured UV broad-line luminosities and lags used to validate the model BLR.","marker":"De Rosa et al. (2015)"},{"why":"Defines the cross-correlation lag measurement used on all simulated light curves.","marker":"White & Peterson (1994)"},{"why":"Provides the alternative bowl-shaped BLR geometry used to test the geometry-dependence of DC delay spectra.","marker":"Goad et al. (2012)"}],"fun_headline_variants":["BLR diffuse glow skews AGN disk size estimates","Diffuse continuum from gas biases AGN delay maps","Gas glow confounds AGN accretion disk measurements","Nonlinear correction needed for AGN time delays","BLR gas emission distorts disk reverberation lags"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The correction recipe assumes the underlying accretion-disk continuum in each longer-wavelength band is a scaled, lag-less replica of the 1157 Å driver; if the disk itself has wavelength-dependent delays of several days, comparable to the diffuse-continuum delays, the measured lags cannot be separated using Equation 3.","fun_headline_variants_meta":{"raw":{"variants":["BLR diffuse glow skews AGN disk size estimates","Diffuse continuum from gas biases AGN delay maps","Gas glow confounds AGN accretion disk measurements","Nonlinear correction needed for AGN time delays","BLR gas emission distorts disk reverberation lags"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00021,"raw_usage":{"total_tokens":1502,"prompt_tokens":1126,"completion_tokens":376,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":742,"completion_tokens_details":{"reasoning_tokens":300}},"tokens_in":742,"tokens_out":376,"duration_ms":508606,"temperature":1.0,"reasoning_tokens":300,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:57:17.215703+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the diffuse-continuum fraction $x$ in each continuum band from a high signal-to-noise spectrum of an AGN with a measured inter-band delay spectrum. Equation 3 predicts the delay should approach zero as $x\\to0$ and follow $(1-A)x/(1-Ax)$ at intermediate $x$; observing multi-day, wavelength-dependent lags in the lowest-$x$ bands, or a delay-versus-$x$ curve that cannot be fitted with a single $A$, would falsify the lag-less disk assumption and demonstrate that the disk itself contributes wavelength-dependent delays.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior modeling of BLR diffuse-continuum contamination that this work extends across 1000–10000 Å with new constraints."},{"cited_title":"et al.\\ 2017, RMxAA 53, 385","cited_arxiv_id":null,"evidence_quote":"The photoionization code used to compute the cloud emission grids."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the adopted NGC 5548 spectral energy distribution that fixes the underlying continuum level."},{"cited_title":"et al.\\ 2016, \\ 821, 56, ( Paper iii )","cited_arxiv_id":null,"evidence_quote":"Supplies the 2014 continuum light curves and rms variability amplitudes used to drive and normalize the simulations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports observed UV-optical inter-band delays that define the observational target."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies measured UV broad-line luminosities and lags used to validate the model BLR."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the cross-correlation lag measurement used on all simulated light curves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the alternative bowl-shaped BLR geometry used to test the geometry-dependence of DC delay spectra."}],"review_version":1}