{"id":"74ed0d3f-cc92-4636-ab17-c402585a9265","arxiv_id":"1908.03950","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"An independent inversion pipeline reproduces JSOC's horizontal flow maps but indicates that JSOC's sound-speed perturbation maps are inflated by cross-talk and are sensitive to depths different from those labeled.","lead":"This paper checks the public JSOC time-distance helioseismology maps of solar flows and sound-speed changes against an independent inversion pipeline. It finds the flow maps are reliable near the surface but the sound-speed maps are inflated by cross-talk and their labeled depths do not match the actual sensitivities.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Cross-talk explanation for JSOC sound-speed excess is not directly tested because JSOC sound-speed averaging kernels are not used; the RMS excess could also reflect resolution or noise differences.","rationale":"The paper's horizontal-flow reproduction is convincing: correlation coefficients 0.86-0.94 and comparable RMS values demonstrate that the two pipelines see the same flow structures. The non-localization of JSOC flow averaging kernels is also directly supported by the computed kernels and is not in question. The concern is specific to the sound-speed amplitude claim. The reader's weakest assumption (external model amplitude) is related but not identical; I identify a more direct internal gap: the JSOC sound-speed averaging kernels are never obtained or used, so the cross-talk contribution to the sound-speed maps is never quantified. The RMS excess could in principle be produced by other differences between the pipelines, such as horizontal resolution, vertical weighting, or noise. A single test using the JSOC sound-speed kernels would settle whether cross-talk actually explains the excess. Without this test, the conclusion that JSOC sound-speed products are 'strongly affected' by cross-talk remains plausible but not demonstrated, so the verdict should remain CONDITIONAL on this verification.","tokens_in":28147,"tokens_out":5336,"duration_ms":55281,"concrete_test":"Obtain or compute the JSOC sound-speed averaging kernels K^{cs}_{flow} for the first four target depths (they are products of the same RLS inversion and should be available from the JSOC pipeline). Using the JSOC horizontal-flow maps v_x, v_y and these kernels, form δcs_xtalk(r) = ∫ [K^{cs}_x v_x + K^{cs}_y v_y] d r' dz (with appropriate model-S weighting). Compare the RMS and spatial correlation of δcs_xtalk with the difference map JSOC_δcs - OUR2_δcs (or JSOC_δcs - OUR3_δcs). If δcs_xtalk accounts for most of the 8-9 m/s RMS excess and correlates with the difference map, cross-talk is confirmed and the central claim stands. If it does not, the excess must be attributed to resolution, noise, or real signal, and the cross-talk claim should be weakened. This test is independent of the external model assumption about the true sound-speed amplitude.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim that JSOC sound-speed perturbations are strongly affected by cross-talk rests on an indirect comparison. The paper obtains JSOC averaging kernels only for the horizontal flow components (Section 3.1: 'we also managed to obtain the averaging kernels for the horizontal flows for the first four depths'), not for sound speed. The cross-talk term specific to sound speed is the leakage from the (larger-amplitude) flows into the sound-speed averaging kernel, K^{cs}_{flow}. Without these kernels, the actual cross-talk contribution to the JSOC sound-speed maps cannot be computed. The evidence offered instead is that (i) MC-SOLA inversions with similar vertical localization give lower RMS, and (ii) a synthetic study (Korda & Švanda 2019) indicates cross-talk can be about half the sound-speed estimate in such inversions. But the RMS excess (JSOC 18 m/s vs OUR 9-10 m/s at 2 Mm) could also be produced by a narrower horizontal averaging kernel in the JSOC sound-speed inversion, by a different vertical weighting not captured by the flow kernels, or by unaccounted noise in the sound-speed channel. The paper does not compare JSOC sound-speed noise levels or horizontal widths. Thus the attribution to cross-talk, which is the load-bearing step linking the observed amplitude excess to the advertised bias, is underdetermined by the data actually shown. The external assumption that true quiet-Sun sound-speed perturbations are an order of magnitude below flows (Section 4.2) narrows the possibilities but does not single out cross-talk.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper compares the standard JSOC time–distance inversion products for horizontal flows and sound-speed perturbations with maps produced by the authors' independent MC-SOLA pipeline applied to the same SDO/HMI Dopplergrams. The authors perform three types of inversions: a JSOC-like inversion using the same travel-time geometries, a JSOC-like-target inversion using a richer set of travel times, and a JSOC-indicated-target inversion using localized Gaussian target functions. They report that horizontal flow maps are well reproduced (correlations 0.82–0.94), while JSOC sound-speed maps have systematically larger RMS amplitudes, which they attribute to cross-talk with the larger-amplitude flows. They also show that the JSOC averaging kernels for the horizontal flows are broad in depth and not localized around the labeled depths, and they argue that this makes the labeled depth structure of the public products misleading. The paper concludes that JSOC inversions are representative of near-surface layers but that the sound-speed amplitudes are likely overestimated through cross-talk.","tokens_in":28412,"tokens_out":4047,"duration_ms":42973,"significance":"If the central conclusion holds, the paper has a direct practical impact: users of the public JSOC time–distance products should not interpret the labeled depths as localized sensitivity depths, and sound-speed amplitudes should be treated with caution. The comparison is genuinely useful because it is an external validation of a widely used data pipeline by an independent inversion code, and it makes use of the actual JSOC averaging kernels for the horizontal flows as target functions. The horizontal-flow reproduction is a solid positive result, and the non-localization of the flow kernels is well documented with quantitative indicators (Tables 1 and 4). The main weakness is that the sound-speed cross-talk attribution, which is the load-bearing claim in the abstract, is not directly tested with JSOC sound-speed averaging kernels and depends on an external assumption about the magnitude of quiet-Sun sound-speed perturbations.","major_comments":[{"comment":"The abstract and Section 4.2.2 assert that JSOC sound-speed perturbations are 'strongly affected by the high level of the cross-talk' based on the RMS excess and the prior synthetic study Korda & Svanda (2019). However, Section 3.1 reports that averaging kernels were obtained only for the horizontal flows, not for the JSOC sound-speed inversions. The sound-speed-specific cross-talk term, K^{cs}_{flow}, is therefore never computed for the JSOC setup. The observed RMS excess (JSOC 18 m/s vs. OUR2 9 m/s at 2 Mm) could also arise from differences in horizontal averaging width, vertical weighting, or noise between the two pipelines. The paper should either obtain the JSOC sound-speed averaging kernels, present a synthetic test that quantifies the expected cross-talk amplitude for the actual JSOC setup, or revise the claim to state that cross-talk is a plausible but not yet demonstrated explanation.","section":"Section 4.2.2 and Conclusions"},{"comment":"The interpretation of the larger JSOC sound-speed RMS as contamination presupposes that quiet-Sun sound-speed perturbations are genuinely about an order of magnitude smaller than horizontal flow velocities, citing Rempel (2014) and DeGrave et al. (2014). This assumption is not tested against the data presented in this paper. If the true sound-speed perturbations near 1–3 Mm are larger than those simulations suggest, part or all of the JSOC excess could be real signal. The authors should quantify how sensitive their cross-talk conclusion is to plausible variations in the assumed sound-speed amplitude, or provide an independent observational constraint on that amplitude in the analyzed region.","section":"Section 4.2, first paragraph"}],"minor_comments":[{"comment":"There is a typo in the sentence 'These ocillations are best observed in Doppler shifts' — 'ocillations' should be 'oscillations'.","section":"Section 1.1"},{"comment":"In the 'JSOC-indicated target' rows of Table 6 the correlation column is labeled 'corr(OUR2, OUR3)', unlike the 'corr(OUR1, JSOC)' and 'corr(OUR2, JSOC)' used elsewhere. Since the text explains that no JSOC vertical-flow maps are available, this label is understandable, but it should be defined in the table caption for clarity.","section":"Table 6"},{"comment":"The statement that the lower RMS in the JSOC-like inversion is 'caused by a larger-than-expected smoothing in the horizontal direction' is plausible but not quantified; giving the horizontal widths or a smoothing comparison of the two kernels would strengthen the point.","section":"Section 4.1.1"},{"comment":"The near-zero or negative correlations between the JSOC sound-speed maps at 1–3 and 3–5 Mm are noted as unexpected, but no further investigation is offered. A brief comment on whether this could be tested with the authors' pipeline would help the reader interpret the table.","section":"Section 3.1, Table 2"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the horizontal-flow validation is a useful contribution. The main concern is that the abstract and conclusions overstate the sound-speed cross-talk result relative to what the data directly show; this is fixable either by obtaining the relevant JSOC kernels or by tempering the claim. I would be willing to reconsider after revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this paper compares the public JSOC time–distance products (RLS inversions) with an independent MC-SOLA pipeline and makes two claims. First, the JSOC averaging kernels for horizontal flows are not localized at the indicated depths—maps labeled 0–1, 1–3, and 3–5 Mm all see essentially the same near-surface layers. Second, JSOC sound-speed maps have higher RMS than the authors' equivalent inversions (18 vs 9 m/s at 2 Mm), which they attribute to cross-talk from flows. The first claim is well supported; the second is plausible but less secure than the abstract suggests.\n\nWhat the paper does well: it directly computes and displays the JSOC averaging kernels, giving the depth-smearing claim a quantitative anchor (mean depths all near 3 Mm for shallow targets). The horizontal-flow reproduction with correlations around 0.9 is a solid external benchmark—their pipeline can recreate the JSOC flow maps when given similar kernels and travel times. The three-tiered comparison (JSOC-like, JSOC-like target, JSOC-indicated target) is a clean way to separate the role of the target function from the role of the data. The authors are also honest about limitations, including that deep sound-speed inversions are not meaningful with 24h averaging.\n\nThe soft spots are real but not fatal. The cross-talk attribution is the weakest link: the paper only obtained JSOC averaging kernels for horizontal flows, not for sound speed. The RMS excess could in principle also come from differences in horizontal smoothing, noise levels, or vertical weighting of the sound-speed channel. The authors rely on a prior synthetic study (Korda & Švanda 2019) and the external expectation that quiet-Sun sound-speed perturbations are an order of magnitude below flows. That makes the story coherent, but it is not a direct measurement of the JSOC sound-speed cross-talk kernel. The abstract and conclusions state the cross-talk explanation more firmly than the data shown. Additionally, the map comparisons use a single 24-hour day and the RMS/correlation differences have no error bars. No code or data are shipped.\n\nWho this is for: anyone who uses the JSOC hmi.tdVinvrt_synopHC products for near-surface flow or sound-speed studies. The depth-mislocalization point alone is worth knowing, since it affects many published interpretations.\n\nRecommendation: send it to peer review. It is a careful, reproducible-in-spirit comparison with a load-bearing but addressable weakness. Ask the authors to either obtain the JSOC sound-speed averaging kernels (if feasible) or soften the abstract to say the sound-speed excess is 'consistent with cross-talk' rather than 'strongly affected by' it. With that revision, it would be a solid contribution.","headline":"A careful pipeline comparison that convincingly shows JSOC depth labels are misleading; the cross-talk explanation for the sound-speed amplitude excess is plausible but not directly tested.","tokens_in":28962,"tokens_out":2806,"would_cite":true,"duration_ms":28221,"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":"The publicly released JSOC sound-speed maps for the quiet Sun are inflated by contamination from the larger-amplitude horizontal flows, while their averaging kernels are not localised at the labelled depths.","keywords":["time-distance helioseismology","SOLA","regularised least squares","cross-talk","averaging kernels","sound-speed perturbations","horizontal flows","SDO/HMI"],"falsifier":"Because flow cross-talk should imprint the flow geometry on the sound-speed maps, one could measure the spatial correlation between the JSOC sound-speed maps and the divergence of the JSOC horizontal flow maps: a strong positive correlation would confirm the cross-talk mechanism, while a null correlation would leave the amplitude excess unexplained. Alternatively, run the JSOC-style regularised least-squares inversion on synthetic travel times with zero sound-speed perturbations and known flows; if it returns large sound-speed RMS, cross-talk is directly demonstrated.","tokens_in":27930,"feed_emoji":"☀️","tokens_out":7322,"duration_ms":66067,"temperature":0.7,"pith_summary":"This paper asks whether the routinely produced sound-speed and flow maps from the SDO/HMI time-distance helioseismology pipeline say what their depth labels claim. Comparing them with an independent inversion pipeline that can invert all quantities at once and explicitly minimise the leakage between them, the authors reproduce the horizontal flow maps well but find that the public sound-speed perturbation maps are inflated by cross-talk from the flows. They further show that the JSOC averaging kernels are not localised around the indicated target depths: maps labelled 0-1, 1-3, and 3-5 Mm mostly see the same near-surface layers down to about 4 Mm. If right, studies that use these products as depth-resolved sound-speed measurements overestimate amplitudes and misplace depth structure, while horizontal-flow studies are largely unaffected.","feed_headline":"Solar sound-speed maps are inflated by flow leakage","feed_subtitle":"Labeled depths in public sound-speed maps mix one near-surface layer, so amplitudes run high.","key_machinery":"The machinery is the multichannel subtractive optimally localised averaging (MC-SOLA) scheme, in which the inversion for horizontal flows, vertical flow, and sound-speed perturbations is performed at once and the cost function includes a term that minimises the integrals of the off-diagonal averaging kernels $K^\\alpha_\\beta$ for $\\alpha \\neq \\beta$, the cross-talk. The averaging kernels themselves, which connect the inverted estimate to the true subsurface quantities, are the diagnostic objects: their mean depth and vertical extent let the paper quantify how localised each product really is. A JSOC-style regularised least-squares inversion is the counterpoint: its cost function fits travel times and applies smoothing, but the shape of the averaging kernel never enters the solution, so cross-talk cannot be suppressed. Three inversion setups, JSOC-like, JSOC-like target, and JSOC-indicated target, isolate the effect of the target function from the effect of the travel-time set.","core_discovery":"The central claim is that the JSOC sound-speed perturbation products are strongly affected by a high level of cross-talk, which leads to larger amplitudes in the inversions, while the horizontal flow components are faithfully reproduced. Using the JSOC averaging kernels as targets, the independent SOLA pipeline recovers flow maps correlated with JSOC at 0.86-0.94 at shallow depths, but the sound-speed maps show RMS values around $9\\text{--}10~\\mathrm{m\\,s^{-1}}$ against $18~\\mathrm{m\\,s^{-1}}$ for JSOC, with the best correlation of 0.64 at 2 Mm depth. The authors attribute part of the excess to positively correlated contamination from the flows, consistent with their earlier synthetic-data finding that cross-talk can make up about half of an inverted sound-speed estimate. The same comparison shows that JSOC averaging kernels for the first three indicated depths all peak near 2 Mm and extend from the surface to roughly 4 Mm, so the depth labels in the public products do not correspond to the actual localisation; only the 5-7 Mm map reaches a mean sensitivity near 5.5 Mm.","pith_inferences":["If the true quiet-Sun sound-speed perturbations are closer to the upper range allowed by convection simulations, the JSOC excess would be partly signal; the cross-talk explanation could be tested by checking whether the excess amplitude scales with flow amplitude or flow divergence across active and quiet regions.","The same comparative protocol could be applied to other routinely produced local-helioseismic inversions, such as density or magnetic perturbations, where the regularised least-squares method is used without kernel constraints; cross-talk may be inflating those products too.","The depth-localisation figures suggest a practical rule for consumers: for a product to be called depth-resolved, the averaging kernel's mean depth should lie within the labelled bin and its width should be smaller than the bin spacing, a criterion the JSOC sound-speed and vertical-flow products do not meet.","An automated pipeline of the SOLA type could in principle be run at the same cadence as JSOC products, since the added degrees of freedom are a matter of cost-function design rather than new observations."],"forward_implications":["Users of the public JSOC sound-speed maps should treat the labelled depths as approximate and the amplitudes as upper limits: the maps mix the surface-to-4-Mm layer, and part of their RMS is contamination from the larger-amplitude flows.","Studies using JSOC flow maps for the horizontal components remain on firmer ground: correlations with the independent pipeline are high (0.86-0.94) at shallow depths, with comparable structure and amplitudes.","Apparent vertical coherence of near-surface flows across depth bins (correlations of 0.96-0.99 between the top three depth maps) is likely an artefact of the broad, poorly localised averaging kernels rather than evidence for physically coherent depth structure.","Deep sound-speed and vertical-flow inversions from 24-hour-averaged travel times cannot achieve signal-to-noise above unity; the depth labels for those products give a misleading impression of localisation.","Combining difference and mean travel-time geometries with ridge-filtered measurements in one inversion lowers both the noise and the cross-talk, so a routine product built this way would be more trustworthy."],"supporting_citations":[{"why":"Describes the JSOC regularised least-squares pipeline and noise estimates whose products are the object of the comparison.","marker":"Zhao et al. (2012)"},{"why":"Supplies the independent MC-SOLA pipeline and the synthetic-data evidence that cross-talk can form about half of an inverted sound-speed estimate.","marker":"Korda & Švanda (2019)"},{"why":"Earlier work suggesting biases and poor depth sensitivity in the JSOC products, which this study verifies with averaging kernels.","marker":"DeGrave & Jackiewicz (2015)"},{"why":"Magnetoconvection model used as the reference expectation that quiet-Sun sound-speed perturbations are an order of magnitude below horizontal flow amplitudes.","marker":"Rempel (2014)"},{"why":"Second model-based reference for the expected small amplitude of quiet-Sun sound-speed perturbations.","marker":"DeGrave et al. (2014)"},{"why":"Validation of the SOLA-type inversion and the result that 24-hour-averaged travel times support signal-to-noise above unity only for the shallowest depths.","marker":"Švanda et al. (2011)"},{"why":"Defines the multichannel SOLA formulation used in the independent pipeline.","marker":"Jackiewicz et al. (2012)"},{"why":"Provides the Born-approximation sensitivity kernels used for the travel-time forward model in the comparisons.","marker":"Birch & Gizon (2007)"},{"why":"Model S sound-speed profile used to convert JSOC fractional sound-speed perturbations into velocity units before comparison.","marker":"Christensen-Dalsgaard et al. (1996)"}],"fun_headline_variants":["Sound-speed depth labels mislead; cross-talk inflates amplitudes","Solar sound-speed maps: depths wrong, amplitudes high","Flow maps reliable, but sound-speed maps inflated by cross-talk","Cross-talk inflates sound-speed, mislabels depths"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that quiet-Sun sound-speed perturbations are genuinely small, roughly an order of magnitude below the horizontal flow amplitudes, as magnetoconvection models predict; if real near-surface sound-speed perturbations are larger than these models say, part or all of the excess amplitude in the JSOC maps could be real signal rather than cross-talk.","fun_headline_variants_meta":{"raw":{"variants":["Sound-speed depth labels mislead; cross-talk inflates amplitudes","Solar sound-speed maps: depths wrong, amplitudes high","Flow maps reliable, but sound-speed maps inflated by cross-talk","Cross-talk inflates sound-speed, mislabels depths"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001442,"raw_usage":{"total_tokens":5849,"prompt_tokens":1024,"completion_tokens":4825,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":640,"completion_tokens_details":{"reasoning_tokens":4758}},"tokens_in":640,"tokens_out":4825,"duration_ms":32481,"temperature":1.0,"reasoning_tokens":4758,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:56:43.608786+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Because flow cross-talk should imprint the flow geometry on the sound-speed maps, one could measure the spatial correlation between the JSOC sound-speed maps and the divergence of the JSOC horizontal flow maps: a strong positive correlation would confirm the cross-talk mechanism, while a null correlation would leave the amplitude excess unexplained. Alternatively, run the JSOC-style regularised least-squares inversion on synthetic travel times with zero sound-speed perturbations and known flows; if it returns large sound-speed RMS, cross-talk is directly demonstrated.","supporting_citations":[{"cited_title":"S., et al","cited_arxiv_id":null,"evidence_quote":"Describes the JSOC regularised least-squares pipeline and noise estimates whose products are the object of the comparison."},{"cited_title":"& S vanda , M","cited_arxiv_id":null,"evidence_quote":"Supplies the independent MC-SOLA pipeline and the synthetic-data evidence that cross-talk can form about half of an inverted sound-speed estimate."},{"cited_title":"& Jackiewicz , J","cited_arxiv_id":null,"evidence_quote":"Earlier work suggesting biases and poor depth sensitivity in the JSOC products, which this study verifies with averaging kernels."},{"cited_title":"2014, , 788, 127","cited_arxiv_id":null,"evidence_quote":"Second model-based reference for the expected small amplitude of quiet-Sun sound-speed perturbations."},{"cited_title":"C., Gizon , L., et al","cited_arxiv_id":null,"evidence_quote":"Defines the multichannel SOLA formulation used in the independent pipeline."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Born-approximation sensitivity kernels used for the travel-time forward model in the comparisons."}],"review_version":1}