{"id":"3e8fbcd3-ca22-4ff9-be10-76fca0e09f03","arxiv_id":"2506.23693","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Using eight years of HARPS-N Sun-as-a-star radial velocities, the authors measure a clear variation in the supergranulation timescale over the solar cycle, strongly anti-correlated with the sunspot number.","lead":"This paper measures how the Sun's supergranulation pattern, a source of noise in planet-hunting radial velocity measurements, changed across most of an 11-year activity cycle. It finds that the characteristic timescale of supergranulation lengthens near activity minimum and shrinks as the sunspot number rises.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reported 8–14 day residuals are the weakest link: Appendix B injects a single fixed-amplitude activity signal, not one whose amplitude tracks the sunspot cycle, so a two-SHO GP could absorb activity-dependent residual power and mimic the tau_SG–SSN anti-correlation.","rationale":"The reader's CONDITIONAL verdict is appropriate, and its weakest_assumption identifies the same load-bearing point: the two-component SHO model must map the true supergranulation timescale to the fitted tau_SG without activity-dependent bias. I have sharpened this into a concrete, testable gap. The paper's own validation in Appendices A and B covers reduced daily sampling and one fixed-amplitude activity signal, but not the regime that would produce the observed anti-correlation as an artifact: activity residuals whose amplitude scales with sunspot number. The paper's positive evidence—two correction methods, small Spearman p-values, stable granulation parameters, constant jitter—means the physical claim may well be true, and the two methods agreeing is a genuine point in its favor. But before the result is used to motivate survey design or non-stationary noise models, the SSN-scaled null simulation is a cheap, decisive check. Because this is exactly the gap the reader flagged and because the verdict is already CONDITIONAL, my read does not move the verdict. I would not REJECT the paper on this basis, since the concern is empirical and testable rather than a demonstrated internal contradiction.","tokens_in":23921,"tokens_out":8277,"duration_ms":100390,"concrete_test":"Run the identical 12-week-chunk GP pipeline on synthetic quiet-Sun RVs built from the actual HARPS-N time stamps with constant tau_SG (whole-cycle mean), constant granulation and white-noise parameters, plus a quasi-periodic residual at 13.5 d (and optionally 8.4 d) whose amplitude in each chunk is proportional to the observed monthly sunspot number, normalized to the residual power in Figure 3 at cycle maximum. Repeat for 100 realizations, apply the same 100-point cut and outlier rejection, and compare the distribution of Spearman rho and p-values against the observed values (p<1e-2 SDO, p<1e-4 YARARA). If the null model produces comparable anti-correlations, the central claim is not supported; if recovered tau_SG remains flat, the concern is refuted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.3's central claim—that tau_SG anti-correlates with sunspot number in both SDO- and YARARA-corrected quiet-Sun RVs—rests on the fitted SHO timescale being an unbiased estimator of the physical supergranulation timescale. Section 3 fixes the model to exactly two aperiodic SHO kernels plus white noise; Section 2.2 reports that both corrected data sets retain residual power at Prot/2 (~14 d) and Prot/3 (~8 d), so the actual residuals are not a stationary two-component process. Appendix B injects one quasi-periodic 13.5-day signal into one simulated year and finds no bias, but it does not scale the amplitude with the solar cycle, does not reproduce the 4-week/12-week chunking and outlier rejection, and does not vary the activity-state. If the 8–14 day residual amplitude grows with SSN, the aperiodic SHO kernel can trade variance between its long-timescale tail and the white-noise term; the 'jitter is constant at 0.20–0.25 m/s' check does not test whether the SHO parameters themselves move. Agreement between SDO and YARARA is reassuring but not decisive, because both time series exhibit the same 14-day residuals and both start from the same HARPS-N spectra. The anti-correlation could therefore be a byproduct of the model absorbing activity-dependent residual power rather than a physical cycle in supergranulation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses eight years of HARPS-N Sun-as-a-star radial velocities (July 2015 to November 2023) to measure the granulation and supergranulation properties of the quiet Sun after subtracting active-region contributions with two independent methods: an SDO/HMI image-based correction and the YARARA spectral decorrelation. The quiet-Sun RVs are modelled in 4-week, 12-week, and 1-year chunks with a two-component aperiodic simple-harmonic-oscillator Gaussian process plus white noise (Section 3). The paper reports that the supergranulation timescale tau_SG is longest at solar minimum and is strongly anti-correlated with the SILSO relative sunspot number, with Spearman p-values below 1e-2 for SDO and below 1e-4 for YARARA for 3-month chunks, while granulation parameters remain stable. It also uses a Fisher-information approach to argue that a roughly 23-night campaign with rapid target switching can constrain tau_SG to 30% precision for up to 10 stars, and it discusses plausible physical explanations for a supergranulation cycle.","tokens_in":24216,"tokens_out":8921,"duration_ms":98164,"significance":"If the central claim holds, the supergranulation timescale in disk-integrated solar RVs is not stationary over the magnetic cycle, with direct consequences for modelling stellar noise in EPRV surveys and for the interpretation of granulation-driven RV variability. The study's strengths are the use of two independent activity corrections, the explicit reporting of correlation p-values, and the effort to validate the GP model with injection-recovery tests in the appendices. The observing-strategy analysis is concrete and falsifiable, and it provides a useful planning estimate for future campaigns. However, the physical interpretation rests on the assumption that the fitted SHO timescale is an unbiased estimator of the true supergranulation timescale across the activity cycle, and the current validation does not yet establish this. The paper is therefore a potentially valuable contribution whose central claim needs additional support before it can be accepted.","major_comments":[{"comment":"The abstract and the discussion overstate the amplitude of the cycle variation. Table 2 gives mean tau_SG at solar minimum of 1.27 +/- 0.13 d (YARARA) and 1.60 +/- 0.16 d (SDO), versus 0.61 +/- 0.07 d and 0.80 +/- 0.10 d at the beginning of Cycle 25, i.e. a factor of about 2, not an order of magnitude. The order-of-magnitude range appears only in the 4-week-chunk scatter discussed in Section 4.2, which is not the cycle trend; the Discussion's 'factor of three' also exceeds the Table 2 ratios. Please harmonize the abstract, Section 4.3, and Section 6 with the quantitative results.","section":"Abstract, §4.3, §6"},{"comment":"The injection-recovery test does not rule out activity-dependent contamination of tau_SG. Section 2.2 reports residual power at Prot/2 ~14 d and Prot/3 ~8 d in both corrected datasets, so the actual residuals are not a stationary two-component aperiodic process. Appendix B injects a single quasi-periodic 13.5-day signal at fixed amplitude into one simulated year and finds no bias in the fitted parameters, but it does not scale the injected amplitude with sunspot number, does not reproduce the 4-week/12-week chunking, outlier rejection, and gap patterns, and does not cover the range of activity states present in the real data. If the 8-14 d residual power grows with SSN, the aperiodic SHO kernel and the white-noise term can trade variance, shifting tau_SG; the constant 0.20-0.25 m/s jitter check in Section 6 does not test whether the SHO parameters themselves move. Please add multi-amplitude injection-recovery tests spanning the observed SSN range with the actual chunking and outlier rejection, or otherwise demonstrate that tau_SG is an unbiased estimator of the supergranulation timescale throughout the cycle.","section":"§4.3, Appendix B"},{"comment":"The model assumes that the quiet-Sun residuals are exactly described by two independent aperiodic SHO kernels with Q=1/sqrt(2) plus white noise, and the fitted tau_SG is interpreted as the physical supergranulation timescale. This assumption is load-bearing. The agreement between the SDO and YARARA corrections is reassuring but not decisive: both corrections start from the same HARPS-N spectra, both residual sets contain the same 14-day peaks, and the authors themselves note in Section 4.1 that the YARARA S-index decorrelation may remove part of the supergranulation signal. The paper should either validate the two-SHO model against injected signals with activity levels representative of all cycle phases, or test an alternative model that includes a quasi-periodic term and show that the tau_SG-SSN anti-correlation is robust to the model choice.","section":"§3 (Eqs. 6-7), §4.1"}],"minor_comments":[{"comment":"The sentence containing 'weather the activity-correction was done with YARARA or SDO' should read 'whether'; the typo appears in the discussion of the 14-day and 8-day activity residuals.","section":"§2.2"},{"comment":"The caption states that the injected activity signal has period = 13.5 hours, while the text says 13.5 days; please correct the inconsistency.","section":"Appendix B, Fig. B1 caption"},{"comment":"The Al Moulla et al. row lists sigma_G and sigma_SG without uncertainties; please state whether these are central values without published errors or add the uncertainties.","section":"Table 2"},{"comment":"The statement that the 4-week-chunk scatter 'is real' and not an artefact of the limited chunk duration is supported only by a reference to O'Sullivan & Aigrain (2024); please show the relevant injection-recovery result or temper the wording.","section":"§4.2"},{"comment":"The displayed derivative of the SHO kernel with respect to omega appears garbled in the typeset version; please verify the equation and its sign structure.","section":"Eq. (13)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope and I do not see citation-pattern concerns. The main risk is that the physical interpretation as a supergranulation cycle is under-validated: the empirical anti-correlation may be robust, but the current injection-recovery tests do not eliminate the possibility that activity-dependent residual power at 8-14 days is absorbed by the aperiodic SHO model. I would encourage the editor to require the extended activity-residual injection tests described in major comment 2 before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth reading. The genuine new result is the first GP-based, disk-integrated measurement of the supergranulation timescale across most of a solar cycle, and the fact that two very different activity corrections—SDO image subtraction and YARARA spectral decorrelation—give the same qualitative answer: tau_SG is longest near minimum and anti-correlates with sunspot number (Spearman p < 1e-2 for SDO, < 1e-4 for YARARA). That cross-method agreement is the paper's strongest evidence, and it is real. The paper also does the right things around it: KS tests on parameter distributions, shifting chunk boundaries, comparing structure functions to Lakeland et al. (2024) to address the apparent contradiction with that work, and a Fisher-information observing strategy analysis that gives concrete, plausible survey guidance (23 nights to characterize tau_SG to 30% for 10 stars). The observational strategy section is a solid contribution in its own right.\n\nThe soft spots are in the central claim. The abstract says \"order of magnitude variation\" in the timescale, but Table 2 shows a factor 2–3 between cycle minimum and active phases; the order-of-magnitude swings appear in the 4-week chunk scatter, which the paper itself attributes to real short-term variability, not the cycle. That overstatement should be fixed. More importantly, the validation in Appendix B injects a single fixed-amplitude 13.5-day activity signal into one simulated year. It does not scale that amplitude with sunspot number, even though the real corrected data retain residual power at 14 and 8 days in both pipelines. If those residuals grow with activity—which is plausible—the aperiodic SHO kernels can absorb them and shift tau_SG without the white-noise term moving much. The \"jitter is constant\" check does not test whether the SHO parameters themselves are biased. The SDO/YARARA agreement is reassuring but not decisive, since both start from the same HARPS-N spectra and both show the same residuals. I would not call the result wrong, but the load-bearing validation is thinner than the claim requires. Data and code are not public yet (HARPS-N RVs are described in Dumusque et al., submitted), so the p-values cannot be checked independently at this point.\n\nBottom line: a promising, well-written paper with a thought-provoking result and a genuinely useful survey-planning section. The cycle-anti-correlation is probably real, but the abstract overstates the size and the injection-recovery test should be expanded to cycle-dependent residual amplitudes before the result is used to set survey strategy. I would send it to review; the referees should push on Appendix B.","headline":"Two independent activity corrections agree that the supergranulation timescale in disk-integrated solar RVs anti-correlates with the cycle, but the abstract overstates the amplitude and the injection-recovery test is too thin to fully rule out a modeling artifact.","tokens_in":24862,"tokens_out":2377,"would_cite":true,"duration_ms":25185,"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":"The quiet Sun's supergranulation timescale varies by up to an order of magnitude across the solar cycle, peaking at activity minimum and anti-correlating with sunspot number.","keywords":["supergranulation","solar radial velocities","solar activity cycle","Gaussian process regression","Sun-as-a-star","HARPS-N","stellar noise","exoplanet radial velocity surveys"],"falsifier":"Re-fit the same 2015-2023 HARPS-N quiet-Sun RVs with a Harvey-style kernel with a free power-law index, or with a quasi-periodic component added, and check whether the $\\tau_{\\mathrm{SG}}$ anti-correlation with sunspot number survives; alternatively, measure supergranule lifetimes from resolved SDO/HMI Dopplergrams over the same cycle and compare their evolution with the disk-integrated $\\tau_{\\mathrm{SG}}$.","tokens_in":23687,"feed_emoji":"☀️","tokens_out":6318,"duration_ms":61858,"temperature":0.7,"pith_summary":"This paper tries to establish that supergranulation, the Sun's large-scale horizontal convection pattern, produces a radial-velocity noise component whose characteristic timescale is not fixed but changes by roughly an order of magnitude over the 11-year activity cycle, with the longest timescale at activity minimum. Using eight years of Sun-as-a-star observations from HARPS-N, the authors subtract the contribution of magnetically active regions in two independent ways, one image-based and one purely spectroscopic, and model the residual quiet-Sun radial velocities as the sum of two stochastic processes: granulation and supergranulation. Granulation parameters stay stable, but the supergranulation timescale is strongly anti-correlated with the relative sunspot number, and this anti-correlation appears with both correction methods. If true, stellar noise in exoplanet radial-velocity surveys is non-stationary over activity-cycle timescales, and treating it as a fixed, unchanging process needs revision.","feed_headline":"Supergranulation timescale swings with the solar cycle","feed_subtitle":"Quiet-Sun RV noise is not stationary: its characteristic timescale anti-correlates with sunspot number, complicating Earth-twin searches.","key_machinery":"The argument is carried by a two-component Gaussian Process with aperiodic simple-harmonic-oscillator kernels (quality factor $Q=1/\\sqrt{2}$), fitted independently to 4-week chunks of quiet-Sun radial velocities, with the supergranulation component forced to have a lower frequency than the granulation component. For each chunk the fit yields amplitude $S_0$ and angular frequency $\\omega_0$ for both components, converted to timescale $\\tau=2\\pi/\\omega_0$ and standard deviation; the resulting timescales are then correlated with the mean relative sunspot number over the same chunk. A Fisher-information calculation, based on derivatives of the covariance matrix with respect to $S_0$ and $\\omega_0$, translates different observing cadences into the expected fractional uncertainty on the supergranulation timescale, which is used to predict how many nights are needed to characterise supergranulation in other stars.","core_discovery":"The central discovery is that the disk-integrated quiet-Sun radial-velocity signal attributed to supergranulation carries a cycle-dependent timescale: $\\tau_{\\mathrm{SG}}$ is longest at solar minimum, drops by a factor of roughly three during the rising and decaying phases, and anti-correlates with the relative sunspot number ($p<10^{-2}$ for the image-corrected data, $p<10^{-4}$ for the spectroscopically corrected data at 3-month chunks). The authors interpret this as a physical property of supergranulation rather than an artefact, because the two correction methods have limitations that would pull in opposite directions, and because their injection tests for reduced daily sampling and for a 13.5-day activity residual do not reproduce the effect. They also find that the supergranulation timescale measured from 4-week chunks shows large scatter that disappears at 3-month and 1-year chunk lengths, and they reconcile their results with the earlier structure-function analysis by converting their Gaussian-process covariance into structure functions.","pith_inferences":["A testable consequence the paper leaves implicit: if $\\tau_{\\mathrm{SG}}$ is activity-dependent, then applying the same Gaussian-process analysis to long-baseline radial-velocity time series of other Sun-like stars, alongside an activity indicator such as the S-index, should reveal a similar anti-correlation between the fitted convective timescale and activity level.","The factor-of-two change in timescale is larger than the roughly 10% changes in supergranule size reported in imaging studies, suggesting that the disk-integrated timescale may be modulated by something other than cell size alone, perhaps the distribution of cell lifetimes or the magnetic network; resolved-imaging lifetime measurements over a full cycle would be a discriminating test.","Because the spectroscopic correction linearly decorrelates against the S-index, part of the apparent cycle could in principle be caused by activity-dependent removal of supergranulation; the agreement with the image-based correction argues against this, but injecting activity-correlated supergranulation signals into the pipeline would settle it."],"forward_implications":["Radial-velocity noise models for Sun-like stars must allow the supergranulation timescale to vary with activity state; a fixed kernel will misestimate the covariance on multi-year baselines.","Exoplanet surveys that deliberately target low-activity stars to reduce spot and facula signals may unintentionally encounter a stronger, longer-timescale supergranulation signal.","Characterising supergranulation in other stars requires dedicated campaigns of at least about 23 nights, but rapid switching between targets can monitor up to 10 stars at once to roughly 30% precision in $\\tau_{\\mathrm{SG}}$.","Granulation parameters remain stable over the cycle, so the granulation component can be calibrated once and treated as stationary, while the supergranulation component cannot.","The effect disappears when the data are analysed with daily sampling or structure functions, so non-parametric methods on sparsely sampled data can miss the cycle dependence."],"supporting_citations":[{"why":"Supplies the two-component aperiodic-SHO Gaussian Process method and the injection-recovery validation used to measure $\\tau_{\\mathrm{SG}}$.","marker":"O'Sullivan & Aigrain (2024)"},{"why":"Provides the SDO-corrected quiet-Sun RV dataset and the structure-function analysis whose non-detection is reconciled here.","marker":"Lakeland et al. (2024)"},{"why":"Gives the literature Harvey-model supergranulation parameter values the results are compared against.","marker":"Al Moulla et al. (2023)"},{"why":"Defines the magnetogram threshold and the two-component active-region RV model used for the image-based activity correction.","marker":"Haywood et al. (2016)"},{"why":"Calibrates the image-based active-region subtraction on HARPS-N solar RVs.","marker":"Milbourne et al. (2019)"},{"why":"Extends and applies the image-based activity-correction procedure used here.","marker":"Haywood et al. (2022)"},{"why":"Introduces the spectroscopic post-processing and S-index decorrelation used for the second quiet-Sun dataset.","marker":"Cretignier et al. (2021)"},{"why":"Provides the SILSO relative sunspot numbers used for the correlation analysis.","marker":"Clette & Lefèvre (2015)"},{"why":"Supplies the Fisher-information formalism used to predict observing-campaign requirements.","marker":"Gupta & Bedell (2024)"}],"fun_headline_variants":["Supergranulation timescale anti-correlates with sunspots","Sun's supergranulation rhythm slows at activity minimum","Supergranulation cycle throws off Earth-twin RV searches","Sun's granulation timescale swings over 11-year cycle","Quiet-Sun RV noise timescale varies with solar activity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that after active regions are removed, the quiet-Sun radial velocities are exactly the sum of two independent aperiodic SHO processes with fixed $Q=1/\\sqrt{2}$, so that the fitted supergranulation timescale is an unbiased measure of the true physical timescale rather than a product of the model or of the activity correction.","fun_headline_variants_meta":{"raw":{"variants":["Supergranulation timescale anti-correlates with sunspots","Sun's supergranulation rhythm slows at activity minimum","Supergranulation cycle throws off Earth-twin RV searches","Sun's granulation timescale swings over 11-year cycle","Quiet-Sun RV noise timescale varies with solar activity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000176,"raw_usage":{"total_tokens":1280,"prompt_tokens":927,"completion_tokens":353,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":543,"completion_tokens_details":{"reasoning_tokens":270}},"tokens_in":543,"tokens_out":353,"duration_ms":4384,"temperature":1.0,"reasoning_tokens":270,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:34:26.167501+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-fit the same 2015-2023 HARPS-N quiet-Sun RVs with a Harvey-style kernel with a free power-law index, or with a quasi-periodic component added, and check whether the $\\tau_{\\mathrm{SG}}$ anti-correlation with sunspot number survives; alternatively, measure supergranule lifetimes from resolved SDO/HMI Dopplergrams over the same cycle and compare their evolution with the disk-integrated $\\tau_{\\mathrm{SG}}$.","supporting_citations":[],"review_version":1}