{"id":"8d86e901-7f54-4f3d-956e-242e82c80241","arxiv_id":"2505.04511","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"The horizontal advection speed of solar magnetic flux decreases smoothly with increasing vertical field strength, from roughly 110 m/s at 150 G to 10 m/s at 2500 G, and is fit by a fourth-degree polynomial.","lead":"This paper measures how fast magnetic flux moves horizontally across the Sun's surface as a function of magnetic field strength, using six active regions. It finds that flux advection slows from about 110 m/s in weak-field plage to about 10 m/s in sunspot umbrae, confirming that strong fields suppress convection.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"FLCT at 12-minute cadence cannot resolve the ~0.02-pixel displacements in strong fields; Eq. (1)'s absolute speeds are cadence- and kernel-dependent and lack a noise-floor calibration.","rationale":"The paper aims to quantify the suppression of horizontal flux advection with increasing vertical field strength, and the headline result is the polynomial in Eq. (1). For that claim to hold, FLCT must measure advection speeds accurately at the 10-100 m/s level. This is the least secure part of the argument because the implied per-frame displacements are 0.02-0.22 pixels, which is at or below the typical sensitivity of correlation tracking, and because the Bz data are temporally smoothed. The authors' own cadence comparison (Fig. 9) shows that 45-second BLOS speeds are roughly two to three times larger than 12-minute Bz speeds, so the absolute values in Eq. (1) are method-dependent. The trend direction is robust: all six active regions show a monotonic decrease, the BLOS data reproduce the trend, and the result is consistent with Title et al. (1992). This is therefore not a reason to reject the paper, but it is a reason to withhold full acceptance until the tracking calibration is demonstrated. A synthetic FLCT test with known advection velocities would directly settle whether the low speeds in strong fields are real or an artifact of cadence, smoothing, and subpixel tracking. The reader's weakest assumption identifies the same missing noise-floor analysis, and I agree with that assessment. The CONDITIONAL verdict is appropriate and should remain unchanged.","tokens_in":15330,"tokens_out":3124,"duration_ms":33410,"concrete_test":"Run FLCT on synthetic SHARP-like Bz sequences: take a real quiet/plage Bz map, impose known uniform horizontal advection velocities v_true from 0 to 300 m/s, add HMI noise of ~70 G, apply the same 1215-second boxcar temporal averaging and 12-minute sampling, and measure v_recovered with the same 15-pixel FWHM kernel. Plot v_recovered versus v_true for each Bz bin. If v_recovered at 10 m/s is not significantly different from zero, or is biased by more than ~20% at 100 m/s, then Eq. (1) should be recalibrated or explicitly presented as a cadence-dependent proxy rather than as the true advection speed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim, Eq. (1), gives 10 ± 4 m/s at 2500 G and 110 ± 3 m/s at 150 G. At the stated HMI CEA scale of 360 km/pixel and 12-minute cadence, 10 m/s corresponds to 7.2 km, i.e., 0.02 pixels per frame, and even 110 m/s is only 0.22 pixels. The paper selects a 15-pixel FWHM Gaussian kernel in Section 2.2 based on visual persistence, not on calibration against known displacements. In addition, the Bz data are temporally averaged with a 1215-second boxcar and an FWHM of 720 seconds, as noted in Appendix .1, which smooths short-lived motions and can bias 12-minute velocities downward. The authors' own BLOS comparison in Fig. 9 shows that measured speeds depend strongly on cadence: 45-second BLOS speeds are roughly two to three times larger than 12-minute Bz speeds. Thus Eq. (1) is not an intrinsic physical relation unless the tracking method is calibrated, and the error bars in Figs. 3 and 6 are data spreads rather than tracking uncertainties. No synthetic test or noise-floor analysis establishes the minimum detectable velocity or the bias at 0.02-pixel displacements. The qualitative trend of decreasing speed with increasing field strength is consistent across six active regions and with prior work, so the concern is not about the direction of the trend but about the quantitative calibration that the abstract and Eq. (1) present as the main result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper measures the horizontal advection speed of magnetic flux as a function of vertical field strength (Bz) in six active regions using Fourier Local Correlation Tracking (FLCT) on HMI SHARP Bz magnetograms at 12-minute cadence. The authors report a monotonic decrease in the average advection speed from about 110 m/s at 150 G to about 10 m/s at 2500 G, fit a fourth-degree polynomial (Eq. 1), and compare results with BLOS magnetograms at several cadences. They interpret the trend as quantitative confirmation that strong magnetic fields suppress convective advection of flux and propose the relation as an empirical input for future MHD models of coronal heating.","tokens_in":15663,"tokens_out":5470,"duration_ms":45064,"significance":"If the quantitative relation were well calibrated, this would be the first uniform measurement of flux advection speed across the full range from network/plage to sunspot umbrae, and it would provide a useful empirical input for coronal heating models. Strengths of the study include the use of six stable, non-flaring active regions, consistent data processing, and cross-checks against BLOS data at multiple cadences and against earlier results (Title et al. 1992; Sobotka et al. 2012). The qualitative trend—decreasing speed with increasing field strength—is robust across all six ARs and agrees with prior work. However, the absolute speed scale and therefore the polynomial in Eq. (1) are not yet established as physical, owing to the missing calibration of FLCT at sub-pixel displacements and the known cadence and temporal-smoothing dependencies.","major_comments":[{"comment":"The paper provides no noise-floor or minimum-detectable-velocity analysis for FLCT. At the stated CEA scale of 360 km/pixel and 12-minute cadence, the quoted 10 m/s at 2500 G corresponds to a displacement of about 0.02 pixels per frame, and even 110 m/s at 150 G is only 0.22 pixels. Without synthetic tests or calibration against known displacements, the absolute speeds in Eq. (1) and in the abstract (110 ± 3 and 10 ± 4 m/s) cannot be taken as physical advection speeds; the error bars in Figs. 3 and 6 are spreads of the data, not tracking uncertainties. Since Eq. (1) is the central quantitative result, this gap is load-bearing.","section":"Section 2.2, Eq. (1)"},{"comment":"The authors' own comparison shows that 45-second BLOS speeds are two to three times larger than 12-minute Bz speeds, and Section 2.2 states that the 15-pixel kernel was chosen by visual persistence rather than calibration. Because the paper claims in Section 1 to 'establish a general relation,' Eq. (1) as presented is not invariant to the measurement settings. The paper should either calibrate the speeds to make them independent of cadence and kernel, or explicitly present Eq. (1) as a measurement-specific relation with clear caveats about how the absolute values would change.","section":"Section 4, Fig. 9"},{"comment":"The Bz magnetograms used in the main analysis are temporally averaged with a 1215-second boxcar and a cosine weighting function with an FWHM of 720 seconds. This smoothing is likely to suppress short-lived motions and bias the measured 12-minute velocities downward. The paper does not quantify this bias or its impact on Eq. (1). Because the main analysis is based on these smoothed maps, the polynomial in Eq. (1) may underestimate the true horizontal advection speed even at 12-minute cadence.","section":"Appendix .1"}],"minor_comments":[{"comment":"The acronym 'FCLT' should be 'FLCT', and 'Guassian' should be 'Gaussian'.","section":"Section 2.2"},{"comment":"The red error bars, described as standard error, are mostly hidden inside the diamond markers; consider using caps or a different marker style so that the small standard errors are visible to the reader.","section":"Figure 3 caption"},{"comment":"The figure caption should explicitly state that the blue error bars are divided by four for display, rather than leaving this information only in the main text.","section":"Section 3.3, Fig. 7"},{"comment":"The appendix headings appear as '.1' and '.2' instead of 'A.1' and 'A.2'; please fix the formatting.","section":"Appendix headings"},{"comment":"The 'Max Flare' entry for AR 12108 is '–'; using 'None' would be consistent with the text and clearer to readers.","section":"Table 1"},{"comment":"Reporting uncertainties for the polynomial coefficients would allow readers to assess the fit quality and would strengthen the quantitative claim.","section":"Eq. (1)"}],"recommendation":"major_revision","confidential_remarks":"The central quantitative claim (Eq. 1) is not yet supported because the FLCT measurements lack calibration at sub-pixel displacements and the absolute speeds are demonstrably cadence- and kernel-dependent. The qualitative trend is consistent and valuable, but the paper needs either a noise-floor calibration and synthetic tests, or a reframing of Eq. (1) as a measurement-dependent fit. I recommend major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nTwo things to know. This is the first uniform, multi-active-region measurement of horizontal flux advection speed versus vertical field strength from 150 to 2500 G, using one instrument, one cadence, one tracker, including sunspot umbrae. And the qualitative finding—advection slows steadily as field strength increases—is solid and consistent across six ARs and with earlier work.\n\nThe paper does several things well. The AR selection is careful (non-flaring, within 30 deg of disk center, 24h of SHARP Bz). The trend is cross-checked against BLOS at 45s, 3, 6, and 12 min cadences, and against prior results. The spatial structure shown in Fig. 5 and the radial profile across a sunspot in Fig. 7 give confidence that the suppression is real, not an artifact of one region. The prose is clear about the method's limits, including the cadence dependence of absolute speeds.\n\nThe soft spot is the quantitative polynomial, Eq. (1). At 10 m/s and 12-min cadence with 360 km/pixel, the per-frame displacement is about 0.02 pixels. The paper gives no noise-floor or minimum-detectable-velocity analysis for FLCT at these speeds. The 15-pixel kernel is chosen for visual persistence, not calibrated against known displacements. And the authors' own Fig. 9 shows that 45-s speeds are two to three times higher than 12-min speeds. So Eq. (1) describes what FLCT returns at this particular cadence and kernel, not an intrinsic physical relation. The error bars in Figs. 3 and 6 are data spreads, not tracking uncertainties, and the polynomial coefficients are fit without uncertainties. That does not undermine the qualitative trend, which is strong. But it does mean the numeric curve should be treated as provisional, with a calibration step before it is used in MHD boundary conditions.\n\nA few minor points: the BLOS curves at 45-s show a bump at high field strengths that the authors attribute to p-modes; that is plausible but not tested. The appendix on correlation analysis is useful but does not address the noise floor. The paper would be stronger with a synthetic-shift test or a minimum-velocity estimate.\n\nWho is this for? Solar observers and modelers who need a parameterization of flux advection suppression. They will get a well-documented empirical trend, but they should not hard-code Eq. (1) without calibration. I would bring it to a reading group as a good example of tracking-method limits, and I would cite the qualitative trend. A serious referee should engage with it; the revisions I'd ask for are a noise-floor calibration, uncertainty on the fit coefficients, and a more prominent caveat that the absolute speeds are cadence-dependent.\n\nRecommendation: send to peer review with requested revisions.","headline":"A solid, well-documented first uniform measurement of flux advection suppression, but the headline polynomial's absolute speeds are cadence- and kernel-dependent and need a noise-floor calibration before use in MHD models.","tokens_in":16221,"tokens_out":3043,"would_cite":true,"duration_ms":26573,"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 Sun's magnetic flux advection slows from 110 to 10 m/s as vertical field strength rises from 150 to 2500 G.","keywords":["solar photosphere","solar magnetic fields","solar coronal heating","magnetic flux advection","local correlation tracking","FLCT","magnetoconvection","sunspots"],"falsifier":"Re-run the FLCT analysis on the same magnetograms but with a 9-pixel or 6-pixel kernel and with 6-minute cadence; if the mean speeds at 150-200 G do not stay within the reported 110 ± 3 m/s, or if the curve's shape changes materially, the specific polynomial is an artifact of the kernel and cadence choices rather than a robust physical relation.","tokens_in":15152,"feed_emoji":"☀️","tokens_out":6300,"duration_ms":50019,"temperature":0.7,"pith_summary":"This paper tries to establish a quantitative, empirical relation between the strength of the vertical magnetic field and the speed at which magnetic flux is advected horizontally across the solar surface. Using Fourier Local Correlation Tracking on 24 hours of HMI SHARP magnetograms for each of six active regions, it finds that the average advection speed declines steadily from about 110 m/s at 150 G to about 10 m/s at 2500 G, a trend fit by a fourth-degree polynomial. This matters for coronal heating because convection-driven shuffling of magnetic footpoints is thought to heat the corona, and the new curve quantifies how strongly that shuffling is suppressed as field strength grows. If correct, the relation gives modelers a direct observational input for magneto-convection simulations and for scaling laws of coronal loop heating.","feed_headline":"Sun's flux advection slows from 110 to 10 m/s","feed_subtitle":"Six active regions, 24 hours of magnetograms: a quantitative curve linking field strength to flux advection speed.","key_machinery":"The central tool is Fourier Local Correlation Tracking (FLCT), applied to 12-minute-cadence HMI SHARP Bz magnetograms. FLCT measures the horizontal displacement of magnetic-flux patterns between consecutive frames by cross-correlating Gaussian-weighted subimages (here with a 15-pixel FWHM kernel spanning ~5.4 Mm, about five granules), then converts the displacement into a speed. The argument is carried by binning measured speeds in 50 G-wide Bz bins, averaging over 24 hours per active region, and fitting the mean curve with a fourth-degree polynomial. The Bz threshold of 150 G (~2σ noise) defines the weak-field endpoint, and the choice of Bz rather than BLOS or |B_total| is justified by lower noise and by Bz being the vertical magnetic flux density.","core_discovery":"The paper reports that horizontal advection of magnetic flux, measured with FLCT on 12-minute-cadence HMI SHARP Bz magnetograms of six non-flaring active regions, decreases monotonically with increasing vertical field strength Bz. Mean speeds fall from 110 ± 3 m/s in the 150–200 G bin (network and plage) to 10 ± 4 m/s in the 2400–2500 G bin (sunspot umbra), with a plateau near 1200–2000 G and a slight bump before the drop to near-zero speeds at the strongest fields. The combined mean is fit by $v_h = -1.55\\times10^{-14} x^4 + 7.06\\times10^{-11} x^3 - 7.97\\times10^{-8} x^2 - 3.92\\times10^{-5} x + 0.11$ (km/s), where $x$ is Bz in G. The authors take this as quantitative confirmation that stronger magnetic fields increasingly suppress convection-driven flux advection, and as the first uniform measurement of that suppression across network, plage, penumbra, and umbra.","pith_inferences":["Because FLCT with a 15-pixel kernel averages over scales of ~5 granules, Eq. (1) should be read as a mesogranular-scale advection law; granular-scale advection of weak flux may be considerably faster, and testing the same method at higher spatial resolution could yield a steeper curve at the weak-field end.","At 2500 G the measured 10 ± 4 m/s is comparable to the expected tracking noise for 12-minute cadence, so the strong-field end of the curve may represent an upper bound on umbral advection rather than a resolved value.","The polynomial fit is empirical; a physically motivated form (e.g., advection speed scaling with the ratio of magnetic to gas pressure) could be fitted to the binned means and compared, which would test whether the plateau and bump have dynamical meaning."],"forward_implications":["The empirical curve gives magneto-convection and coronal heating simulations a direct boundary condition: how fast photospheric footpoints are shuffled as a function of local field strength.","The relation can be folded into coronal loop heating scaling laws to include loops rooted in sunspot umbrae, which current field-strength-and-length laws fail to fit.","The monotonic decrease quantitatively confirms and extends the earlier plage-only trend of Title et al. (1992) to the full range from network to umbra.","The plateau and the bump near the penumbra-umbra boundary, if real, point to additional velocity contributions (penumbral filaments, moat flows, possible p-mode oscillations) that are not pure advection."],"supporting_citations":[{"why":"Provides the FLCT algorithm used to measure horizontal advection speeds from the magnetograms.","marker":"Fisher & Welsch 2008"},{"why":"Original local correlation tracking method on which FLCT is based.","marker":"November & Simon 1988"},{"why":"Motivates the study by showing coronal loop brightness depends on field strength and convection freedom at the loop feet.","marker":"Tiwari et al. 2017"},{"why":"Earlier measurement of speed versus field strength in a plage region, the trend the present result extends.","marker":"Title et al. 1992"},{"why":"Describes the SHARP vector magnetogram data product used here.","marker":"Bobra et al. 2014"},{"why":"Prior optical-flow (DAVE4VM) analysis of SHARP data using similar kernel size, providing comparison and p-mode interpretation.","marker":"Liu et al. 2013"},{"why":"Supplies the auto-correlation method used to judge reliability and persistence of tracked speeds across cadences.","marker":"Welsch et al. 2012"}],"fun_headline_variants":["Stronger solar fields suppress flux advection: 110 to 10 m/s","Magnetic field strength throttles Sun's flux advection: 110→10 m/s","Quantified: Solar flux advection falls from 110 to 10 m/s with field","Sun's flux advection speed drops with magnetic field: 110→10 m/s","Solar flux advection curve: 110 m/s at 150 G to 10 m/s at 2500 G"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire quantitative relation rests on the assumption that the motions FLCT measures in Bz maps at 12-minute cadence with a 15-pixel kernel are the true horizontal advection speeds of magnetic flux driven by convection, accurate even at the ~10 m/s level where the per-frame displacement is only ~0.02 pixels.","fun_headline_variants_meta":{"raw":{"variants":["Stronger solar fields suppress flux advection: 110 to 10 m/s","Magnetic field strength throttles Sun's flux advection: 110→10 m/s","Quantified: Solar flux advection falls from 110 to 10 m/s with field","Sun's flux advection speed drops with magnetic field: 110→10 m/s","Solar flux advection curve: 110 m/s at 150 G to 10 m/s at 2500 G"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000855,"raw_usage":{"total_tokens":3776,"prompt_tokens":1072,"completion_tokens":2704,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":688,"completion_tokens_details":{"reasoning_tokens":2585}},"tokens_in":688,"tokens_out":2704,"duration_ms":18880,"temperature":1.0,"reasoning_tokens":2585,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:26:25.660550+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the FLCT analysis on the same magnetograms but with a 9-pixel or 6-pixel kernel and with 6-minute cadence; if the mean speeds at 150-200 G do not stay within the reported 110 ± 3 m/s, or if the curve's shape changes materially, the specific polynomial is an artifact of the kernel and cadence choices rather than a robust physical relation.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior optical-flow (DAVE4VM) analysis of SHARP data using similar kernel size, providing comparison and p-mode interpretation."},{"cited_title":"T., Kusano, K., Yamamoto, T","cited_arxiv_id":null,"evidence_quote":"Supplies the auto-correlation method used to judge reliability and persistence of tracked speeds across cadences."}],"review_version":1}