{"id":"aa631b7e-8467-49e8-8bcb-dd994a37e7c9","arxiv_id":"2509.10233","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Over 30 winters, persistent warmth in the East/Japan Sea lines up with sea-surface height and along-front currents, while winds and heat fluxes act only as short-lived, non-persistent forcing.","lead":"This study maps how winter sea-surface temperature anomalies in the East/Japan Sea co-vary with winds, heat fluxes, and ocean currents under positive and negative Arctic Oscillation phases, finding that ocean circulation provides the lasting memory. The result points to which fields to monitor for subseasonal marine-heatwave risk.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Cut-and-stitch of JFM winters without handling year-boundary discontinuities can bias every DFA/DCCA exponent; the headline H≈1.4–1.5 and persistence hierarchy are not secure until this is tested.","rationale":"I share the reader's judgment. The cut-and-stitch concatenation is the single most load-bearing concern because it sits upstream of every univariate H and bivariate H_XY and rho_DCCA estimate; if it biases exponents, all three robust features in the abstract lose quantitative support. The paper otherwise has real strengths: iterative-AAFT surrogates preserve marginals and approximate power spectra, BH-FDR control across cells and scales is appropriate, the explicit R^2>=0.90 masking avoids over-interpreting poorly fit H_XY values, and the physically coherent signs (negative heat-flux coupling, positive SSHA coupling) are reassuring. I also considered the AO threshold inconsistency in Section 2.1.4: 'AO > −0.8σ' literally includes AO+ winters in the AO− composite. This is a serious error, but it reads as a sign typo (intended 'AO < −0.8σ'), and correcting it would not change the data-construction assumption that is the focus here. The MHW-susceptibility claim is an over-reach from variance/H coexistence to event statistics, but it is interpretive and secondary to the descriptive cross-correlation maps. The sign convention in Eqs. (6)–(7) is under-specified because F_XY^2 can be negative, yet the later use of rho_DCCA in Eq. (9) partially mitigates this; it is less central than stitching. Therefore the verdict remains CONDITIONAL pending the segment-aware re-analysis, with no change from the reader's verdict.","tokens_in":13791,"tokens_out":7937,"duration_ms":75902,"concrete_test":"Re-estimate all DFA/DCCA exponents by treating each winter as a separate realization: compute F(s) per winter, average log F(s) across winters, then fit the slope, and repeat for the cross-fluctuation functions. Compare the resulting H and H_XY maps with Figures 1, 2–12, and A1–A3. If H in the EKB–SPF corridor drops by more than about 0.2, or if the spatial pattern of significant H_XY shifts materially, the reported near-ballistic persistence and the ocean-dominated persistence hierarchy are artifacts of stitching.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central persistence and cross-persistence results are estimated from a single continuous series formed by concatenating JFM segments across winters (Section 2.1.4). SSTA anomalies are defined against a 1993–2022 day-of-year climatology, so each winter segment retains its own year-specific mean; concatenating introduces a level discontinuity at every year boundary. DFA with m=1 detrending removes within-segment linear trends, but a segment that straddles a year boundary contains a slope break that linear detrending cannot remove, inflating the detrended residual. The fraction of boundary-straddling segments grows with scale s (for roughly 900 days per AO phase and boundaries every 90 days, at s=50 about a quarter of segments contain a boundary), so F(s) is preferentially inflated at large scales and the OLS log–log slope H is biased upward. The same mechanism affects DCCA F_XY(s) and hence H_XY and rho_DCCA whenever the predictor also has year-to-year offsets, which is true for all fields used here. Section 2.2.4 reports only a scale-band robustness check; no segment-aware or per-winter analysis is provided. Thus the headline H≈1.4–1.5 in the EKB–SPF corridor and the 'oceanic fields dominate persistent coupling' hierarchy rest on an untested data-construction assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes 30 winters (1993–2022) of daily SSTA, atmospheric, flux, and oceanic fields in the East/Japan Sea, using grid-point DFA and DCCA to estimate Hurst exponents, cross-Hurst exponents, and DCCA coefficients over 5–50-day scales, separately for Arctic Oscillation positive (AO+) and negative (AO−) winters. Significance is assessed with iterative-AAFT surrogates and Benjamini–Hochberg false-discovery-rate control, and univariate/cross-scaling fits are screened by an R² threshold. The authors report that during AO+ winters the EKB–SPF corridor shows high SSTA variance and near-ballistic persistence (H ≈ 1.4–1.5); that persistent SSTA coupling is dominated by oceanic fields (SSHA, geostrophic meridional velocity) while synoptic winds and turbulent heat fluxes provide strong but non-persistent tendencies; and they frame the results as a two-tier process of mesoscale/advective organization versus synoptic/heat-flux forcing.","tokens_in":14042,"tokens_out":3649,"duration_ms":33790,"significance":"If the results hold, this would be a useful regional contribution with a transferable methodological template: the combination of pre-declared R² quality thresholds on the log–log fits, iAAFT surrogate nulls, and BH-FDR control across grid cells and scales is careful and reproducible, and the data sources are clearly documented. The paper’s proposed hierarchy (oceanic fields organize persistent coupling; atmospheric synoptic forcing and turbulent fluxes act as fast, non-persistent tendencies) is physically plausible and consistent with prior work on winter air–sea interaction in the East/Japan Sea. However, two load-bearing assumptions—the definition of the AO− phase and the concatenation of winter segments into a continuous daily series—are not adequately supported, so the central quantitative claims are not yet secure.","major_comments":[{"comment":"The phase threshold is internally inconsistent: the text states 'Winters with JMF AO > +0.8σ ... were tagged AO+, and those with AO > −0.8σ were tagged AO−'. If taken literally, the AO− set includes all winters with AO greater than −0.8σ, which also contains the AO+ winters, making the phase composites overlapping and invalid. This must be corrected to AO < −0.8σ (or equivalent) and the resulting winter counts reported. Because every AO+ versus AO− comparison in Sections 3 and 4 depends on this partition, this is a central, load-bearing issue.","section":"Section 2.1.4, AO phase selection"},{"comment":"The DFA/DCCA analyses are performed on a series formed by concatenating JFM winter segments ('cut-and-stitch') without accounting for discontinuities at year boundaries. Each winter segment is about 90 days, and the concatenation joins 31 March to 1 January of the following winter; the SSTA fields, although defined relative to a day-of-year climatology, still contain year-specific offsets, producing an abrupt jump at every boundary. With linear (m = 1) detrending, DFA segments that straddle a boundary contain a step-like discontinuity that cannot be removed by a linear fit, inflating the fluctuation functions F(s) and F_XY(s) preferentially at scales where such segments are common. Since the headline H ≈ 1.4–1.5 and the cross-Hurst hierarchy are estimated from this stitched series, the authors must either demonstrate that boundary effects do not bias the scaling estimates (e.g., by a segment-aware analysis that drops or bridges boundary-straddling segments, or by comparing with per-winter H estimates) or the persistence hierarchy cannot be regarded as established. Section 2.2.4 reports only a scale-band robustness check and does not address this data-construction premise.","section":"Section 2.1.4, cut-and-stitch concatenation"}],"minor_comments":[{"comment":"The abbreviation 'JMF' appears consistently in the AO phase-selection paragraph, but the correct winter season everywhere else is 'JFM' (January–February–March). Please correct this typo.","section":"Section 2.1.4"},{"comment":"The sentence about oceanic drivers reads 'meridional geostrophic flow (V10)' maps the advective corridors; however, V10 in Section 2.1.2 denotes the 10-m meridional wind. The intended variable is geo-VA, the meridional geostrophic velocity anomaly. Please fix this notation clash.","section":"Section 4.2"},{"comment":"The abstract refers to the cross-Hurst exponent as λ, while the main text consistently uses H_XY (also written h_XY in places). Please standardize the symbol throughout.","section":"Abstract and Section 2.2.1"},{"comment":"Two equations are both labeled (10): the initial Monte-Carlo p-values and the R²-conditioned p_H. The second equation should be numbered (11) to avoid confusion when readers refer to the quality-control step.","section":"Section 2.2.2, Eqs. (10) and (11)"},{"comment":"The caption states 'Blanks in (C) and (F) mark insignificance at the 95% level', but the methods use Benjamini–Hochberg FDR control at q = 0.05, not a per-test 95% confidence threshold. Please align the caption with the actual inference procedure.","section":"Figure 1 caption"},{"comment":"No confidence intervals are reported for H or H_XY. Given that the scaling estimates drive the interpretation, adding a surrogate-based or block-bootstrap interval would materially aid interpretation, even if only for the headline EKB–SPF corridor values.","section":"Section 2.2.4 and throughout"}],"recommendation":"major_revision","confidential_remarks":"The two major comments are both fixable within the manuscript's scope, but each is load-bearing. The AO− threshold appears to be a typographical error, yet as written it undermines every phase-contrast result. The cut-and-stitch issue requires either new analysis or a convincing artifact test; without it, the H ≈ 1.4–1.5 claim and the 'oceanic fields dominate persistence' hierarchy are not established. I would recommend that the editor require the authors to address both before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper gives the first grid-point, AO-phase-stratified DFA/DCCA maps of SSTA coupling for the East/Japan Sea, and the methodology is mostly careful; but the headline H≈1.4–1.5 is built on a data-construction premise that needs to be tested before I'd trust it.\n\nThe genuinely new piece is that specific combination of diagnostics: grid-point Hurst exponents, cross-Hurst exponents, and DCCA coefficients over 5–50 days, screened with iAAFT surrogates and BH-FDR, for two AO phases. The sign patterns are physically coherent: ATMP positive, UA10 negative, turbulent heat fluxes negative and memory-light, SSHA basin-wide positive, geo-V giving the clearest advective signature. The surrogate testing and FDR control are done carefully, and masking cells where the log–log fit is poor is a reasonable, honest choice. The two-tier interpretation—mesoscale/advection organizes persistence, synoptic forcing is strong but short-memory—is a sensible way to read those maps.\n\nThe soft spot is not minor. The analysis concatenates JFM segments from multiple winters into one continuous series and runs DFA/DCCA on that. Year boundaries are discontinuities in the anomaly series, and linear detrending (m=1) cannot absorb the jumps. For scales of 5–50 days on a ~900-day series, a substantial fraction of segments will straddle a boundary, inflating F(s) preferentially at large scales and biasing H and H_XY upward. All the headline numbers—H≈1.4–1.5 in the EKB–SPF corridor, the persistence hierarchy—come from this stitched series, and there is no segment-aware or per-winter check. Until that's tested, the specific exponent values and the 'ocean fields dominate persistent coupling' claim aren't secure.\n\nTwo smaller issues: the AO− threshold is mis-specified in Section 2.1.4 (written as AO > −0.8σ, almost certainly a typo for AO < −0.8σ), and the MHW-susceptibility language goes slightly beyond what is shown—the paper never analyzes MHW data, so 'increased susceptibility' should be framed as a hypothesis, not an implication.\n\nNet: the paper deserves a serious referee. The maps and framework are useful, but the load-bearing analysis needs to be redone with handling of the yearly concatenation—for example, DFA on individual winters and averaging fluctuation functions, or explicit treatment of boundaries—plus a corrected AO definition and qualified MHW language. If the authors can do that, this becomes a solid regional contribution. I wouldn't cite it in its current form, but I'd send it to review and tell the authors the stitching issue is the gate.","headline":"Useful regional DFA/DCCA mapping, but the headline H values rest on an untested concatenation of winter segments that needs fixing.","tokens_in":14626,"tokens_out":3794,"would_cite":false,"duration_ms":33234,"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":"During positive Arctic Oscillation winters, winter sea-surface temperature anomalies in the East/Japan Sea show near-ballistic persistence (Hurst exponent about 1.4–1.5) along the East Korean Bay–subpolar-front corridor, with long-lasting…","keywords":["Arctic Oscillation","East/Japan Sea","sea surface temperature anomaly","detrended fluctuation analysis","detrended cross-correlation analysis","Hurst exponent","marine heatwaves","cross-persistence"],"falsifier":"Recompute all Hurst and cross-Hurst exponents separately for each individual winter season, or after explicitly removing year-boundary discontinuities from the stitched series; if the 1.4–1.5 exponents and the ocean-dominated cross-persistence collapse toward $0.5$ or lose their spatial pattern, the stated hierarchy would be an artifact of the concatenation.","tokens_in":13528,"feed_emoji":"🌊","tokens_out":9984,"duration_ms":67319,"temperature":0.7,"pith_summary":"The paper argues that winter sea-surface temperature anomalies in the East/Japan Sea are organized by two separate tiers of influence. Persistent, scale-invariant coupling comes from oceanic mesoscale structure—sea-surface height and meridional geostrophic flow—while synoptic winds and turbulent heat fluxes supply strong but essentially memoryless tendencies. Using 30 winters of daily fields, the authors estimate Hurst exponents and detrended cross-correlations on 5–50-day scales at every grid cell, separately for positive and negative Arctic Oscillation winters. The central result is that in positive-AO winters the East Korean Bay–subpolar-front corridor shows unusually high variance and near-ballistic persistence ($H \\approx 1.4$–$1.5$), which the authors interpret as elevated susceptibility to winter marine heatwaves.","feed_headline":"Ocean fields, not winds, carry winter SST memory in East/Japan Sea","feed_subtitle":"H ~1.4–1.5 along the subpolar front in positive Arctic Oscillation winters raises marine-heatwave risk.","key_machinery":"The central machinery is grid-point detrended fluctuation analysis and detrended cross-correlation analysis (DFA/DCCA). At each cell, daily anomaly profiles are divided into 5–50-day segments, a local linear trend is removed, and the fluctuation function is fit as a power law; the slope gives the Hurst exponent $H$ for a single field and the cross-Hurst exponent $H_{XY}$ for a pair, while the normalized DCCA coefficient $\\rho_{\\mathrm{DCCA}}$ supplies the sign and scale-resolved strength of the coupling. Significance is assigned by surrogate-based Monte Carlo testing that preserves each series' amplitude distribution and spectrum, combined with false-discovery-rate control across grid cells and scales. The method's role is to separate persistent, scale-invariant co-evolution from strong but transient correlation.","core_discovery":"The paper claims to establish a two-tier mechanism for winter SST variability in the East/Japan Sea. On 5–50-day scales, SSTA cross-persistence is clear and spatially organized with oceanic fields—above all sea-surface height anomaly and meridional geostrophic velocity—along the East Korean Warm Current and subpolar-front pathways. In contrast, coupling with near-surface air temperature is positive but less persistent, coupling with turbulent heat fluxes is strongly negative with no cross-memory, and zonal wind and wind-stress curl show patchy, sign-changing correlations. The key phase contrast is that during positive Arctic Oscillation winters, SSTA variance and its Hurst exponent peak together ($H \\approx 1.4$–$1.5$) in the East Korean Bay–subpolar-front corridor, which the authors read as a precondition for marine heatwaves. The overall picture is that the atmosphere and fluxes dictate the immediate tendency of SSTA, while the ocean's own height and advection fields decide which anomalies persist.","pith_inferences":["Because the persistence hierarchy is fluxes < atmosphere < ocean, subseasonal predictability of winter SSTA in this basin is likely to come from monitoring sea-surface height and meridional geostrophic transport rather than from heat-flux or wind forecasts.","A transferable extension would apply the same FDR-controlled DCCA analysis to other marginal seas (for example the Yellow Sea or South China Sea) to test whether a two-tier 'ocean organizes, atmosphere forces' structure is generic.","The near-ballistic exponents over 5–50 days suggest that once a warm SSTA pattern is established in AO+ winters along the subpolar front, it may persist well beyond individual synoptic events; if so, winter marine-heatwave warnings could be conditioned on the pre-existing oceanic state rather than on the atmospheric forecast.","A causal test would compare these observational DCCA maps against ocean-model experiments with and without AO-forced wind and flux perturbations, isolating whether the oceanic cross-persistence is produced by advection or by mixed-layer re-emergence."],"forward_implications":["During positive-AO winters, the East Korean Bay–subpolar-front corridor combines high SSTA variance with $H \\approx 1.4$–$1.5$, i.e., near-ballistic persistence, which implies elevated marine-heatwave susceptibility there.","Among atmospheric fields, the 2-m air temperature anomaly shows basin-wide positive $\\rho_{\\mathrm{DCCA}}$ with SSTA and localized cross-persistence, while sea-level pressure and wind-stress curl produce patchy correlations that rarely persist.","Sensible and latent heat flux anomalies couple to SSTA with widespread negative $\\rho_{\\mathrm{DCCA}}$ and essentially no cross-persistence, consistent with fast damping feedbacks.","Sea-surface height anomaly shows the most extensive, AO-phase-stable positive coupling with SSTA, and meridional geostrophic velocity yields the clearest advective cross-persistence along the East Korean Warm Current and subpolar front.","The same grid-point DFA/DCCA workflow, with surrogate testing and false-discovery-rate control, can be transferred to other marginal seas."],"supporting_citations":[{"why":"Documents that positive-AO winters produce warm eddy-like anomalies and increased marine-heatwave days around the East Korean Bay, motivating the corridor focus.","marker":"[4]"},{"why":"Earlier DCCA application to SST and ocean parameters that this study extends to grid-point winter coupling.","marker":"[9]"},{"why":"Introduces detrended fluctuation analysis, the basis for estimating the Hurst exponent.","marker":"[11]"},{"why":"Introduces detrended cross-correlation analysis, the basis for the cross-Hurst exponent.","marker":"[12]"},{"why":"Defines the normalized DCCA coefficient used to measure scale-resolved cross-correlation sign and strength.","marker":"[14]"},{"why":"Provides the stochastic climate-model argument for why atmospheric forcing can produce persistent SST responses through mixed-layer integration.","marker":"[15]"},{"why":"Supplies the iterative amplitude-adjusted Fourier transform surrogate scheme that preserves marginals and spectra for significance testing.","marker":"[17]"},{"why":"Supplies the statistical testing framework for power-law cross-correlated processes, including surrogate-based null distributions.","marker":"[18]"},{"why":"Provides the false-discovery-rate procedure used to control multiple testing across grid cells and scales.","marker":"[19]"},{"why":"Supplies the reanalysis atmospheric and surface-flux fields used in all coupling maps.","marker":"[22]"}],"fun_headline_variants":["Ocean height and currents, not wind, set East/Japan Sea SST persistence","SST memory in East/Japan Sea is ocean-carried, not wind-carried","East/Japan Sea SST persistence: ocean fields, not winds, hold the key","Ocean fields, not winds, give East/Japan Sea SST its long memory","Marine-heatwave risk tied to ocean-set SST persistence in East/Japan Sea"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that stitching the January–February–March daily records of different years into one continuous series does not create artificial jumps that inflate the persistence estimates, since every reported Hurst and cross-Hurst exponent is computed from that stitched record.","fun_headline_variants_meta":{"raw":{"variants":["Ocean height and currents, not wind, set East/Japan Sea SST persistence","SST memory in East/Japan Sea is ocean-carried, not wind-carried","East/Japan Sea SST persistence: ocean fields, not winds, hold the key","Ocean fields, not winds, give East/Japan Sea SST its long memory","Marine-heatwave risk tied to ocean-set SST persistence in East/Japan Sea"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000353,"raw_usage":{"total_tokens":1997,"prompt_tokens":1097,"completion_tokens":900,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":713,"completion_tokens_details":{"reasoning_tokens":794}},"tokens_in":713,"tokens_out":900,"duration_ms":367183,"temperature":1.0,"reasoning_tokens":794,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:56:22.920824+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute all Hurst and cross-Hurst exponents separately for each individual winter season, or after explicitly removing year-boundary discontinuities from the stitched series; if the 1.4–1.5 exponents and the ocean-dominated cross-persistence collapse toward $0.5$ or lose their spatial pattern, the stated hierarchy would be an artifact of the concatenation.","supporting_citations":[{"cited_title":"-Y.; Kim, Y","cited_arxiv_id":null,"evidence_quote":"Documents that positive-AO winters produce warm eddy-like anomalies and increased marine-heatwave days around the East Korean Bay, motivating the corridor focus."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier DCCA application to SST and ocean parameters that this study extends to grid-point winter coupling."},{"cited_title":"-K.; Buldyrev, S.V.; Havlin, S.; Simons, M.; Stanley, H.E.; Goldberger, A.L","cited_arxiv_id":null,"evidence_quote":"Introduces detrended fluctuation analysis, the basis for estimating the Hurst exponent."},{"cited_title":"Detrended Cross -Correlation Analysis: A New Method for A nalyzing Two Nonstationary Time Series","cited_arxiv_id":null,"evidence_quote":"Introduces detrended cross-correlation analysis, the basis for the cross-Hurst exponent."},{"cited_title":"DCCA cross -correlation coefficient: quantifying level of cross -correlation","cited_arxiv_id":null,"evidence_quote":"Defines the normalized DCCA coefficient used to measure scale-resolved cross-correlation sign and strength."},{"cited_title":"Stochastic climate models Part I","cited_arxiv_id":null,"evidence_quote":"Provides the stochastic climate-model argument for why atmospheric forcing can produce persistent SST responses through mixed-layer integration."},{"cited_title":"Improved Surrogate Data for Nonlinearity Tests","cited_arxiv_id":null,"evidence_quote":"Supplies the iterative amplitude-adjusted Fourier transform surrogate scheme that preserves marginals and spectra for significance testing."},{"cited_title":"-Q.; Zhou, W","cited_arxiv_id":null,"evidence_quote":"Supplies the statistical testing framework for power-law cross-correlated processes, including surrogate-based null distributions."},{"cited_title":"Hochberg, Y","cited_arxiv_id":null,"evidence_quote":"Provides the false-discovery-rate procedure used to control multiple testing across grid cells and scales."},{"cited_title":"Bell, B","cited_arxiv_id":null,"evidence_quote":"Supplies the reanalysis atmospheric and surface-flux fields used in all coupling maps."}],"review_version":1}