{"id":"dea57aea-2180-495d-a9cc-b12492c5ddcb","arxiv_id":"2506.09652","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"The 2025 Mandalay earthquake rupture went supershear (~6 km/s), slowed to subshear (~3 km/s) before the video site, then re-accelerated to supershear.","lead":"A security camera recorded the ground breaking during the 2025 Mandalay earthquake, and combining that footage with seismic and satellite data shows the rupture slowed from supershear to subshear speed, then sped up again. This is the first direct read-out of how rupture speed changed along the fault during a single earthquake.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The video's own phase timings imply τP≈3 s and rd≈99 km, yet the paper's simulations use the allowed upper bound τP≈7 s to place rd≈65 km next to the Sentinel-2 slip minimum; this post hoc selection carries the barrier-causation claim.","rationale":"In good faith, the paper's primary observational result—a video showing an S-type phase arriving roughly two seconds before surface slip, combined with an NPW rupture arrival at ~50 s—does kinematically force a departure from uniform supershear rupture. The 2-D/3-D polarity comparison adds support, and the published, peer-reviewed status plus independent Latour et al. (2025) interval estimate argue against dismissing the observation. However, the paper's central causal claim ('the slip minimum caused the deceleration') rests on a parameter choice that is not forced by the data. The video's visible T=30/33/35 timestamps imply τP≈3 s, which gives rd≈99 km. The paper then chooses τP≈7 s, the largest value allowed by the NPW inequality, because it moves rd to 65 km and closer to the 40–60 km slip minimum. The authors are transparent about this selection, but transparency does not remove the circularity: the same slip minimum is used both to motivate the causal story and to select the rupture scenario that makes the story spatially plausible. If the video's own timing is used, the spatial correlation disappears. This is the same weakest assumption identified by the reader, so the verdict remains CONDITIONAL. A reanalysis with τP=3 s, or better, a spontaneous dynamic-rupture simulation with an independently derived stress-drop distribution, would settle whether the barrier-causation claim is real or an artifact of the chosen end-member scenario.","tokens_in":11095,"tokens_out":25783,"duration_ms":305240,"concrete_test":"Recompute rd and Vsub from Eqs. (A.2)-(A.3) using the video-visible timings τP≈3 s and τS≈2 s (the earliest case) instead of the selected τP≈7 s, and overlay the resulting rd on the Sentinel-2 slip profile. If rd≈99 km and does not overlap the 40–60 km slip minimum, the barrier-causation claim is an artifact of choosing the maximum allowed τP; if rd still falls within or dynamically connects to the slip-minimum region, the claim survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's causal claim—that the Sentinel-2 slip minimum at 40–60 km from the epicenter caused the rupture to decelerate—is carried by the choice of the 'latest case' in Supp. A.2. The video's visible phase timestamps give τP≈3 s and τS≈2 s, which yields rd≈99 km and Vsub≈2.74 km/s. The authors instead simulate τP≈7 s, the maximum allowed by the NPW causality bound (S.1), yielding rd≈65.2 km, because 65 km is 'more comparable' to the slip minimum. Since the paper admits the exact P onset cannot be determined and that P could have arrived before T=30, τP is only a lower bound; the data do not prefer the latest case. The simulation parameters are thus selected, within the allowed range, to make the barrier-causation story look good. With the video's own timing, the deceleration point is ~40–60 km south of the slip minimum, and the claimed spatial correlation disappears. The independently reported S-before-rupture interval and the 2-D/3-D eastward-acceleration polarity support the qualitative supershear→subshear→supershear sequence, but they do not constrain τP, and therefore do not support the specific rd/barrier association.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes the 2025 M_w7.7 Mandalay, Myanmar earthquake and claims, from a CCTV video of surface rupture, the GEOFON station NPW record, kinematic wavefield simulations, and Sentinel-2 pixel-offset slip estimates, that the southward rupture initially propagated at supershear speed (~6 km/s), decelerated to subshear speed (~2.7–3.1 km/s) before reaching the camera at 124 km from the hypocenter, and then re-accelerated to supershear speed to reach NPW at t_N = 50 s. The deceleration is inferred from an S-wave observed ~2 s before rupture onset in the video combined with the NPW arrival time. The authors further associate the inferred deceleration point (rd = 65–99 km) with a Sentinel-2 slip minimum at 40–60 km south of the epicenter and propose a low-stress-drop segment boundary as the cause of the temporary deceleration. The paper includes 2-D BEM and 3-D FDM simulations of fault-normal acceleration to argue that the observed eastward acceleration before rupture is consistent with the deceleration scenario but not with a constant supershear scenario.","tokens_in":11144,"tokens_out":5857,"duration_ms":70069,"significance":"If the central inference holds, this is a rare direct observation of a supershear-to-subshear-to-supershear rupture sequence, and the paper combines a unique video dataset, near-field seismic data, and satellite geodesy. The kinematic-constraint argument in Section 2.1 and Supplementary A.2 is internally consistent and does not, by itself, depend on the barrier-causation story: the S-before-rupture timing at CCTV and the t_N = 50 s arrival at NPW force a subshear segment somewhere between the hypocenter and NPW. The paper also provides open code and data links, and the two independent simulation approaches (BEM and FDM) reproduce the qualitative fault-normal acceleration pattern. However, the quantitative deceleration point and the causal link to the slip minimum rest on a parameter choice (the 'latest case' τP ≈ 7 s) that the authors select because it places rd near the Sentinel-2 slip minimum. The video-derived τP ≈ 3 s gives rd ≈ 99 km, well away from the 40–60 km slip low.","major_comments":[{"comment":"The choice of the 'latest case' (τP ≈ 7 s, rd = 65.2 km, Vsub = 3.13 km/s) for the numerical simulations is explicitly justified by the statement that 'rd = 65.2 is more comparable to the local minimum of the surface slip.' This is a post hoc selection: the video's own observed P-to-S interval gives τP ≈ 3 s and rd ≈ 98.8 km, whereas τP is only constrained as a lower bound. The data therefore do not prefer rd = 65 km over rd = 99 km, and the claimed spatial coincidence between the deceleration point and the 40–60 km slip minimum is largely produced by the parameter choice. Since the barrier-causation narrative in Sections 3 and 4 rests entirely on this coincidence, the paper must either provide a rigorous treatment of τP uncertainty (e.g., a range of rd and Vsub with a likelihood or posterior) or explicitly refrain from claiming that the low-slip segment caused the deceleration.","section":"Supplementary Material A.2"},{"comment":"The entire quantitative framework depends on t_N = 50 s, but this value is read from a figure in Lai et al. (2025) and corrected by 2 s because of an assumed origin-time difference. The paper provides no uncertainty or alternative values for t_N. If t_N were 48 s or 52 s, the allowed τP range, rd, and Vsub would all shift by several kilometers and by up to ~0.2 km/s. The authors should show the sensitivity of rd and Vsub to t_N and to the assumed V_P = 6.0 km/s and V_S = 3.5 km/s before presenting 65–99 km and 2.7–3.1 km/s as definitive ranges.","section":"Section 2.1 and Supplementary A.2"},{"comment":"The kinematic simulations are presented as strong support for the deceleration scenario, but the 'decelerating' input uses exactly the rd = 65.2 km and Vsub = 3.13 km/s values that were selected in Supplementary A.2 because they match the slip minimum. The comparison therefore demonstrates that a model built to reproduce the video's S-before-rupture timing also produces an eastward acceleration pulse before rupture, not that the data uniquely require that specific deceleration point. The paper should frame this as a consistency check rather than independent confirmation, and should test whether the acceleration polarity remains eastward for the earliest case (rd ≈ 99 km, Vsub ≈ 2.74 km/s) over the full allowed range.","section":"Section 2.2 and Figure 3"},{"comment":"The Sentinel-2 slip minimum at 40–60 km is estimated from north-component pixel offsets with an 80 m posting, a 320 m correlation window, and exclusion of 160 m on each side of the fault. The paper does not report formal uncertainties, a resolution test, or an assessment of possible atmospheric and orbital artifacts for this particular local minimum. Because this slip low is the only independent evidence for the low-stress-drop barrier invoked in the causal explanation, its robustness should be quantified before it is used to explain the rupture deceleration.","section":"Section 3 and Figure 2b"}],"minor_comments":[{"comment":"The earthquake magnitude notation appears as 'Mw7.7' in several places; it should be typeset as M_w7.7 for consistency with seismological convention.","section":"Throughout"},{"comment":"The description of the gate displacement is confusing: the paper states that the ground accelerated east (left side of the video frames) and that the sliding gate was 'displaced apparently westward due to inertia.' Please define the mapping between video-left and geographic east explicitly and reconcile this with the sign convention used for the eastward acceleration in Figure 3.","section":"Section 2.2"},{"comment":"The sentence 'The authors thank Dr. Ryo Okuwaki for their helpful comments' uses a plural pronoun for a singular named individual; this should be rephrased.","section":"Acknowledgements"},{"comment":"The reference list contains garbled characters, for example 'Kahramanmara¸ s' in the METU-EERC entry; these need to be corrected.","section":"References"},{"comment":"The caption says 'All units are arbitrary,' which is fine, but the text in Section 2.2 refers to 'significant eastward acceleration' without specifying the normalization used for the two simulations; please state whether the amplitudes are normalized separately in each panel.","section":"Figure 3"}],"recommendation":"major_revision","confidential_remarks":"The paper is timely and interesting, and the kinematic deceleration inference has a non-circular core that may well be correct. However, the specific deceleration-point location and the barrier-causation story are vulnerable because the 'latest case' was chosen to match the Sentinel-2 slip minimum. A major revision that quantifies the τP and t_N uncertainties, tests the full rd range, and softens the causal claim would make the paper acceptable. I would not recommend rejection on the current evidence, but the quantitative causal statement in the abstract and conclusions needs to be downgraded or properly supported."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result here is the qualitative rupture history: the Mandalay rupture propagated at supershear speed, decelerated below the S-wave speed before reaching the CCTV camera, and re-accelerated to supershear afterward. That claim is kinematically forced once you accept two readings: the ~2 s S-wave-before-rupture interval in the video (independently reported by Latour et al.) and the NPW rupture arrival at t_N = 50 s. The video is a genuinely unprecedented observation, and the authors use it cleverly. The eastward acceleration polarity preceding rupture is a nice independent diagnostic that discriminates subshear from supershear, and they reproduce it with both 2-D and 3-D simulations. Code and data are available, and the paper is clearly written.\n\nThe soft spot is exactly where the stress-test note points. The video's own phase timings give τP ≈ 3 s, which yields r_d ≈ 99 km and V_sub ≈ 2.74 km/s. The authors instead simulate the 'latest case' with τP ≈ 7 s, r_d ≈ 65 km, because that puts the deceleration point near the Sentinel-2 slip minimum at 40–60 km. But they admit the exact P onset cannot be determined and that τP is only a lower bound. So the data do not prefer the latest case; the choice is post hoc, and it carries the barrier-causation narrative. With the video's own timing, the deceleration point is 40–60 km south of the slip minimum and the spatial correlation disappears. This does not undermine the qualitative supershear→subshear→supershear sequence, but it means the specific r_d and the barrier association are suggestive, not established.\n\nOther issues are minor by comparison: the origin-time ambiguity (USGS vs Lai, t_N = 50 vs 48 s) is not propagated into the error budget; the waveform comparison uses arbitrary units and only two scenarios; and the Sentinel-2 slip profile has no uncertainties. These are fixable with modest effort.\n\nWho should read this: anyone working on near-field rupture dynamics or supershear transitions. The video analysis alone makes it worth a look, and the kinematic argument is elegant. I would cite it for the qualitative sequence, not for the barrier location. If this crossed my desk as a new submission, I would send it to review; it deserves referee time, with the request that the authors confront the τP selection issue head-on and propagate the origin-time uncertainty.","headline":"The qualitative supershear-subshear-supershear sequence is robust and worth knowing; the specific deceleration point and barrier story rest on a post hoc parameter choice that the data do not prefer.","tokens_in":11941,"tokens_out":1731,"would_cite":true,"duration_ms":20794,"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 2025 Mandalay earthquake's rupture began supershear, dropped to subshear, then became supershear again.","keywords":["Mandalay earthquake","Sagaing fault","supershear rupture","subshear rupture","rupture deceleration","CCTV video","kinematic rupture simulation","satellite slip distribution"],"falsifier":"A decisive check would be a second near-fault record between the deceleration band and NPW: if its rupture-arrival time requires a sustained subshear segment or a much earlier recovery, the inferred re-acceleration fails. More immediately, re-reading the CCTV frames with sub-pixel image correlation to fix the exact P, S, and slip times would distinguish the two end-members (deceleration near 99 km versus near 65 km); if the S-wave does not consistently lead the rupture by about two seconds, the deceleration claim loses its quantitative anchor.","tokens_in":2069,"feed_emoji":"⚡","tokens_out":3924,"duration_ms":126506,"temperature":0.7,"pith_summary":"The paper argues that the Mw7.7 Mandalay earthquake of 28 March 2025 did not rupture at one uniform speed. It claims that the rupture started at supershear speed near the epicenter, slowed to about 3 km/s before passing a site 124 km to the south, and then re-accelerated to supershear, reaching a seismic station 246 km away roughly 50 seconds after initiation. The evidence combines a CCTV video showing S-waves arriving about two seconds before surface slip, kinematic simulations that reproduce the observed ground acceleration only with a deceleration, and a satellite-derived slip minimum 40–60 km from the epicenter interpreted as a low-stress-drop barrier. If correct, this documents a complete supershear-to-subshear-to-supershear rupture sequence and ties the speed change to a geological segment boundary. It also shows that a single security camera can provide rupture-timing data where no seismometers exist.","feed_headline":"Mandalay quake rupture slowed to subshear, then sped back up","feed_subtitle":"Video and one seismic station trace a supershear-to-subshear-to-supershear sequence tied to a low-slip barrier.","key_machinery":"The load-bearing device is a rupture-front timing diagram. The P-wave, S-wave, and rupture-front arrivals read from the video fix the order at $r = 124$ km (P by $T = 30$ s, S at $T \\approx 33$ s, slip at $T \\approx 35$ s), and the condition that the rupture must reach NPW at $t = 50$ s enforces the speed history. From these, with $V_P = 6.0$ km/s and $V_S = 3.5$ km/s, the deceleration point and subshear speed are solved geometrically ($r_d = 65$–$99$ km, $V_{sub} = 2.7$–$3.1$ km/s). The paper validates this with kinematic simulations: a two-second sine-squared slip pulse with the deceleration scenario produces an eastward acceleration pulse before slip onset in both 2-D and 3-D models, while a constant 4.92 km/s supershear pulse does not. The interpretation also uses a spectral relation between slip and stress drop to connect the observed slip minimum to a low-stress-drop barrier.","core_discovery":"On 28 March 2025, the Mw7.7 Mandalay earthquake ruptured a roughly 450 km stretch of the Sagaing fault. Combining a CCTV video taken 124 km south of the hypocenter with the near-fault seismic record at Naypyidaw (246 km away) and satellite-measured surface slip, the paper reconstructs the rupture's along-fault speed. Its central claim: the rupture left the epicenter at supershear speed, decelerated to about 2.7–3.1 km/s in the band 65–99 km from the hypocenter, and then re-accelerated to supershear, reaching the southern station about 50 s after initiation. The deceleration is signaled in the video by an S-wave arriving roughly two seconds before the rupture front, and in the ground-motion pattern by an eastward fault-normal acceleration that kinematic simulations reproduce only if the rupture slows down. The paper ties the slowdown to a low-slip (2–3 m) segment at 40–60 km, interpreted as a low-stress-drop barrier near 21.5N, and proposes that the energy-release-rate feedback of unstable rupture speeds explains the temporary return to subshear.","pith_inferences":["The timing data alone allow two end-member deceleration points, roughly 99 km for a 3 s P-wave duration and roughly 65 km for the latest case the paper simulates; matching the satellite slip minimum at 40–60 km is what selects the nearer point, so the barrier link is not uniquely determined by the timing itself.","If the inferred sequence is typical, large strike-slip ruptures crossing segment boundaries should frequently show speed drops, and dynamic rupture models with heterogeneous stress drop could predict where and whether the rupture recovers to supershear.","Near-fault strong-motion hazard may depend on where a rupture switches regime, since particle motion and radiation patterns differ between subshear and supershear segments; this suggests that mapping low-slip barriers before an earthquake matters for hazard estimation."],"forward_implications":["Average rupture speeds can hide dramatic along-fault speed changes; for this earthquake, the mean 4.92 km/s to NPW is a mix of roughly 6 km/s, about 3 km/s, and then supershear again.","Low-slip, low-stress-drop segments on strike-slip faults can act as barriers that temporarily drop a supershear rupture below the S-wave speed.","Fault-normal ground motion near a fault carries a fingerprint of the rupture speed regime: the sign and timing of the pre-slip acceleration pulse changes between subshear and supershear propagation.","A single CCTV camera can serve as a de facto seismic station, and with one waveform record plus an origin time it can constrain rupture-velocity changes.","Satellite slip maps can help locate where rupture speed changes are expected by revealing stress-drop minima."],"supporting_citations":[{"why":"Supplies the origin time, epicenter, focal depth, and roughly 450 km fault length used to set the timing geometry and zero time.","marker":"USGS (2025)"},{"why":"Provides the near-fault record at NPW whose rapid displacement at about 50 s after initiation gives the southern rupture-arrival constraint.","marker":"Lai et al. (2025)"},{"why":"Analyzes the same CCTV video, finding the S-wave arrives about 1.8 s before the rupture front, the central timing observation the paper builds on.","marker":"Latour et al. (2025)"},{"why":"Teleseismic inversion showing initial supershear propagation near the hypocenter and a potency-density minimum near 21.5N, giving the background and the barrier location.","marker":"Inoue et al. (2025)"},{"why":"2-D steady-state pulse solutions showing how fault-normal ground motion reverses pattern between subshear and supershear rupture.","marker":"Dunham & Archuleta (2005)"},{"why":"3-D dynamic rupture simulations of subshear and supershear ground motion used to support the pattern comparison.","marker":"Abdelmeguid et al. (2024)"},{"why":"Establishes a spectral relation between slip and stress drop used to infer a low-stress-drop barrier from the slip minimum.","marker":"Andrews (1980)"},{"why":"Numerical heterogeneous stress-drop models showing how slip distribution maps to stress drop, supporting the barrier interpretation.","marker":"Andrews & Barall (2011)"},{"why":"Provides the energy-release-rate and unstable-speed-interval arguments used to explain why the rupture reverts to subshear and later recovers.","marker":"Freund (1990)"},{"why":"Gives the boundary-integral kernel used for the 2-D kinematic slip-pulse simulations that reproduce the observed pre-slip acceleration.","marker":"Tada & Madariaga (2000)"}],"fun_headline_variants":["Mandalay quake rupture hit the brakes, then floored it","Video capture: Mandalay rupture's supershear speed dip","Earthquake speed wobble: supershear to subshear to supershear","Mandalay M7.7 rupture decelerated, then re-accelerated","Fault rupture slows on low-slip segment, then races again"],"cache_read_input_tokens":13824,"weakest_assumption_plain":"All derived speeds depend on the reading that the CCTV video shows the P-wave by $T = 30$ s, the S-wave at about $T = 33$ s, and surface slip at about $T = 35$ s at the camera site, together with assumed wave speeds $V_P = 6.0$ km/s and $V_S = 3.5$ km/s and the 50 s rupture arrival at NPW; if that timing or those values are wrong, the deceleration and its location lose their quantitative support.","fun_headline_variants_meta":{"raw":{"variants":["Mandalay quake rupture hit the brakes, then floored it","Video capture: Mandalay rupture's supershear speed dip","Earthquake speed wobble: supershear to subshear to supershear","Mandalay M7.7 rupture decelerated, then re-accelerated","Fault rupture slows on low-slip segment, then races again"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000737,"raw_usage":{"total_tokens":3309,"prompt_tokens":975,"completion_tokens":2334,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":2238}},"tokens_in":591,"tokens_out":2334,"duration_ms":15824,"temperature":1.0,"reasoning_tokens":2238,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:45:59.402011+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive check would be a second near-fault record between the deceleration band and NPW: if its rupture-arrival time requires a sustained subshear segment or a much earlier recovery, the inferred re-acceleration fails. More immediately, re-reading the CCTV frames with sub-pixel image correlation to fix the exact P, S, and slip times would distinguish the two end-members (deceleration near 99 km versus near 65 km); if the S-wave does not consistently lead the rupture by about two seconds, the deceleration claim loses its quantitative anchor.","supporting_citations":[{"cited_title":"M7.7--2025 mandalay, burma (myanmar) earthquake","cited_arxiv_id":null,"evidence_quote":"Supplies the origin time, epicenter, focal depth, and roughly 450 km fault length used to set the timing geometry and zero time."},{"cited_title":"M., Htwe, Y","cited_arxiv_id":null,"evidence_quote":"Provides the near-fault record at NPW whose rapid displacement at about 50 s after initiation gives the southern rupture-arrival constraint."},{"cited_title":"Direct Estimation of Earthquake Source Properties from a Single CCTV Camera","cited_arxiv_id":"2505.15461","evidence_quote":"Analyzes the same CCTV video, finding the S-wave arrives about 1.8 s before the rupture front, the central timing observation the paper builds on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Teleseismic inversion showing initial supershear propagation near the hypocenter and a potency-density minimum near 21.5N, giving the background and the barrier location."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"2-D steady-state pulse solutions showing how fault-normal ground motion reverses pattern between subshear and supershear rupture."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"3-D dynamic rupture simulations of subshear and supershear ground motion used to support the pattern comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes a spectral relation between slip and stress drop used to infer a low-stress-drop barrier from the slip minimum."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Numerical heterogeneous stress-drop models showing how slip distribution maps to stress drop, supporting the barrier interpretation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the energy-release-rate and unstable-speed-interval arguments used to explain why the rupture reverts to subshear and later recovers."},{"cited_title":"& Madariaga, R","cited_arxiv_id":null,"evidence_quote":"Gives the boundary-integral kernel used for the 2-D kinematic slip-pulse simulations that reproduce the observed pre-slip acceleration."}],"review_version":1}