{"id":"155f307d-2369-481f-addd-9e07eee47b4a","arxiv_id":"2502.02469","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"A dose-rate threshold alarm for Mars SEP events is tested on MSL/RAD data, but the 30-minute warning and zero-false-alarm claims depend on threshold tuning and excluding weak events.","lead":"A simple alarm based on radiation dose rates measured by the Curiosity rover could warn future Mars astronauts to shelter when a solar particle storm begins. The authors test it on 21 historical solar events and claim at least 30 minutes of warning, though weak events with shorter times are excluded.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 30-minute lead-time guarantee is not supported for the system as described: lead times are retrospective, four detected events have 2-17 min lead times and are excluded, and the two-measurement confirmation needed for near-zero FAR consumes 15 min.","rationale":"The paper's operational contribution—a simple threshold on RAD dose-rate E with archived data and a clear FAR analysis—is real, and much of the methodology is reproducible. The load-bearing problem is that the headline guarantee is stated without the qualifications that the authors themselves introduce in Section 4. Lead time is defined post hoc relative to the final total dose, so it is not a forward-looking warning. The minimum is obtained only after discarding four triggered events with 2-17 minute lead times, and the near-zero-FAR configuration (two consecutive measurements) further subtracts 15 minutes. Under that configuration the shortest retained lead time is 18 minutes, below the advertised 30. This does not destroy the value of the system as a last-resort backup, but it means the abstract and Section 4 must be revised to state the guarantee only for the quick single-measurement alert (with its 25% surface FAR) or for moderate-to-large events, after subtracting confirmation delay. Because the data are public, a simple recomputation can settle the point. The reader's CONDITIONAL verdict is appropriate; no change needed.","tokens_in":22604,"tokens_out":10885,"duration_ms":99900,"concrete_test":"Using the archived MSL/RAD dose-rate E data, recompute the lead time and time-to-peak for all events that trigger the 25% threshold, but start the clock at the second consecutive 15-minute measurement above 1.25× background (the official alert). Include the four low-dose events and do not discard them. Report the minimum lead time and minimum time-to-peak. If either minimum is below 30 minutes, the headline 30-minute guarantee is falsified for the system as configured for near-zero FAR.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline claim in the abstract and Section 4—'at least 30 minutes' to avoid peak dose and >90% of cumulative dose—rests on a lead-time definition that is not a real-time warning horizon. In Figure 3 (right) and Table 1, lead time is the time from the 25%-threshold crossing to the moment 10% of the event's final total SEP dose has accumulated. This quantity is only known retrospectively, after the event total is fixed. Worse, the minimum lead time is computed after excluding four events that did trigger the system (2012-01-23, 2014-09-10, 2023-03-12, 2024-09-02) with lead times of 17, 2, 5, and 8 minutes. These are dismissed as 'not prioritized,' but they are detected by the system and would give an astronaut 2-17 minutes, not 30. The near-zero FAR claim introduces a second problem: Figure 4 shows single-measurement FAR at 25% is 25% on the Martian surface; zero FAR requires two consecutive measurements, which the text says sacrifices 15 minutes of warning. Subtracting 15 minutes from the tabulated lead times makes even the shortest retained event, 2024-09-05 (33 min), drop to 18 min, and the shortest time-to-peak events fall below 30 min as well. Thus the abstract's unqualified guarantee does not follow from the paper's own definitions and data.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript describes a nowcasting system for solar energetic particle (SEP) events relevant to future Mars missions. The system uses the MSL/RAD dose rate E in 15-minute cadence: an event is triggered when the dose rate exceeds a five-day linear background fit by 25%. Using the 21 SEP events observed by RAD (5 during the 2011-2012 cruise, 16 on the Martian surface through September 2024), the authors compute shelter durations, times to peak, lead times (defined as the time from trigger to the accumulation of 10% of the total event dose), and total background-removed doses. They also simulate false alarm rates over the full dataset for single- and two-measurement triggers and for thresholds between 5% and 30%. The central claims are that the system can provide astronauts with at least 30 minutes to avoid the peak dose and the majority of the cumulative dose, and that it achieves a near-zero false alarm rate both in deep space and on the Martian surface.","tokens_in":22883,"tokens_out":12177,"duration_ms":107221,"significance":"If the claims were supported as stated, this would be a valuable, minimal-cost, last-resort warning layer for Mars missions, complementing forecast tools such as REleASE and ESPERTA that have not been validated for Mars. The strengths of the paper are real: it is built on 13 years of actual RAD flight data; the algorithm is simple and clearly specified, including a two-step confirmation logic; the false-alarm simulation is a systematic and appropriate test of the method; the negative result for dose rate B in Appendix 6 is honestly documented; the analysis data are openly provided on Zenodo; and the detector-specific scope limitation at the end of Section 5 is stated explicitly. However, the headline quantitative claims go beyond what the paper's own tables establish. The 30-minute lower bound is obtained after excluding detected events with 2-17 minute lead times; the near-zero false alarm rate requires a two-measurement confirmation that costs 15 minutes of warning time; the lead time is a retrospective quantity; and the 25% threshold is selected on the same in-sample data used to report the FAR.","major_comments":[{"comment":"The abstract's claim that the system 'can provide astronauts with at least 30 minutes' is not supported by Table 1 as presented. Four events that trigger the system have lead times of 17 minutes (2012-01-23), 2 minutes (2014-09-10), 5 minutes (2023-03-12), and 8 minutes (2024-09-02); the minimum among the events retained as prioritized is 33 minutes (2024-09-05 on the surface, with 35 minutes for 2012-03-13 in deep space). Section 4 excludes the four short-lead-time events as 'not prioritized' because of negligible dose, but the exclusion is not applied consistently by the tabulated doses: the excluded 2012-01-23 event has a total SEP dose of 12.71 µGy, which is larger than the retained 2013-04-10 event (9.22 µGy). The universal statement in Section 4 ('an astronaut should not move more than 30 minutes away from a shelter') is therefore obtained by removing the counterexamples, and it should be reformulated as conditional on a stated minimum event dose, or replaced by the empirical distribution of lead times.","section":"Abstract; Section 4, Table 1"},{"comment":"The lead time used in Table 1 and in the 30-minute claim is defined retrospectively: it is the interval from the 25%-threshold crossing to the time at which 10% of the event's final total SEP dose has accumulated. Because the final total is known only after the event has ended, a real-time system cannot know, at the moment of triggering, how much lead time a given event will provide. The sentence in Section 4 ('to propose a universal lower limit... we therefore assume the minimum lead time, which is approximately 30 minutes') is thus an empirical property of the selected historical events, not an operational guarantee, unless the manuscript states and defends an additional assumption about future dose-accumulation rates. The wording should distinguish the retrospective characterization of historical events from a forward-looking warning guarantee.","section":"Section 3, Figure 3 (right); Section 4"},{"comment":"The abstract's conjunction of 'at least 30 minutes' with 'near-zero false alarm rate' is internally inconsistent with the paper's own results. Figure 4 shows that with a single measurement at the 25% threshold the FAR is 25% on the Martian surface and zero in deep space; the zero FAR in both environments is reached only with the two-measurement confirmation, which the text states sacrifices 15 minutes of warning time. Subtracting 15 minutes from the tabulated lead times, the shortest retained surface lead time (2024-09-05, 33 minutes) becomes 18 minutes, and the shortest retained deep-space lead time (2012-03-13, 35 minutes) becomes 20 minutes, both below 30 minutes. The abstract and the Summary should state explicitly which claim belongs to which warning mode: the 30-minute lead time belongs to the immediate single-measurement notification (25% surface FAR), while the near-zero FAR belongs to the official alert delivered 15 minutes later.","section":"Section 4.1, Figure 4; Abstract; Section 5"},{"comment":"The near-zero FAR result is selected in-sample. The section scans trigger thresholds from 5% to 30% on the same 21-event dataset used to report the performance, and the uniform 25% threshold is justified by the FAR values obtained in that scan. With only 5 deep-space and 16 surface events, a zero-FAR outcome carries a wide binomial confidence interval; the error bars in Figure 4 acknowledge the statistical uncertainty, but the abstract's unqualified 'near-zero false alarm rate' should be presented as an in-sample property of this dataset with the small-sample caveat stated explicitly.","section":"Section 4.1, Figure 4"}],"minor_comments":[{"comment":"Event identifications are inconsistent across tables: '2014-08-31' in Table 4 corresponds to '2014-09-02' in Table 1; the event described in the text as May 19, 2024 appears as 2024-05-20 in Table 1 and as 2023-05-19 in Table 4; and events dated 2024-10-05 and 2024-10-09 appear in Tables 3 and 4 but are absent from Table 1.","section":"Table 4"},{"comment":"The observation 'it can be observed that time to peak ≳ lead time' is violated by the largest surface event, 2024-05-20, which has a time to peak of 0h53m and a lead time of 0h57m; the sentence should be qualified.","section":"Section 4"},{"comment":"The citation for astronaut career dose limits appears in the text as '(?, ?)', and the reference list includes 'HandWiki' (a wiki) and 'Richardson et al., n.d.'; these are not acceptable as final references for a journal submission.","section":"Section 1; References"},{"comment":"The caption of Figure 3 and the surrounding text define lead time as 'the period in which an astronaut can avoid at least 90% of the total SEP dose'; this definition should state explicitly that the interval is measured from the 25%-threshold crossing, not from the physical onset of the event, so that the distinction is not left to the reader.","section":"Section 3, Figure 3"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for Space Weather and draws on a unique long-duration RAD dataset, and the authors have done the field a service by releasing the analysis data. My principal concern is that the abstract and Summary are written to maximize the headline claims (at least 30 minutes warning; near-zero false alarm rate) without the qualifications that the body of the paper itself imposes: a retrospective lead-time definition, a two-measurement confirmation that costs 15 minutes, four detected events with 2-17 minute lead times that are excluded from the minimum, and an in-sample threshold selection. All of these are straightforwardly fixable in a revision, and I do not see the need for new analysis as long as the claims are re-stated to match the evidence. The reference list also needs professional cleanup (a '(?, ?)' placeholder and a wiki citation)."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a useful operational paper with a real dataset and a clear method, but the headline claims are stronger than the analysis supports. If you read it for the event catalog and the false-alarm comparison, it's fine; if you take the \"at least 30 minutes\" and \"near-zero false alarms\" at face value, you'll be misled.\n\nWhat's new: they apply a simple dose-rate threshold nowcast to the full MSL/RAD record (cruise plus surface), give lead times and shelter durations for 21 events, and provide an explicit FAR study across thresholds. The data are open (Zenodo). That is a legitimate extension of Posner and Strauss's simulation work to real measurements, and the detector comparison in the appendix (dose rate E vs B) is a genuinely useful negative result: dose rate B is unusable for nowcasting due to RTG and temperature effects.\n\nThe soft spots are exactly where the stress-test lands. First, the \"lead time\" is defined retrospectively: the time from threshold crossing until 10% of the final total SEP dose has accumulated. Only after the event is over can you know that 10% point. A real-time warning gives you the threshold crossing; the lead time is a retrospective quality metric, not a guaranteed warning horizon. Second, the 30-minute lower bound is obtained after excluding four events with 2-17 minute lead times, dismissed as \"not prioritized\" because their total doses are small. That exclusion is defensible if the goal is acute risk, but the abstract does not say it. Third, the zero FAR on the surface requires two consecutive measurements; a single measurement gives 25% FAR, and the confirmation costs 15 minutes. Subtracting that from the retained events, the 30-minute guarantee evaporates.\n\nThe threshold selection (25%) is done on the same data used to evaluate FAR, so a held-out validation would be more convincing. I don't think this is fatal for an operational nowcast—the concept is sound and the system would still be a useful last resort—but the authors need to qualify the claims and either use prospective validation or reposition the results as retrospective analysis.\n\nBottom line: a careful referee could get this into shape. I'd send it to review, with the clear expectation that the abstract be rewritten to match the actual definitions of lead time and FAR. Worth citing for the event list and the dose-rate comparison.","headline":"A useful operational nowcast for Mars SEPs with open data, but the 30-minute and near-zero-FAR claims need qualification before the abstract matches the analysis.","tokens_in":23519,"tokens_out":2098,"would_cite":true,"duration_ms":20179,"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":"A 25% rise in local dose rate can warn Mars crews of solar particle storms at least 30 minutes ahead.","keywords":["solar energetic particles","nowcasting","Mars radiation environment","MSL/RAD","dose rate E","astronaut warning","false alarm rate","space weather"],"falsifier":"Find or record an SEP event with a significant total dose (comparable to the ones flagged as threatening, e.g., above roughly 100 µGy background-removed) whose lead time, measured from the 25%-above-background trigger to accumulation of 10% of the total SEP dose, is less than 30 minutes; or record any two-consecutive-measurement trigger above the 25% threshold that is not followed by a real SEP event, which would break the claimed zero false-alarm rate.","tokens_in":22358,"feed_emoji":"☀️","tokens_out":5798,"duration_ms":51881,"temperature":0.7,"pith_summary":"Solar energetic particle (SEP) storms are a serious radiation hazard for crews traveling to or living on Mars, and the existing forecast tools were built for Earth. This paper argues that a simple local warning system can serve as a reliable last resort: whenever the tissue-equivalent dose rate measured by a RAD-like detector rises 25% above its five-day background and a second consecutive measurement confirms the rise, an astronaut should head to shelter. Reconstructing 21 SEP events recorded by the MSL/RAD instrument, the authors find that for the events that matter, this trigger would have arrived at least 30 minutes before the peak dose rate and before 10% of the cumulative event dose had accumulated, while missing only minor events whose added dose is negligible. If this holds, future Mars missions can set a plain operational rule: keep within 30 minutes of shelter, and use a dosimeter threshold as the fallback alarm when forecasts fail.","feed_headline":"Dose spike gives Mars astronauts a 30-minute warning","feed_subtitle":"A 25%-above-background threshold with a second-measurement check caught every significant SEP event with zero false alarms.","key_machinery":"The load-bearing object is the dose rate E, the tissue-equivalent absorbed dose rate recorded by the plastic scintillator in the Radiation Assessment Detector, which has a large geometric factor and is insensitive to the rover's radioisotope power source. The method couples two simple components: a rolling five-day linear background fit with a fixed 25% (1.25×) trigger threshold, and a two-consecutive-measurement confirmation that suppresses false triggers from heavy-ion outliers and post-Forbush-decrease recoveries. The paper's lead time is defined retrospectively as the interval from the trigger onset to the time when 10% of the background-removed total event dose has accumulated, which is the quantity that supports the 30-minute lower limit.","core_discovery":"The paper's central claim is that the onset of a dangerous SEP event can be nowcast, without any solar observation, from the local dose rate alone. Using the plastic-scintillator channel (dose rate E) of MSL/RAD, the authors define the background by a linear fit to the previous five days and trigger a warning when the current 15-minute dose rate exceeds 1.25 times that background; a second consecutive measurement converts a preliminary notice into a confirmed alert. Tested against five deep-space and sixteen surface SEP events, the two-measurement version achieves a zero false-alarm rate at the 25% threshold in both environments, and the authors state that a confirmed trigger means near-certainty of an ongoing SEP event. From the distribution of lead times, they propose a universal lower bound of about 30 minutes as the time a warned astronaut has to reach shelter while still avoiding the peak dose rate and more than 90% of the total SEP dose.","pith_inferences":["The same thresholding logic should transfer to any tissue-equivalent dosimeter with low statistical noise; the paper's own comparison of dose rates B and E shows that the method fails when the detector has poor statistics, so the practical precondition is a detector with a large geometric factor.","The two-measurement confirmation deliberately costs one 15-minute cadence of warning time; in a real deployment, a shorter measurement interval (e.g., one-minute data where available) could recover most of that lost time while keeping the false-alarm benefit.","Because the lead-time metric is defined retrospectively from the total event dose, the 30-minute guarantee is a statement about the 21 historical events; a standalone prospective test on the next solar maximum would be needed to validate it in real time.","The system is a nowcast, not a forecast: it cannot warn before particles arrive, so its role is complementary to solar-observatory-based predictions, and mission design should still plan shelter access time as the binding constraint."],"forward_implications":["A Mars mission can adopt an operational rule that astronauts on extravehicular activity or surface traversals remain within roughly 30 minutes of a shelter, because the nowcast is designed to give at least that much warning for dose-significant events.","The system runs entirely on local dosimeter data, so it works in deep space and on the surface alike and serves as a fallback whenever solar-observation-based forecasts are unavailable or fail.","Shelter duration can be communicated automatically: astronauts stay sheltered until the dose rate drops back below the 25% threshold, which the paper's Table 1 shows can range from about 1.5 hours to more than 3 days.","For the largest recorded events, the avoided dose is operationally meaningful: about 8.9 mGy for the March 2012 deep-space event and about 1.56 mGy for the May 2024 surface event, the latter being roughly 1.6 times the annual public dose limit.","The threshold can be lowered to 15% in deep space and 18% on Mars if the small increase in Mars false-alarm rate is accepted, or kept at 25% for a single uniform rule."],"supporting_citations":[{"why":"Defines the RAD instrument and its measurement capabilities, including the dose-rate channels used here.","marker":"(Hassler et al., 2012)"},{"why":"Supplies the cruise-phase radiation measurements that document the deep-space SEP events and background.","marker":"(Zeitlin et al., 2013)"},{"why":"Describes the MSL mission context for the rover and the RAD measurement campaign.","marker":"(Grotzinger et al., 2012)"},{"why":"Characterizes the long-term Martian radiation environment and dose-rate variations that the background fit must absorb.","marker":"(Guo et al., 2021)"},{"why":"Provides the earlier warning-time analysis for Mars exploration that motivates the operational shelter-access limit.","marker":"(Posner & Strauss, 2020)"},{"why":"Documents the September 2017 Mars surface SEP event using RAD, among the events in the nowcast test set.","marker":"(Ehresmann et al., 2018)"},{"why":"Reviews SEP prediction model accuracy, establishing the need for a last-resort nowcast.","marker":"(Whitman et al., 2023)"}],"fun_headline_variants":["Mars dose spike gives 30-minute SEP warning","Dose rate triggers nowcast Mars solar storms","Zero false alarms: Mars SEP nowcast via dose","30-minute heads-up: Mars SEP from local dose","Deep-space SEP alert from dose rates alone"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 30-minute guarantee assumes that a real-time warning sounded the moment dose rate rises 25% above background will always come at least 30 minutes before 10% of the eventual event dose has accumulated, because the authors derive that bound from the timing of 21 past events and set aside the several historical events with shorter lead times as too minor to matter.","fun_headline_variants_meta":{"raw":{"variants":["Mars dose spike gives 30-minute SEP warning","Dose rate triggers nowcast Mars solar storms","Zero false alarms: Mars SEP nowcast via dose","30-minute heads-up: Mars SEP from local dose","Deep-space SEP alert from dose rates alone"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000288,"raw_usage":{"total_tokens":1717,"prompt_tokens":1003,"completion_tokens":714,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":619,"completion_tokens_details":{"reasoning_tokens":642}},"tokens_in":619,"tokens_out":714,"duration_ms":7750,"temperature":1.0,"reasoning_tokens":642,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T12:03:04.554763+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find or record an SEP event with a significant total dose (comparable to the ones flagged as threatening, e.g., above roughly 100 µGy background-removed) whose lead time, measured from the 25%-above-background trigger to accumulation of 10% of the total SEP dose, is less than 30 minutes; or record any two-consecutive-measurement trigger above the 25% threshold that is not followed by a real SEP event, which would break the claimed zero false-alarm rate.","supporting_citations":[{"cited_title":", Zeitlin, C","cited_arxiv_id":null,"evidence_quote":"Defines the RAD instrument and its measurement capabilities, including the dose-rate channels used here."}],"review_version":1}