{"id":"9449adaf-5187-4622-8be7-655de7be061f","arxiv_id":"2608.11161","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Occultation of stars by LEO/GPS satellites and the Moon can yield <2 mas stellar angular diameter measurements for ~10^4 stars per year, given microsecond photon-counting detectors.","lead":"A new analysis shows that stars briefly eclipsed by artificial satellites or the Moon can be measured with sub-milliarcsecond angular resolution using fast photon-counting detectors. If it works, it would let astronomers measure the sizes of thousands of stars per year without giant telescopes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Real satellite edges, not half-planes: unmodelled shape/orientation effects can bias the fringe-derived stellar diameters at the claimed sub-mas precision.","rationale":"The paper is a careful feasibility study; the statistical pipeline and the resolution definition are internally consistent, and the event-rate estimates are transparently approximate. I agree with the reader's conditional verdict. In stress-testing the chain, the least secure link is not the photon statistics or the Bayesian machinery but the physical model of the mask. The central claim requires that the observed fringe pattern be the half-plane pattern up to a few percent, because the stellar-diameter signal is itself only a few percent of the fringe amplitude. For MEO, the Fresnel scale in the mask plane is about 2 m, so metre-scale structural details of a GPS satellite cannot be ignored without proof. The paper cites reference [44] for the straight-edge approximation, but that reference concerns lunar and asteroid occultations, where the limb is smooth over many Fresnel zones; it does not cover a structured spacecraft. This is an external-fidelity concern rather than an internal inconsistency, so it does not refute the proposal but makes it conditional. I would not move the reader's verdict; a pilot observation or a public realistic-diffraction code release would be the natural condition. If the shape test shows bias comparable to sigma_theta, the '10^1 to 10^2 stars to 0.5 mas' claim and the sub-2 mas rates for satellite events would need revision. The Moon-only path is less affected because the lunar limb is a much cleaner straight edge, but the satellite path is the novel part of the central claim.","tokens_in":19851,"tokens_out":21086,"duration_ms":215047,"concrete_test":"Take a public CAD model of a GPS Block IIF or Starlink v2 satellite and compute the full near-field Fresnel diffraction pattern for a representative occultation geometry, including panel gaps and mast shadows. Generate Poisson light curves at the n0 values used in Figure 5, with the silhouette orientation frozen and also rotating across the roughly 100 microsecond fringe window. Run the paper's straight-edge pipeline on these light curves and compare the recovered theta* to the injected value. If the bias exceeds the sigma_theta reported in Figure 5 (approximately 0.3 to 0.5 mas), the straight-edge assumption is the limiting error and the headline numbers must be re-derived; if the bias is sub-dominant, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 2.1 models every satellite as an opaque, locally straight half-plane (Eq. 1), and the entire inference of theta* in Section 3 uses this single-edge Fresnel pattern. The load-bearing requirement is that the diffracting edge is straight and opaque over several Fresnel zones. For the MEO case that yields the best results, D_F = sqrt(lambda * D_OM / 2) is approximately 2.2 m at D_OM = 2 x 10^4 km, comparable to the size, gaps, and appendages of a GPS satellite. A real satellite is a finite, partly transparent, multi-edged body; its Fresnel pattern is a coherent superposition of edge waves from several boundaries, not the half-plane pattern. Orientation changes during the event alter the projected silhouette. The claimed 0.5 mas diameter precision corresponds to a roughly 1% fringe-amplitude change for an approximately 5 mas star at the n0 = 30 photons per microsecond rate used in Figure 5, and unmodelled 3D structure can plausibly produce such amplitude distortions. Without quantifying this, the reported theta_res and sigma_theta values are conditional on an idealized mask.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Fung et al. propose using high-speed photon-counting detectors to record Fresnel diffraction patterns during occultations of bright stars by artificial satellites (Starlink in LEO, GPS-like satellites in MEO) and by the Moon, and to infer stellar angular diameters below the telescope diffraction limit. They derive a straight-edge Fresnel model (Eq. 1), define the angular resolution as the 95% upper bound of the posterior for a point source (Eq. 7), simulate Poissonian light curves for MEO occultations, validate a Bayesian inference pipeline on injected diameters (Fig. 5), and compute event rates for LEO, MEO, and lunar masks (Eqs. 11-13). The headline claims are <2 milliarcsecond resolution for ~10^4 stars per year, with ~10^1-10^2 stellar diameters constrained to better than 0.5 milliarcseconds, using SPINA-type detectors with ~1 microsecond timing.","tokens_in":20081,"tokens_out":8820,"duration_ms":90315,"significance":"If the idealized edge model is adequate, this is a genuinely clever and potentially transformative route to population-scale stellar angular diameters, complementary to intensity interferometry and far cheaper than constructing very large diffraction-limited telescopes. The Bayesian treatment of the photon-starved regime, the explicit statistical definition of resolution, and the validation with injected signals are clear strengths. However, the significance of the artificial-satellite results is conditional on treating each satellite as an opaque straight half-plane; the paper does not yet demonstrate this for real spacecraft. The lunar-occultation component is on much firmer ground, since the lunar limb is a long-established straight-edge approximation.","major_comments":[{"comment":"The half-plane Fresnel model is not justified for artificial satellites. For GPS-like MEO, the Fresnel scale D_F = sqrt(lambda*D_OM/2) is about 2 m at lambda = 500 nm, comparable to the bus, solar panels, and gaps of a navigation satellite; for LEO, D_F is roughly 0.3-0.4 m, comparable to Starlink dimensions. A finite, multi-edged, partially transparent screen produces a coherent superposition of edge waves rather than the single-edge pattern of Eq. (1), and the projected silhouette changes with attitude on the ~100 microsecond timescale of the event. Since the claimed <0.5 mas precision corresponds to a small change in fringe amplitude, unmodelled three-dimensional structure can plausibly bias the inferred stellar diameter. The authors should simulate Fresnel diffraction from realistic satellite silhouettes (or at least from rectangular or gapped screens) over a range of orientations and show that the inferred diameter bias is below the quoted precision.","section":"§2.1, Eq. (1); Table 1"},{"comment":"The validation pipeline assumes theta(t) = omega*t with the angular speed known from external tracking, and it does not marginalize over the projected edge velocity or orientation. Section 2.2 notes that omega*cos(phi) can be determined empirically, but this is not implemented in the likelihood (Eq. 8) or in the bias check (Fig. 5). For satellite events, an error in the assumed angular rate or a time-varying attitude directly shifts the phase of the diffraction fringes and can mimic the amplitude suppression that constrains theta_*. The reported sigma_theta values are therefore conditional on exactly known kinematics; a sensitivity analysis or marginalization over omega*cos(phi) is needed before the quoted precision can be taken as realistic.","section":"§3.1, Eq. (8); Fig. 5"},{"comment":"The event-rate calculation assumes circular non-overlapping orbits and multiplies the single-satellite swept solid angle by N_mask. Real LEO constellations have many satellites in shared orbital planes, so track overlap can be substantial, and visibility effects such as Earth shadow, sky background near the Moon, and satellite illumination are not included. The authors correctly label this as a coarse estimate, but the abstract's '~10^4 stars every year' and the cumulative rate curves in Fig. 8 are headline numbers. A rate estimate using current ephemerides and accounting for orbit overlap would calibrate the claim; alternatively the numbers should be presented as upper limits or order-of-magnitude estimates.","section":"§4.2, Eqs. (12)-(13); Fig. 8"}],"minor_comments":[{"comment":"The phase-misalignment term 'pi*theta_*^2/8' is dimensionally inconsistent as written; it should be pi*theta_*^2/(8*theta_F^2) in angular units, or the equation should be expressed using normalized variables x = theta/theta_F and a = theta_*/theta_F.","section":"§2.1, Eq. (5)"},{"comment":"The statement that occultation resolution is independent of wavelength relies on a flat SED with f_lambda proportional to lambda and on a cancellation between the Fresnel-scale growth and photon-counting SNR; this should be presented as a special illustrative case, not as a general property of the method.","section":"§2.3"},{"comment":"The text contains a typo: 'Kelper's law' should read 'Kepler's law'.","section":"§2.2"},{"comment":"The middle panel states that bias is consistent with zero, but no error bars or numeric bias values are shown; please include the measured bias and scatter so the reader can verify the claim.","section":"Fig. 5"},{"comment":"The source code is available only 'on reasonable request'; since the paper presents reproducible simulations, making the code public would strengthen the manuscript.","section":"Code and Data Availability"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the straight-edge model for artificial satellites; if the authors cannot demonstrate robustness to realistic satellite silhouettes and attitude variations, the artificial-satellite claims should be substantially softened. The lunar occultation component is more secure. The paper fits the scope of astro-ph.IM and is potentially publishable after the mask-fidelity and kinematic-uncertainty issues are addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a solid feasibility study with one load-bearing assumption that worries me. The idea—using artificial satellites as occulting masks to measure stellar angular diameters via Fresnel diffraction—is genuinely new, and the authors do the work properly: clear formalism, a sensible statistical resolution definition, a Bayesian pipeline that recovers injected diameters without bias, and realistic detector parameters from SPINA. The event-rate estimates are a useful first cut. The lunar occultation part is on well-trodden ground.\n\nThe soft spot is the straight-edge model for satellites (Section 2.1). The paper cites the standard straight-edge approximation, but that is derived for the Moon, where the Fresnel zone D_F ~ 0.01 m is tiny compared with the lunar limb. For GPS-class MEO satellites, D_F ~ 2 m, comparable to the satellite body, solar panels, and gaps. A real satellite is a finite, multi-edged, partly transparent object whose projected silhouette changes as it rotates. The diffraction pattern is a coherent sum of edge waves, not the half-plane pattern. The claimed 0.5 mas precision corresponds to roughly a 1% fringe-amplitude change at the photon rates used in Fig. 5; unmodelled shape features can plausibly produce distortions of that size. The authors note the straight-edge condition but never quantify it for satellites. This does not kill the lunar occultation results, and it may be fixable by modeling actual satellite shapes (or arguing that the first edge dominates and is straight over many D_F), but without that, the satellite resolution and rate numbers are conditional on an idealized mask.\n\nMinor points: the event-rate calculation assumes circular non-overlapping orbits and an approximate star-count calibration; acceptable for a first estimate but worth tightening. The code is 'available on reasonable request'; for a simulation paper, that is a reproducibility barrier.\n\nBottom line: the paper deserves a serious referee. I would send it out, with the instruction that the straight-edge assumption needs a quantitative check or a revised claim. If the authors can show that the multi-edge structure does not bias the inferred diameters, this becomes a genuinely useful proposal. As it stands, it's a promising feasibility study whose headline numbers should be taken with a grain of salt.","headline":"A promising feasibility study for measuring stellar diameters via satellite occultations, but the headline numbers rest on an unvalidated straight-edge model for satellites that needs a careful check.","tokens_in":20573,"tokens_out":4595,"would_cite":true,"duration_ms":43746,"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":"Eclipses of bright stars by artificial satellites and the Moon can measure stellar angular diameters below the optical diffraction limit, down to about 0.1 milliarcseconds for the brightest targets.","keywords":["occultation imaging","stellar angular diameters","Fresnel diffraction","photon counting","artificial satellite constellations","lunar occultation","super-resolution astronomy","Bayesian inference"],"falsifier":"Occult a bright star whose angular diameter is already known from long-baseline interferometry with a variety of satellites, using a microsecond photon counter; if the recovered diameters disagree between satellites, or if a point-like star produces fringes that deviate from the straight-edge Fresnel prediction, the edge model is falsified. A tabletop version would diffract a laser around a razor edge with added panel gaps and test whether the predicted high-order fringes survive.","tokens_in":19658,"feed_emoji":"🛰️","tokens_out":10351,"duration_ms":88778,"temperature":0.7,"pith_summary":"Most stars are too small and distant for even the largest telescopes to resolve, because the diffraction limit of a 10-metre telescope sits at tens of milliarcseconds at optical wavelengths. This paper argues that watching a star wink out behind a fast-moving foreground object bypasses that limit, because the star's angular size is encoded in the time-dependent diffraction pattern of the occulting edge. Using artificial satellites in low and medium Earth orbit, and the Moon, the authors estimate that sub-2-milliarcsecond resolution becomes available for roughly $10^4$ stars per year, with a few dozen to a hundred stellar diameters measured to better than 0.5 milliarcseconds. The practical requirement is microsecond-level photon counting, which fast silicon photomultiplier instruments have already demonstrated.","feed_headline":"Satellite eclipses can measure star sizes below the diffraction limit","feed_subtitle":"Fast photon timing turns passing satellites into sub-milliarcsecond measurements for thousands of stars.","key_machinery":"The load-bearing object is the Fresnel diffraction integral for a straight-edged mask, which acts as a 'negative point-spread function': instead of forming an image, the occulting edge blocks light and imprints a damped oscillatory fringe pattern on the total detected flux. The angular Fresnel scale $\\theta_F = \\sqrt{\\lambda D_{\\mathrm{OM}}/2}/D_{\\mathrm{OM}}$ sets the fringe spacing, and a star's angular diameter $\\theta_*$ suppresses the fringe amplitudes through an amplitude-suppression factor and a small phase shift, approximated by a WKB expression. This converts spatial structure on the stellar disc into a temporal signal that an ultra-fast photon counter can record, with the fringe period fixing $\\theta_F$ and the fringe damping fixing $\\theta_*$.","core_discovery":"The paper's central claim is that occultation imaging is no longer a rare-event technique: the thousands of artificial satellites now in orbit make it a population-scale survey tool. When a satellite edge crosses the line of sight, it casts a Fresnel diffraction pattern that sweeps past the observer on microsecond timescales, and the finite angular diameter of the background star damps the amplitude of the higher fringes in that pattern. The authors simulate photon-counting observations, analyse them with a Bayesian model that handles Poisson shot noise, and show that stellar diameters are recovered without bias. In their estimates, navigation satellites in medium Earth orbit deliver sub-milliarcsecond resolution for bright stars and outperform lunar occultations once the background star is dimmer than roughly tenth magnitude, while low-Earth-orbit satellites provide hundreds to thousands of intermediate-resolution events per night. The resolution limit is set by photon shot noise and the angular Fresnel scale, not by telescope aperture.","pith_inferences":["Beyond the paper: if real satellites are not clean straight edges, the predicted fringe pattern would be corrupted; modelling each satellite as a compound mask with panels, gaps, antennas, and orientation would yield concrete predictions for how recovered diameters and event rates change.","Beyond the paper: the same time-domain logic applies to any artificial object with a known ephemeris, including debris and future constellations, so the survey power of the method will grow as the orbital population changes.","Beyond the paper: pairing these angular diameters with existing parallax measurements would turn the method into a physical-radius census for supergiants and binary stars, which is where the population-level payoff would be largest."],"forward_implications":["Occultations by medium-Earth-orbit navigation satellites yield about ten events per night at resolutions better than roughly 3 milliarcseconds, independent of lunar phase.","Lunar occultations reach about 0.1 milliarcsecond resolution for the brightest stars, at a rate of roughly once per week; navigation satellites take over for fainter stars, where the Moon's scattered light dominates the noise.","Low-Earth-orbit satellite constellations can in principle generate more than one hundred thousand sub-diffraction-limited events per day, with the practical rate limited by telescope pointing overhead to a few hundred per night.","The required detector speed already exists in demonstrated fast photon-counting imagers with nanosecond-scale timing, and in large Cherenkov telescope arrays.","Because precision scales with photon shot noise and the Fresnel scale, the method's resolution depends only weakly on aperture size and wavelength, so it complements rather than replaces diffraction-limited telescopes."],"supporting_citations":[{"why":"supplies the straight-edge approximation for the mask geometry, the basis of the Fresnel diffraction model.","marker":"[44]"},{"why":"demonstrates the nanosecond-scale inter-frame delay of a fast photon-counting imager, establishing that the required detector speed exists.","marker":"[41]"},{"why":"shows the dark count rate of that imager is negligible over a typical occultation timescale.","marker":"[42]"},{"why":"provides photometric and trajectory characterization of low-Earth-orbit satellites, fixing the assumed angular speeds.","marker":"[45]"},{"why":"supplies the stellar luminosity function and star counts used to compute occultation event rates.","marker":"[46]"},{"why":"gives the brightness of the dark side of the Moon, used to model the dominant noise in lunar occultations.","marker":"[49]"},{"why":"demonstrates sub-0.1 milliarcsecond stellar diameter measurement by asteroid occultation with fast Cherenkov telescope arrays, the observational baseline this paper extends.","marker":"[36]"},{"why":"quantifies the low-Earth-orbit satellite population that sets the number of available masks.","marker":"[37]"}],"fun_headline_variants":["Satellite eclipses shrink star measurements to sub-milliarcsecond","Microsecond eclipses by satellites resolve stars beyond diffraction","Artificial satellite occultations yield precise stellar diameters","Satellite shadows map stellar sizes with microsecond timing","Eclipses by thousands of satellites enable star size survey"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire prediction rests on treating each satellite as a clean, opaque, straight-edged screen; a real satellite with solar panels, gaps, antennas, and changing orientation could distort the Fresnel fringe pattern from which the stellar diameter is read.","fun_headline_variants_meta":{"raw":{"variants":["Satellite eclipses shrink star measurements to sub-milliarcsecond","Microsecond eclipses by satellites resolve stars beyond diffraction","Artificial satellite occultations yield precise stellar diameters","Satellite shadows map stellar sizes with microsecond timing","Eclipses by thousands of satellites enable star size survey"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000726,"raw_usage":{"total_tokens":3259,"prompt_tokens":954,"completion_tokens":2305,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":570,"completion_tokens_details":{"reasoning_tokens":2227}},"tokens_in":570,"tokens_out":2305,"duration_ms":15134,"temperature":1.0,"reasoning_tokens":2227,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:59:25.226581+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Occult a bright star whose angular diameter is already known from long-baseline interferometry with a variety of satellites, using a microsecond photon counter; if the recovered diameters disagree between satellites, or if a point-like star produces fringes that deviate from the straight-edge Fresnel prediction, the edge model is falsified. A tabletop version would diffract a laser around a razor edge with added panel gaps and test whether the predicted high-order fringes survive.","supporting_citations":[{"cited_title":"Comparison of the Disk Diffraction Pat- tern with the Straight-Edge Diffraction Pattern in Occultations,","cited_arxiv_id":null,"evidence_quote":"supplies the straight-edge approximation for the mask geometry, the basis of the Fresnel diffraction model."},{"cited_title":"Initial on-sky performance testing of the single-photon imager for nanosecond astrophysics (spina) system,","cited_arxiv_id":null,"evidence_quote":"demonstrates the nanosecond-scale inter-frame delay of a fast photon-counting imager, establishing that the required detector speed exists."},{"cited_title":"False alarm rate-based statistical detection limit for astronomical photon detectors,","cited_arxiv_id":null,"evidence_quote":"shows the dark count rate of that imager is negligible over a typical occultation timescale."},{"cited_title":"Photometric charac- terization and trajectory accuracy of Starlink satellites: implications for ground-based astronomical surveys,","cited_arxiv_id":null,"evidence_quote":"provides photometric and trajectory characterization of low-Earth-orbit satellites, fixing the assumed angular speeds."},{"cited_title":"Gaia Data Release 2. Summary of the contents and survey properties,","cited_arxiv_id":null,"evidence_quote":"supplies the stellar luminosity function and star counts used to compute occultation event rates."},{"cited_title":"The colour of the dark side of the Moon,","cited_arxiv_id":null,"evidence_quote":"gives the brightness of the dark side of the Moon, used to model the dominant noise in lunar occultations."},{"cited_title":"Direct measurement of stellar angular diameters by the VERITAS Cherenkov telescopes,","cited_arxiv_id":null,"evidence_quote":"demonstrates sub-0.1 milliarcsecond stellar diameter measurement by asteroid occultation with fast Cherenkov telescope arrays, the observational baseline this paper extends."},{"cited_title":"The Low Earth Orbit Satellite Population and Impacts of the SpaceX Starlink Constellation,","cited_arxiv_id":null,"evidence_quote":"quantifies the low-Earth-orbit satellite population that sets the number of available masks."}],"review_version":1}