{"id":"f7212fa9-0fba-409a-8e1e-a74d5a665804","arxiv_id":"1908.05482","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Diffuse glow around cities in nighttime satellite imagery is real atmospheric scattering of artificial light, not sensor blooming.","lead":"Nighttime satellite images show a fuzzy glow around bright cities, and that glow was long written off as a camera artifact. This paper compares ground-based sky brightness measurements with four satellite systems and concludes the glow is real city light scattered by the atmosphere.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central scattering claim rests on an asserted up/down symmetry and correlation evidence; without radiative transfer modeling, the identification of satellite glow as ground skyglow is not established.","rationale":"The reader's conditional verdict is appropriate, and the weakest assumption identified by the reader matches my own: the inference depends on comparison pixels being free of direct emission and on the ground and space signals sharing a common scattering source. The paper deserves credit for its strongest positive evidence: the persistence of the glow across multiple sensors with different optics and resolutions (DMSP, VIIRS DNB, ISS) argues strongly against instrumental blooming, and the residual distribution is a plausible signature of a clean skyglow population mixed with direct-light pixels. The concern is not that the claim is false, but that the causal identification is under-supported. The 'Physical base' paragraph asserts up/down symmetry rather than testing it, and the regression evidence alone does not exclude confounders such as distance-to-city gradients or low-level direct emissions. A radiative transfer closure test would settle whether the correlation actually reflects scattering; until then, CONDITIONAL remains the right verdict. I would not move to reject, because the multi-sensor correlation is real and the physical argument is reasonable, only unverified.","tokens_in":9580,"tokens_out":6650,"duration_ms":78214,"concrete_test":"Perform a 3D radiative transfer simulation over the Madrid region for one VIIRS DNB overpass (e.g., May 2014) using the actual city light source map from the UCM survey, a published angular emission distribution, and the aerosol optical depth and vertical profile for the overpass date. Simulate both the top-of-atmosphere upwelling radiance at the dark pixels and the zenith sky brightness at the SQM stations, then compare both to the observed data. If a single set of parameters reproduces both observed quantities within their uncertainties (e.g., about 0.3 mag in sky brightness and the corresponding DNB radiance), the scattering interpretation is confirmed; if the required parameters are implausible or the two predictions decouple, the correlation does not establish that the satellite glow is the same skyglow detected from the ground.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that diffuse glow around cities in satellite imagery is atmospheric scattering, not sensor blooming. The load-bearing step is in the Methods section 'Physical base', where up/down symmetry of Rayleigh and Mie scattering for horizontal light is invoked to equate ground zenith sky brightness with satellite upwelling radiance. This symmetry is asserted, not modeled. The empirical evidence is a regression between UCM SQM values and radiance in dark satellite pixels (Figs. 3-5). That correlation is consistent with scattering, but also with both variables sharing a common distance-to-city gradient, or with low-level direct emission contaminating nominally 'unlit' pixels. The paper's double-Gaussian residual interpretation and the brightness thresholds (19 mag/arcsec2 for DMSP, 18.8 for ISS) are post hoc ways of selecting a 'skyglow-only' population; they do not independently establish that those pixels contain no direct light. The World Atlas comparison is partly circular because the UCM data were used to calibrate the Atlas. If the symmetry assumption fails, for example because city emission is not predominantly horizontal or aerosols have vertical gradients and forward-peaked phase functions, then the ground-space correlation could be real yet not demonstrate the same upward-scattered skyglow; the central claim would be unproven.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates the diffuse glow surrounding cities in nighttime satellite imagery, often dismissed as instrumental 'blooming.' The authors compare ground-based zenith sky brightness measurements from the Universidad Complutense Madrid (UCM) SQM survey with satellite radiance from four very different imaging systems: DMSP (5 km), VIIRS DNB (750 m), and two ISS astronaut-photograph HDR composites (240 m and 54 m). They find strong correlations between ground sky brightness and satellite radiance in pixels away from directly lit areas, interpret the diffuse glow as upward-scattered artificial skyglow rather than a sensor artifact, and propose that future space radiometers could monitor artificial skyglow. They also compare VIIRS-derived sky brightness with the World Atlas of artificial sky brightness and discuss other contributors to diffuse light such as surface albedo, airglow, fog, and residual blooming.","tokens_in":9906,"tokens_out":6580,"duration_ms":58386,"significance":"If established, the paper's central claim would change how the diffuse glow in nighttime satellite imagery is interpreted and would open the door to space-based monitoring of artificial skyglow. The study has real strengths: it uses four independent sensors with very different point-spread functions, all showing a consistent correlation with independent ground-based SQM data, which is strong evidence that the glow is a real optical signal rather than a single-sensor artifact. The paper is clearly written, uses public data, and candidly acknowledges that not all diffuse light is skyglow. However, the identification of the glow specifically as atmospheric scattering of city light rests on an unquantified symmetry assumption and on post hoc pixel-selection choices, and the World Atlas comparison is partly circular because the same UCM survey was used in the Atlas calibration. These issues prevent the paper, in its current form, from fully supporting the strong claim in the abstract.","major_comments":[{"comment":"The central inference of the paper depends on the assertion that 'because of the up/down symmetry of Rayleigh and Mie diffusion for light traveling horizontally, it is reasonable to suppose that the same amount of light that is scattered to the ground from the zenith is also scattered towards space from the same volume of atmosphere.' This is an assumption, not a derivation. The relationship between ground zenith sky brightness and space-borne upwelling radiance depends on the angular distribution of city light emission, the aerosol phase function (which is strongly forward-peaked for Mie scattering), the vertical profile of scatterers, and the observation geometry. The paper does not provide a radiative transfer calculation or a quantitative model that connects the two quantities for Madrid's source distribution. Without such a calculation, the observed correlation could also arise from a common distance-to-city gradient or from low-level direct emission in the nominally 'unlit' pixels, so the abstract's claim that the glow is 'a real detection of light scattered by the atmosphere' is not yet established by the evidence presented.","section":"Methods, 'Physical base'"},{"comment":"The linear fits that define the ground-space relationship are made after rejecting bright bins: the caption of Fig. 3 says the fit uses 'the darkest values and rejecting the bins brighter than 19 mag/arcsec2,' and Fig. 4 uses 18.8 mag/arcsec2. These thresholds are post hoc, and the double-Gaussian decomposition of the residuals (Fig. 6) attributes the second component to direct light contamination without an independent test. Because the regression itself defines the expected sky brightness, the residual analysis is partly circular. The authors should show that the correlation and the width of the main residual peak are robust to (a) varying the brightness threshold, (b) excluding the pixels closest to the city center, and (c) including distance to the city as a control variable. Without such checks, the selection of a 'skyglow-only' population is not justified.","section":"Results, Figs. 3-5 and Fig. 6"},{"comment":"The comparison with the World Atlas model (Fig. 13) is not an independent validation, because the manuscript itself states in the Methods that 'The UCM sky brightness survey was used for the calibration of the \"World Atlas of artificial sky brightness\".' The apparent 'perfect fit' between VIIRS-derived values and the Atlas therefore contains a circular component. A stronger test would use a sky-brightness model or ground survey that was not part of the Atlas calibration, or a first-principles radiative transfer model driven by the observed VIIRS radiance field.","section":"Results, 'Comparison with the World Sky atlas' and Methods"},{"comment":"The paper describes the evidence as 'strong correlation' and 'similar dispersion' but does not report the coefficient of determination, fit parameters, uncertainties, or sample sizes for the regressions in Figs. 3-5. These numbers are necessary for the reader to assess the strength of the central correlation and to support the claim that the dispersion is similar across four sensors with very different PSFs. Visual inspection of 2D histograms is not sufficient for a quantitative claim of this kind.","section":"Results, 'Relationship between sky brightness and diffuse light'"}],"minor_comments":[{"comment":"The manuscript refers to the diffuse glow as 'blooming' but later distinguishes residual instrument blooming (Fig. 15); the terminology should be clarified early so that 'blooming' is not used for two different phenomena.","section":"Throughout"},{"comment":"The statement 'The changes in fit are produced by the different mean composition of the atmosphere when the image was taken' is speculative, since no atmospheric composition data are presented for the imaging epochs.","section":"Discussion"},{"comment":"The sentence 'We can conclude that the radiance observed in the regime between 0.2-5 nW/cm2/sr is dominated by diffuse light' introduces a specific radiance range without explaining how the endpoints were determined.","section":"Discussion"},{"comment":"There is a typographical error in the sentence beginning 'Other components such as natural sky glow...': 'in other to extract' should be 'in order to extract'.","section":"Discussion"},{"comment":"The choice of the months 'closest to the four months second equinox (Aug to Nov)' and the averaging over 2012-2018 should be justified, as seasonal and interannual changes in VIIRS data are known to occur.","section":"Methods and Fig. 13"}],"recommendation":"major_revision","confidential_remarks":"The multi-sensor correlation is compelling evidence that the glow is a real optical signal, and the paper is likely publishable after revision. However, the identification of the glow specifically as atmospheric scattering of city light currently rests on an unquantified symmetry argument and post hoc selections. I would require either a radiative transfer calculation connecting ground and space radiance for the Madrid region, or a clear reframing of the conclusion as 'consistent with atmospheric scattering' rather than a demonstrated detection. The circularity in the World Atlas comparison should also be acknowledged explicitly in the text."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper makes a strong observational case that the diffuse glow around cities in nighttime satellite images is real scattered light, not sensor blooming. The multi-sensor correlation with ground SQM data is convincing evidence that the glow tracks sky brightness. But the causal story—that the upwelling radiance seen from space is the same physical population as the downwelling skyglow seen from the ground—rests on a symmetry argument that is asserted, not derived. That is the soft spot, and it deserves to be fixed before publication.\n\nWhat's actually new: earlier work (Zamorano et al.) raised the possibility and showed a DMSP relationship; this paper extends that to VIIRS DNB and two ISS HDR composites, with a consistent ground survey. That is a real step. The residual analysis with the double Gaussian is a sensible way to separate 'skyglow-only' pixels from pixels with direct light, and the fact that the main residual width is similar across sensors with very different PSFs is a good sign. The section acknowledging albedo, airglow, fog, and a residual real blooming component shows the authors are not overclaiming.\n\nThe weaknesses are in proportion. The 'Physical base' Methods section says the up/down symmetry of Rayleigh and Mie scattering for horizontal light makes it 'reasonable to suppose' ground and space signals are related. That is plausible but not quantified; no radiative transfer model or angle-dependent scattering calculation is presented. The brightness cuts (19 mag/arcsec2 for DMSP, 18.8 for ISS) are post hoc, used to select the 'dark' population, and the fits are presented without error bars. The World Atlas comparison is partly circular because the UCM data were used to calibrate that atlas. None of this kills the paper's main observational contribution, but it means the identification of the space signal as skyglow—rather than a distance-to-city gradient or dim direct emission—is not as airtight as the abstract implies.\n\nWho should read it: light pollution researchers and anyone using VIIRS dark pixels as a proxy for sky brightness. It deserves a serious referee; with added modeling and a more transparent selection procedure it could be a solid methods paper. I would send it to review rather than desk reject.","headline":"A convincing multi-sensor correlation that the diffuse glow around cities is real scattered light, but the causal identification needs radiative transfer support and more transparent fitting choices.","tokens_in":10417,"tokens_out":1871,"would_cite":true,"duration_ms":18372,"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":"The diffuse glow around cities in nighttime satellite imagery is real light scattered by the atmosphere, not an instrumental artifact.","keywords":["light pollution","skyglow","nighttime satellite imagery","atmospheric scattering","VIIRS Day/Night Band","DMSP","ISS astronaut photography","urban remote sensing"],"falsifier":"Measure the diffuse glow in unlit pixels near a city with VIIRS while simultaneously measuring ground-level zenith sky brightness, under known aerosol conditions; if the two are uncorrelated once distance to the city is removed, or if the glow can be reproduced by a surface albedo model with no atmospheric scattering, the central claim collapses.","tokens_in":9382,"feed_emoji":"🌆","tokens_out":5410,"duration_ms":51475,"temperature":0.7,"pith_summary":"This paper argues that the diffuse glow surrounding bright cities in nighttime satellite images is not a sensor artifact, but real city light scattered upward by the atmosphere. The authors compare ground-based measurements of artificial sky brightness around Madrid with satellite radiance from four imaging systems with very different resolutions, and find a strong correlation in unlit areas. They conclude that the same scattered light that produces skyglow for a ground observer also produces the diffuse halo seen from space. If correct, the dark pixels of existing and future satellite images become a usable monitor of changes in artificial skyglow.","feed_headline":"Satellite city glow is real scattered light, not sensor blooming","feed_subtitle":"Ground sky-brightness measurements match satellite radiance, so orbit can track light pollution.","key_machinery":"The mechanism is atmospheric scattering, specifically the up/down symmetry of Rayleigh and Mie scattering for near-horizontal light. The paper's central identity is that the same atmospheric volume that scatters city light downward to a ground observer also scatters a nearly equal amount upward to space, so satellite radiance in unlit pixels is an upward view of the same skyglow. This operates through logarithmic regressions between ground-based zenith sky brightness and space-based radiance, with residual analysis separating pixels containing only scattered light from pixels mixed with direct emissions.","core_discovery":"The paper shows that the diffuse light around cities in nighttime satellite imagery is mostly skyglow propagated from urban areas, not detector blooming. Because Rayleigh and Mie scattering are nearly up/down symmetric for the near-horizontal light that dominates ground-level skyglow, light scattered upward toward space and light scattered downward to the ground share the same source. This is supported by correlations between ground sky-brightness data and radiance from DMSP, VIIRS Day/Night Band, and two composites of ISS astronaut photographs; the main residual component has a half width of about 0.35 mag per square arcsecond, roughly a 35 percent radiance difference. A second residual component is attributed to pixels that mix direct ground emissions with scattered light, while the diffuse signal itself is most visible in the darkest pixels.","pith_inferences":["If the correlation holds globally, archived DMSP composites contain a decades-long record of skyglow changes that predate dedicated ground-based skyglow surveys.","The up/down symmetry could be inverted: ground-based skyglow measurements may serve as a calibration for upward-directed light flux, connecting satellite radiance to light-pollution inventories.","Pairing space-derived skyglow with simultaneous aerosol measurements could yield a nighttime aerosol optical thickness proxy, extending daytime remote sensing to dark hours.","The identified pixel-mixing tail implies that low-resolution nighttime-light products systematically blur the boundary between direct emissions and skyglow, so skyglow-free products must either preserve high resolution or subtract a scattering model."],"forward_implications":["A future space-based imaging radiometer could monitor changes in the diffuse artificial skyglow of cities from orbit.","Dark pixels near cities in VIIRS and similar data can be used to build regional or global sky-brightness maps, complementing purely modeled skyglow products.","Changes in street-lighting intensity or spectra that alter ground-observed skyglow should appear as corresponding changes in the diffuse glow around cities.","Satellite radiance in the diffuse regime around cities should be treated as scattered light rather than surface emission, and subtracting a sky-brightness model can isolate direct ground emissions.","The consistency of residuals across four independent imagers indicates that low-resolution composites mix direct and scattered light, so high-resolution data are preferable for skyglow monitoring."],"supporting_citations":[{"why":"Describes the physical behavior of anthropogenic light propagation into the nocturnal environment, establishing scattering as the mechanism behind skyglow.","marker":"[8]"},{"why":"Supplies the improved scattering formula used in sky-brightness models, grounding the expectation of approximate up/down symmetry.","marker":"[13]"},{"why":"Shows that ground-observed skyglow is dominated by light emitted at near-horizontal angles, the key input to the symmetry argument.","marker":"[16]"},{"why":"Provides the ground-based sky brightness survey data around Madrid and earlier suggestion that the satellite diffuse light could be connected to skyglow.","marker":"[17]"},{"why":"Provides the earlier citizen-science comparison between skyglow and DMSP data, including the residual baseline that this paper improves upon.","marker":"[18]"},{"why":"Documents the high dynamic range processing of ISS astronaut images used for the high-resolution comparisons.","marker":"[22]"},{"why":"Bounds the effect of different spectral responses among the sensors used in the comparison.","marker":"[25]"},{"why":"Supports the reality of the diffuse emission through its observed relationship to aerosol optical thickness.","marker":"[30]"},{"why":"Provides the sky-brightness model against which the VIIRS-derived sky brightness values are validated.","marker":"[31]"}],"fun_headline_variants":["City glow in satellite images is real skyglow, not sensor bloom","Nighttime city halo is atmospheric light, not camera artifact","Satellites see true city glow, not detector blooming","Diffuse city light is scattered skyglow, not sensor error","Orbital glow matches ground sky brightness, proving real light"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the unlit pixels used for comparison contain only scattered skyglow, with no direct ground emissions, and that the atmosphere scatters near-horizontal city light about equally upward and downward.","fun_headline_variants_meta":{"raw":{"variants":["City glow in satellite images is real skyglow, not sensor bloom","Nighttime city halo is atmospheric light, not camera artifact","Satellites see true city glow, not detector blooming","Diffuse city light is scattered skyglow, not sensor error","Orbital glow matches ground sky brightness, proving real light"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000436,"raw_usage":{"total_tokens":2162,"prompt_tokens":832,"completion_tokens":1330,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":448,"completion_tokens_details":{"reasoning_tokens":1243}},"tokens_in":448,"tokens_out":1330,"duration_ms":10178,"temperature":1.0,"reasoning_tokens":1243,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:13:02.059386+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the diffuse glow in unlit pixels near a city with VIIRS while simultaneously measuring ground-level zenith sky brightness, under known aerosol conditions; if the two are uncorrelated once distance to the city is removed, or if the glow can be reproduced by a surface albedo model with no atmospheric scattering, the central claim collapses.","supporting_citations":[{"cited_title":"Physical behaviour of anthropogenic light propagation into the nocturnal environment","cited_arxiv_id":null,"evidence_quote":"Describes the physical behavior of anthropogenic light propagation into the nocturnal environment, establishing scattering as the mechanism behind skyglow."},{"cited_title":"Improved scattering formula for calculations of artificial night-sky illumination","cited_arxiv_id":null,"evidence_quote":"Supplies the improved scattering formula used in sky-brightness models, grounding the expectation of approximate up/down symmetry."},{"cited_title":"Campaign of sky brightness and extinction measurements using a portable ccd camera","cited_arxiv_id":null,"evidence_quote":"Shows that ground-observed skyglow is dominated by light emitted at near-horizontal angles, the key input to the symmetry argument."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the ground-based sky brightness survey data around Madrid and earlier suggestion that the satellite diffuse light could be connected to skyglow."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the earlier citizen-science comparison between skyglow and DMSP data, including the residual baseline that this paper improves upon."},{"cited_title":"Variacio´n espacial, temporal y espectral de la contaminacio´n lum´ınica y sus fuentes: Metodolog´ıa y resultados","cited_arxiv_id":null,"evidence_quote":"Documents the high dynamic range processing of ISS astronaut images used for the high-resolution comparisons."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Bounds the effect of different spectral responses among the sensors used in the comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supports the reality of the diffuse emission through its observed relationship to aerosol optical thickness."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the sky-brightness model against which the VIIRS-derived sky brightness values are validated."}],"review_version":1}