moira

Physical Heliacal Visibility Source Ledger

Date: 2026-07-31 Status: core source ledger; Phases 1-6 closed; site-specific moonlight branch quarantined Doctrine: PHYSICAL_HELIACAL_VISIBILITY_ADMISSION_DOCTRINE.md

Supersession notice (2026-08-07): Jones/Paranal entries below document historical source research only. The experiment has no runtime, public API, packaged resource, release gate, or active-roadmap dependency. See the quarantine decision.

Purpose

This ledger identifies the sources allowed to govern the physical heliacal-visibility project, the role of each source, its validity boundary, and whether any code or data may cross into Moira artifacts.

It prevents formulas, coefficients, event meanings, or data identities from being reconstructed from memory.

Access date for every online source below: 2026-07-29 unless separately stated. The REPTRAN module was accessed on 2026-07-30. The Jones solar, ROLO, and Oxford EODG input authorities were accessed on 2026-07-31.

Governing Sources

Source Exact identity Project role Validity and use boundary License or distribution disposition
Kasten and Young, “Revised optical air mass tables and approximation formula” Applied Optics 28 (1989), 4735–4738, DOI 10.1364/AO.28.004735 Preserve and regression-test the current relative-air-mass option Apparent above-horizon optical air mass; not the new directional radiance generator Equation reference only; no table or figure redistribution
Schaefer, “Telescopic Limiting Magnitudes” PASP 102 (1990), 212–229, DOI 10.1086/132629 Preserve the lineage of the current named limiting-magnitude and extinction family Historical composite visibility family; not silently promoted to the new physical baseline Equation/reference use only; no copied paper content
Schaefer, “Astronomy and the Limits of Vision” Vistas in Astronomy 36 (1993), 311–361, DOI 10.1016/0083-6656(93)90113-X Preserve the current named directional twilight and component lineage as a regression family Review/model family spanning multiple applications; every retained equation keeps its existing admitted domain Equation/reference use only; no copied paper content
Krisciunas and Schaefer, “A Model of the Brightness of Moonlight” PASP 103 (1991), 1033–1039, DOI 10.1086/132921 Preserve the current named legacy moonlight option Empirical photometric moonlight model; outside the new moonless twilight baseline Equation/reference use only; any replacement receives a new identifier
Schironi, “The Language of Astronomy” 2024, DOI 10.1515/9783111314532-002, especially pp. 30–31 Names and first/last semantics of the four visible phases summarized from Ptolemy Terminology and event doctrine only; not a numerical visibility model Article CC BY 4.0; cite, do not copy extended text
Knoll, Tousey, and Hulburt, “Visual Thresholds of Steady Point Sources of Light in Fields of Brightness from Dark to Daylight” JOSA 36 (1946), 480–482, DOI 10.1364/JOSA.36.000480 Independent point-source threshold evidence across dark-to-daylight fields One-arcminute steady sources, five young experienced observers, unaided binocular natural-pupil viewing; not an observer-population probability Bibliographic/equation reference only; no table or figure redistribution
Blackwell, “Contrast Thresholds of the Human Eye” JOSA 36 (1946), 624–643, DOI 10.1364/JOSA.36.000624 Underlying contrast-threshold experiment fitted by Crumey Detection data under the paper’s stimulus, adaptation, and probability conditions; not itself an astronomical atmosphere model Bibliographic/equation reference only; no source table redistribution without separate permission
Crumey, “Human Contrast Threshold and Astronomical Visibility” MNRAS 442 (2014), 2600–2619, arXiv 1405.4209, DOI 10.1093/mnras/stu992 Mathematical bridge from contrast-threshold data to astronomical point-source visibility The paper presents a dark-to-daylight achromatic model, but its astronomical verification is mainly dark-sky/scotopic; twilight use requires separate validation Transcribe equations with equation-number receipts; do not copy prose, figures, or tables
Tousey and Koomen, “The Visibility of Stars and Planets During Twilight” JOSA 43 (1953), 177–183, DOI 10.1364/JOSA.43.000177 Independent twilight observational validation Historical Washington, D.C. star cases and stated observing conditions; validation evidence, not a universal coefficient table Bibliographic observations may be encoded as source-owned fixtures with citation; no figure reproduction
Previc, Kosnik, McLin, Dennis, and Goettl, “The Visibility of Point Sources as a Function of Background Luminance, Target Luminance, Eccentricity, Wavelength, and Flicker Rate” AFRL-HE-BR-TR-2005-0138, October 2005, DOI 10.21236/ADA442029 Independent literature-synthesis envelope for observer protocol and sensitivity checks Synthesis of 14 studies after a broader literature review; shows that eccentricity and background matter strongly; not an astronomical event formula Approved for public release/distribution unlimited, but used as reference-only; no tables or figures copied
CIE 191:2010 / CIE TN 004:2016 CIE 191:2010 MES2 system; free technical note CIE TN 004:2016 Named transition between photopic and scotopic spectral weighting A spectral luminous-efficiency system, not a point-source detection threshold; adaptation field and S/P assumptions must be receipted Implement formula from the standard/note; the note itself is not redistributed
CIE photopic spectral luminous efficiency Dataset DOI 10.25039/CIE.DS.dktna2s3, CIE_sle_photopic.csv V(lambda) response data 360–830 nm at 1 nm, linear interpolation, zero extrapolation; the composite uses only the 380–780 nm overlap with the scotopic table Dataset CC BY-SA 4.0; permitted only in the separately licensed data pack with attribution/share-alike notice
CIE scotopic spectral luminous efficiency Dataset DOI 10.25039/CIE.DS.gr6w4b5g, CIE_sle_scotopic.csv V'(lambda) response data Metadata column domain 380–780 nm at 1 nm, linear interpolation, zero extrapolation Dataset CC BY-SA 4.0; permitted only in the separately licensed data pack with attribution/share-alike notice
libRadtran Version 2.0.6, released 2024-12-24, official download, manual Offline reference generator for spectral direct transmission and directional twilight radiance MYSTIC fully spherical reference cases for low Sun and near-horizon lines of sight; runtime use is prohibited GPL; external build tool only, with no source, binary, binding, or runtime dependency copied into Moira
Shettle aerosol model family E. P. Shettle, “Models of aerosols, clouds and precipitation for atmospheric propagation studies,” AGARD Conference Proceedings 454 (1989), as implemented by the pinned libRadtran 2.0.6 source and data Named rural, maritime, urban, and tropospheric aerosol profiles in summer and winter Checkpoint 5 binds all eight supported haze/season optical-depth files and fixes vulcan code 1; the direct-extinction oracle and directional-radiance uses remain distinct Source/data remain inside the external GPL research laboratory; no libRadtran profile file enters Moira or the future data pack
libRadtran REPTRAN module reptran_2024_all.tar.gz, accessed 2026-07-30 from the official download page Full-spectral molecular-absorption research input for libRadtran 2.0.6 REPTRAN fine is the admitted 380-780 nm reference; medium is characterization only External operator-supplied research data; the archive has no embedded notice or license file and is not redistributed by Moira
ESO Advanced Cerro Paranal Sky Model ESO project page Component comparison and validation reference Explicitly developed for Cerro Paranal; adapting it to other sites is not straightforward Code GPLv2; no copying or runtime integration; reference-only
Jones et al., “An advanced scattered moonlight model for Cerro Paranal” A&A 560 (2013), A91, DOI 10.1051/0004-6361/201322433 Candidate separately versioned spectral moonlight component Evaluated for 0.36–0.89 μm and conditioned on Cerro Paranal atmosphere/aerosols; not a scalar K&S replacement or global default Independently derived data/model component only; ESO GPL code remains external and is not copied or linked at runtime
Colina, Bohlin, and Castelli solar reference AJ 112 (1996), 307, DOI 10.1086/118016; NMSU source page; STScI CALSPEC sun_reference_stis_002.fits Independent top-of-atmosphere solar-spectrum authority for the Jones candidate The exact ESO/STIS comparison is bound over the STIS rows; first admission uses only the 380–780 nm response domain External source files are checksum-locked research inputs and are not redistributed in the repository
Kieffer and Stone ROLO lunar model AJ 129 (2005), 2887–2901, DOI 10.1086/430185; USGS publication record; USGS ROLO data Independent lunar-reflectance coefficient authority for the Jones candidate The 32 wavelength rows and four constants are bound directly; first admission is restricted to the 1.55–97 degree empirical phase domain and omits libration terms US Government data are public domain; the external source receipt is retained without copying the ESO member
Oxford EODG Mie routines Oxford EODG Mie page, eodg_mie.tar.gz Falsification authority for reconstructing the Jones aerosol phase function from published particle parameters The identified log-normal Mie procedure does not reproduce ESO mie_m15s1.dat; the undisclosed transform or smoothing may not be invented Archive states no license; external inspection and experiment only, with no source redistribution or runtime dependency
PALACE v1.0 GMD 18 (2025), 4353–4389, DOI 10.5194/gmd-18-4353-2025, data/code DOI 10.5281/zenodo.14064022 Later airglow comparison and component-validation reference Paranal model built from site-specific X-shooter/UVES evidence; not a global default Data CC BY 4.0, code GPLv3; reference-only in the current project

Source Conclusions

Legacy regression boundary

Kasten-Young, Schaefer, and Krisciunas-Schaefer remain authorities for the existing named components only. Their formulas and outputs remain regression protected. They are not blended into clear_sky_naked_eye_point_source_v1, and newer evidence does not mutate their identifiers.

Phase 4 moonlight disposition

Jones et al. 2013 is a spectral radiative-transfer model, not a replacement coefficient for the existing Krisciunas-Schaefer photometric option. Its governing object combines an observed solar spectrum, ROLO lunar albedo, molecular and aerosol scattering, absorption, and single-, double-, and approximated multiple-scattering terms. The paper evaluates the optical 0.36–0.89 μm range and reports a moon-model uncertainty near 0.15 magnitude under its study conditions. Its aerosol decomposition and validation corpus are explicitly Cerro Paranal conditioned.

The Phase 4 admission therefore uses these boundaries:

That component is not yet admitted. The paper does not provide a small machine-readable reference table sufficient to validate an independent engine implementation, so copying equations alone would not close the gate.

The Phase 4 Jones source-audit checkpoint now locks the official 431,651,392-byte SM-01 release and all 18 governing members without redistributing them. The inventory includes the default mie_m15s1.dat aerosol phase function selected by the package dependency map. It also proves that the official package fixture has a 102.1-degree phase angle, outside the 1.55–97-degree empirical ROLO range chosen for the first Moira candidate domain. That fixture remains lineage evidence rather than an admission golden. An isolated in-domain SkyCalc 2.0.9 flux_sml capture is separately locked as an operational comparison, not an independent oracle.

The source audit closes the package, license, candidate-domain, and artifact schema decisions. The follow-on input-authority checkpoint separately classifies the governing inputs. The solar spectrum is independently matched to STIS inside the candidate response domain, and the 32-row ROLO table is independently bound to Kieffer and Stone with its 1.55–97 degree empirical phase limit.

The published particle parameters and identified Oxford EODG Mie routine do not regenerate ESO’s selected mie_m15s1.dat phase table. That table is now explicitly source_owned_checksum_locked_not_reconstructable. It may remain an exact external input to an independent radiative-transfer pilot, but it is not an independent aerosol oracle and no undocumented transform may be invented.

The Phase 4 Jones MYSTIC pilot checkpoint now closes the corrected-v2 bounded 550 nm pilot and pre-holdout threshold gate. The v1 invalidation checkpoint preserves the discovery that the original explicit-aerosol writer assigned physical files to upper rather than lower layer boundaries. Because the v1 builder and validator shared that error, the v1 pilot, threshold, and holdout results are historical but inadmissible. No engine, public API, runtime model, or production data pack consumed them.

The corrected v2 profile has a top marker, an explicit null gap, and physical files bound to their inclusive lower boundaries. Fifteen replacement cases passed independent reconstruction, exact fixed-seed repeat, Monte Carlo convergence, aerosol-representation, lunar-source-linearity, and bounded source-owned operational comparison checks. V2 thresholds were frozen only after pilot measurement and before any of the three fresh holdouts were used. The Phase 4 Jones MYSTIC holdout checkpoint then closes the three replacement 550 nm geometry holdouts, using new seed 271828183, against the frozen 0.005 relative-error ceiling and independently verifies the exact repeat. None of these checkpoints admits a runtime or spectral model.

The Phase 4 Jones MYSTIC lower-boundary checkpoint now resolves the atmosphere limitation in favor of the source-owned geometry: a 2,000 m lower model boundary and explicit 2,640 m observer. It preserves 744 hPa and the same ozone number density at the observer, restores the Jones aerosol column below the observer, exactly reproduces the corrected-v2 control, passes all frozen checks, and independently validates two byte-identical final artifact trees. The observed 0.6-5.5% radiance changes were not used to select the profile. Admission still requires an independently generated spectral grid with frozen interpolation thresholds and untouched spectral holdouts, plus a release/distribution disposition for any derived artifact.

Phase 4 observer-factor disposition

Crumey’s equation-53 field factor is an overall multiplier that may combine target, medium, laboratory scaling, detection practice, and the observer. The paper’s contextual examples do not define a universal personal-factor domain, and its F = 2 value is explicitly notional and illustrative.

The physical observer protocol therefore keeps F = 2 as a fixed, source-receipted singleton. It does not expose an experience, skill, probability, or confidence input. Any future calibrated alternative requires a new observer-protocol identifier, calibration receipt, and validity domain. This decision does not alter the separate legacy VisibilityPolicy.crumey_field_factor contract.

Event doctrine

Schironi records Ptolemy’s four visible configurations as:

That exact four-part language is the authority for the new physical enum. Legacy heliacal, acronychal, and cosmic strings are preserved only as existing software contracts.

Detection criterion

Crumey supplies the selected mathematical family because it:

However, the current CRUMEY_2014_POINT_SOURCE identifier is not widened. The new blackwell_crumey_full_range_point_source_v1 component must be implemented separately and admitted only after:

  1. equation-by-equation transcription receipts;
  2. Blackwell/Knoll threshold checks;
  3. Tousey-Koomen twilight comparisons;
  4. modern independent cases where obtainable; and
  5. explicit residual and validity-domain reporting.

CIE MES2 supplies spectral weighting across adaptation states. It does not replace the detection threshold. The receipt must name the effective weighting and adaptation coefficient.

The AFRL synthesis prevents an implicit “generic observer” claim. The first model therefore fixes a known-location directed-observation protocol and does not extrapolate to peripheral discovery, flicker, casual scanning, or an observer population.

Atmosphere

libRadtran 2.0.6 is selected as an offline reference laboratory, not an engine dependency. Fully spherical MYSTIC cases are selected for difficult low-Sun directional sky radiance. Double-precision DISORT with pseudo-spherical geometry is selected for deterministic direct transmission, with its documented horizontal-irradiance projection removed explicitly. Phase 1 pins all solver, atmosphere, aerosol, absorption, surface, wavelength, random-seed, normalization, and convergence settings.

Phase 1 checkpoint 3 binds the governing libRadtran radiative-transfer theory and cdisort.c implementation receipts. It distinguishes the continuous Chapman integral from the selected surface flux implementation, which applies the bottom-layer midpoint Chapman factor to the full surface vertical optical depth. Independent controlled-atmosphere reconstruction now passes over the admitted 0.25-45 degree true-altitude domain.

Phase 1 checkpoint 4 binds the official external REPTRAN module and validates the 290-level candidate against 579- and 1,157-level controls across all six AFGL named atmospheres. REPTRAN fine is admitted as the clear molecular full-spectral research reference. The result is surface-only and does not itself admit environmental interpolation, response integration, or a runtime table.

Phase 1 checkpoint 5 binds the libRadtran aerosol, pressure, ozone, humidity, Beer-Lambert/Chapman, and delta-M implementation surfaces plus all eight Shettle haze/season optical-depth files. AOD is authoritative at 550 nm with beta = AOD550 * 0.55 ** alpha; aerosol visibility is not a second public control. Default pressure is named-profile-derived, while a measured override is admitted only inside both the 500-1,100 hPa absolute range and a 0.85-1.08 ratio to the profile pressure at observer altitude. Gray ground albedo is radiance-only. Temperature and relative humidity remain profile-derived.

The direct-extinction oracle preserves total aerosol optical depth and uses aerosol_modify ssa set 0 only to prevent delta-M phase-function bookkeeping from contaminating the physical Beer-Lambert line-of-sight extinction quantity. Directional-radiance runs retain physical aerosol scattering. The 73-run evidence also rejects linear unit-AOD scaling over the entire near-horizon domain. These decisions freeze parameter roles and candidate nodes, not interpolation accuracy or a runtime table.

Phase 1 checkpoint 6 closes the observer-altitude and pressure-ratio interpolation study for direct extinction. It uses site-relative 290-level atmospheres, eight altitude nodes from 0 through 5,000 m, five profile-relative pressure nodes, all six named molecular profiles, and bilinear interpolation in extinction magnitude. Across 12,636 withheld spectral values, maximum extinction error is 0.0124663582904496 mag and 95th-percentile error is 0.00404537460338972 mag. The law does not extrapolate and requires all four training corners to pass the existing absolute-pressure and pressure-ratio bounds.

The final Phase 1 radiance artifact uses REPTRAN-fine ALIS over 380-780 nm and the exact CIE photopic and scotopic datasets. A training-only diagnostic selected 531 nm as the common spectral-importance, normalization, and independent-anchor wavelength without executing any holdout. The admitted 64-node response grid has nine untouched off-grid response cases. Photopic/scotopic maximum interpolation errors are 0.354270166272975 and 0.25296630532828035 mag; their 95th-percentile errors remain below the unchanged 0.3-mag ceiling.

The separately licensed moira-physical-heliacal-visibility data pack version 1.0.0 contains only generated response-integrated products, per-cell solver uncertainty, direct-extinction values, error receipts, provenance, and notices. Its root-manifest SHA-256 is 49ac2b68ea105a8e055b27e8d4d70f6cbfe9533f971ef5e6000f0bdd95d6771b. The same immutable pack passed independent Linux and Windows validation. It contains no CIE source table or libRadtran/REPTRAN source, profile, data file, binary, or engine dependency.

The first pack is explicitly a fixed U.S. Standard, rural-summer, sea-level baseline. Earlier environmental and altitude/pressure evidence does not silently create absent pack axes. Requests outside the pack’s exact manifest domain must be typed not_evaluable; broader environmental coverage requires new versioned data-pack evidence.

The initial baseline excludes moonlight and site-specific airglow. Jones, the ESO model, and PALACE remain named comparison or later-component sources. They do not enter the baseline merely because they are newer.

Immutable Source Identities

libRadtran archive

URL:
  https://www.libradtran.org/download/libRadtran-2.0.6.tar.gz
Version:
  2.0.6
Release date:
  2024-12-24
Retrieved bytes:
  154147176
SHA-256:
  64930cc40b6e4a37aa220520974d330fc1563796f466a649b2238131f2d69840
Observed Last-Modified:
  Sat, 18 Jan 2025 10:28:42 GMT
Observed ETag:
  "9301968-62bf87dcdc4c9"

Phase 1 checkpoint 1 independently acquired and verified this identity before the first generator build. Both offline builders re-verify byte count and SHA-256 before every authorized profile. A future hash mismatch stops the build and requires a source-ledger amendment.

REPTRAN module

Official download page:
  https://www.libradtran.org/doku.php?id=download
Direct archive:
  https://www.libradtran.org/lib/exe/fetch.php?media=download%3Areptran_2024_all.tar.gz
Access date:
  2026-07-30
Retrieved bytes:
  698709957
SHA-256:
  55893c80bcc999651bac3bf014ee64aaf602653ba640eb5bebe787a5d8eacce7
Regular archive files:
  292

The 260 files overlapping the libRadtran 2.0.6 source data are byte-identical; the module supplies 32 official files absent from that tree. The separately constructed data root is bound by 1,478 per-file checksums and canonical receipt SHA-256 68f1817782e424ef617dab03ad985a3fbcb91fa2ed0239a8c2de1e8cb6855b59.

No notice, license, README, or citation file is embedded in the REPTRAN archive. The module remains an external research input and is not copied into the repository, engine wheel, or future visibility data pack without a separate distribution decision.

Checkpoint 2 additionally binds the exact libRadtran 2.0.6 implementation files governing altitude insertion, vertical interpolation, molecular-profile combination, and spherical MYSTIC elevation:

src/atmosphere.c
  b900ade7e603260a47fec3efa305577ab6806bbf539021ec028a0c1360099cf8
src/uvspec_lex.l
  174755190e50ecc3099c80a29cb71627c0a33a5e2009d1869c23140095658d89
src/ancillary.c
  97dc576d1cb8f54c40d733cea3d5a56b49a0e7f8f39aa812e55ba7fbe1a7665f
libsrc_c/mystic.c
  d1d981e0dd2e961f7f8991368b92e2179c382bab81c8ae72ed17cd31dbcab87b

Those sources establish why the spherical Monte Carlo probe uses a source-derived atmosphere with the site as its bottom level and a separately bound O4 companion profile. The resulting construction matched all 45 supported deterministic-altitude oracle cases at emitted precision.

Checkpoint 5 additionally binds ten governing libRadtran manual/source files, all eight Shettle tau550 profile files, and the exact uvspec executable used by the 73-run environmental artifact. The admitted root-manifest SHA-256 is e79a250b01f00783f272bae409fa323a94b5c7811375760bf536eaa7de6b0580; the source-owned parameter and holdout contract is recorded in PHYSICAL_HELIACAL_VISIBILITY_PHASE1_ENVIRONMENT_CONTRACT_CHECKPOINT_2026-07-30.md.

Checkpoint 6 binds 5,037 libRadtran runs, 50,768 files, and four preserved failed-design receipts. Its admitted root-manifest SHA-256 is 2264727cf4d1a74bb747aa51cc44e4ba9e703e09c132ab57eb2c0afef863c727. The same immutable bytes passed the independent validator under Linux and Windows; the latter used an immutable tar transport to local NTFS to avoid the WSL UNC per-file traversal penalty. The compact receipt is recorded in PHYSICAL_HELIACAL_VISIBILITY_PHASE1_ALTITUDE_PRESSURE_INTERPOLATION_CHECKPOINT_2026-07-30.md.

ESO SM-01 Jones moonlight source

URL:
  https://ftp.eso.org/pub/dfs/pipelines/skytools/skymodel/SM-01.tar.gz
Release:
  1.0.0
Release date:
  2018-08-30
Retrieved bytes:
  431651392
SHA-256:
  e09b1d62c8af212486f50097fe76d9dcbb242f4fbadf720a4a85be361cc9116b
License:
  GPL-2.0-or-later

The Phase 4 audit binds 18 individual package members and the official flux_sml regression signature. The package remains external: it is not copied, linked, compiled, executed, or installed by Moira. Its test geometry is outside the first candidate phase domain and is used only to confirm official-implementation lineage.

The separately retained SkyCalc 2.0.9 operational comparison has FITS SHA-256 12d7625e1ec1afc718928d873fdb0001d3fd800b19d33a6c5dc2ce135dbbc230 and isolated-component signature SHA-256 8e15e62b5aa5cab32961f3be7ba300f46217d20614e91bdc131aa8ee8b2e1c29. Only those receipts are committed; the FITS bytes are not redistributed.

Jones Phase 4 input authorities

The input-authority checkpoint independently replays the first 1,467 ESO solar rows against STScI CALSPEC and locks the 325 samples inside the 380–780 nm candidate domain. The external STIS file has SHA-256 20133cdf2a402655e3726de6b334bf09d05c9429ebd27ebf5492dc37fc89cfd8; the candidate-domain ESO numeric signature is fcdb33a16166f8ee3e9f894371f3d79e14efddb063753104d547897071be9024.

Kieffer and Stone’s four constants and all 32 published wavelength rows match the locked ESO moonalbedo.dat numeric table. The Table 4 numeric signature is 620b1ca086edda0221a1db7461d69602479c5c584aa403843084786b5608278e.

The Oxford EODG archive used for the aerosol reconstruction falsification has SHA-256 a026c570b2d39988e597fb7ce5b7bc3e451f650a8d6013670e68af5343cf9561. The identified public calculation gives g = 0.595268389115 at 0.55 micrometres, while the selected ESO table gives g = 0.680602583549528 and a materially different phase curve. The repository records the failure and the external source receipt, not the source bytes or an invented correction.

CIE photopic table

Data DOI:
  10.25039/CIE.DS.dktna2s3
File:
  https://files.cie.co.at/CIE_sle_photopic.csv
Metadata:
  https://files.cie.co.at/CIE_sle_photopic.csv_metadata.json
SHA-256:
  ee5d5d17922ae645d4af52cacf6a50bdb9385749f9d2181ca312eb2b08febac2
MD5:
  f389958555461a7d9a7562145e8ca9c0
License:
  CC BY-SA 4.0

The official dataset landing page displayed a different MD5 during the 2026-07-29 audit. The downloaded bytes and the official metadata JSON agreed on the values above. Moira binds SHA-256 plus metadata identity and records the discrepancy; it does not use the stale displayed MD5 as authority.

CIE scotopic table

Data DOI:
  10.25039/CIE.DS.gr6w4b5g
File:
  https://files.cie.co.at/CIE_sle_scotopic.csv
Metadata:
  https://files.cie.co.at/CIE_sle_scotopic.csv_metadata.json
SHA-256:
  6a75d3fdbcbf5e9e9a07478511933eefeda953f3e2cc14b74459e5a099ec3759
MD5:
  3e45714a429d02e5d1f2a752226d7698
License:
  CC BY-SA 4.0

The scotopic metadata description says 360–830 nm, while its column metadata and file coverage are 380–780 nm. The runtime composite uses the actual 380–780 nm overlap and records the metadata inconsistency.

Licensing and Artifact Boundary

The engine source and wheel remain MIT and contain no copied GPL code.

The physical visibility table is a separate artifact:

Artifact:
  Moira visibility data pack
License:
  CC BY-SA 4.0
Runtime acquisition:
  explicit caller action only
Engine dependency:
  optional caller-supplied path
Network during calculation:
  prohibited

The admitted pack attributes the CIE datasets, preserves their DOI, includes the share-alike notice, and identifies the generated libRadtran numerical products and exact generator configuration.

The repository now contains a metadata-only compatibility contract and independent validator, but not the CIE tables or physical LUT. A future release review must still inspect the completed external artifact and notices before distribution. This is an engineering disposition, not legal advice.

Formula Admission Inventory

No formula is admitted merely by appearing in this ledger.

Planned calculation Governing source Phase that must capture exact equation/section Current state
Blackwell/Crumey point-source threshold Crumey 2014 with Blackwell 1946 lineage Phase 2 Implemented from Crumey equations 28 and 34 with fixed F=2; independently checked against all eight public Tousey-Koomen Table I rows
MES2 spectral response CIE 191:2010, CIE TN 004:2016, and CIE TN 007:2017 Phase 2 Implemented with fixed-point solution and same-equation bracketed fallback; both official TN 007 worked examples pass
Spectral radiance to named photometric quantity CIE TN 004:2016 Phase 1/2 Phase 1 source-locks the official CIE tables and admits response-integrated photopic/scotopic data-pack products; Phase 2 owns single-epoch composition and limiting-magnitude propagation
Direct transmission libRadtran 2.0.6 DISORT pseudo-spherical configuration with external REPTRAN fine data Phase 1 Deterministic smoke and exact repeat reproduced; elevated-site construction matched 45 oracle cases; the surface midpoint-Chapman implementation is source-traced; the 290-level clear molecular grid is bounded across all six AFGL profiles; REPTRAN fine is the full-spectral research reference; Checkpoint 5 source-binds AOD550, Angstrom, ozone, pressure, albedo, and all eight Shettle profiles and admits a delta-M-safe aerosol direct-extinction oracle; Checkpoint 6 admits site-relative altitude/pressure interpolation against 12,636 withheld spectral values; the first pack admits a 57-node, 400-bin direct surface with 22,400 untouched holdout bins
Directional twilight radiance libRadtran 2.0.6 MYSTIC configuration Phase 1 The final v9 artifact admits adaptive 531 nm anchored REPTRAN-fine response products over a 4-by-4-by-4 grid, nine untouched response holdouts, per-cell uncertainty, and a typed boundary below -9 degrees
Physical event root Moira-defined margin law and admission doctrine Phase 3 Implemented with certified zero enclosure, four typed phase semantics, exact-pack admission, and independent planetary/stellar validation
Directional terrain horizon Caller-supplied source-receipted apparent-altitude profile plus Phase 3 horizontal-rate certificate Phase 4 Implemented with circular-linear interpolation, 10-degree maximum segment, scalar compatibility, exact slope-derived event certificate, and no missing-direction fallback
Moonlight Jones 2013 or existing named legacy option Phase 4 Legacy K&S identifier preserved; Jones source, license, domain, artifact contract, and solar/ROLO/aerosol authority classifications are frozen; the external 550 nm MYSTIC pilot, pre-holdout thresholds, three sealed geometry holdouts, and source-faithful 2,000 m lower-boundary gate pass; a spectral grid with interpolation validation, release disposition, and runtime admission remain open
Airglow and other separable sky components caller-supplied source-receipted directional output; ESO/PALACE comparison Phase 4 Typed composition and double-count prevention implemented; no built-in airglow, zodiacal, integrated-starlight, artificial-light, or Paranal-global model admitted

Phase implementation documents must quote no more source text than needed, name exact equation or section identifiers, and attach independent numerical fixtures.

Phase 2 Exact Equation Admission - 2026-07-30

The Phase 2 implementation binds the following exact source artifacts:

Artifact Engine use Local source receipt
Crumey 2014 Full-range point-source threshold from equations 28 and 34; equation-28 coefficients a1=5.949e-8, a2=-2.389e-7, a3=2.459e-7, a4=4.120e-4, a5=-4.225e-4; fixed field factor F=2 PDF SHA-256 fa6ef183f9402be4d321bff5fa2c112510f89ca683b534e33c63fdb6538e50a4
CIE TN 004:2016 MES2 governing equations and photopic/scotopic quantity definitions PDF SHA-256 a549fcf5f98ae5fdd959b331dbb91eae99f5fd397bd288ad6b59c43723a4494f
CIE TN 007:2017 Two official MES2 calculation examples and the task-applicability restriction in clause 6 PDF SHA-256 efdd11f4bdf7d77ab3b1fb8e6b94ac89599521eba7425e474bbc82cf34c7877a
Tousey and Koomen 1953 Independent eight-row threshold validation, not coefficient fitting Public HTML SHA-256 4e50f748c6c0de310ceeadcbbcd0a6626a3fccd74fe1063f9cc91640ad3212ef

CIE TN 007 restricts MES2 use to peripheral visual tasks. The admitted observer protocol is therefore known_location_directed_averted_observation_v1: the target position is known and attention is deliberate, but fixation is averted/peripheral after adaptation to the immediate directional field. The engine does not claim foveal equivalence.

CIE 257:2026 is recorded as a current follow-on publication. Its full report was not inspected for this admission, so it is not an equation or coefficient authority for Phase 2.

The source-owned numerical fixture is tests/fixtures/physical_visibility_phase2_equations_v1.json. The Crumey implementation reproduces the public Tousey-Koomen Table I threshold values with a maximum absolute residual of 0.03572 in log10 illuminance (0.0893 magnitude), within the declared fixture acceptance bound.

Phase 2 Planetary Target-Data Admission

The first-candidate planetary audit is complete for Mercury, Venus, Mars, Jupiter, and Saturn:

The separately distributed physical-visibility pack version 1.1 binds:

Source Admitted role Exact identity
Payne et al. planetary spectra Full-phase geometric-albedo spectra used with the locked extraterrestrial solar spectrum to derive each planet’s base photopic and scotopic response integrands Versioned record 10.5281/zenodo.17470005, publication 10.3847/PSJ/ae2feb, CC BY 4.0; five per-file SHA-256 receipts are recorded in the Phase 2 checkpoint artifact
Mallama et al. 2017 Source-domain UBVRI phase/color laws; Mercury remains gray, Venus/Mars/Jupiter use phase-polynomial color, and Saturn uses phase plus effective ring sub-latitude PDF SHA-256 7feb8edb372502cee5dc9c6a7656205e3279353bb38f9a98cbecbe8e8d733f91; DOI 10.1016/j.icarus.2016.09.023
CIE photopic and scotopic datasets Response integration and S/P ratio Dataset DOIs 10.25039/CIE.DS.dktna2s3 and 10.25039/CIE.DS.gr6w4b5g; exact CSV receipts remain source-locked
libRadtran 2.0.6 atlas_plus_modtran Extraterrestrial solar spectral shape used only during offline pack derivation SHA-256 432600ef415706c401a4c0e17c6b733a631f1556a78c3da32e936830288b414b

The 400-bin response weights remain external pack products; no planetary source spectrum or CIE table enters the MIT engine wheel. The engine resolves phase angle and Saturn ring geometry from the same ephemeris context as apparent magnitude. It refuses profile extrapolation beyond the source-owned domains and does not accept caller-supplied planetary response weights.

The exact profile specification is scripts/visibility_reference_lab/phase2_planetary_target_profile_pack_spec.json. The compact source-controlled receipt is tests/artifacts/visibility_reference_lab/phase2_planetary_target_profiles_checkpoint_2026-07-30.json. The admitted 1.1 manifest SHA-256 is f594fd12058cc7f5c7bc9de7f2b06652bef3c0604ef7b0a05a069e54e4026c87.

The current fixed-star field named color_index is not sufficient evidence of a particular color system. Fixed-star admission remains outside Phase 2; no implementation may silently interpret that field.

Phase 2 Numerical-Error Propagation Receipt

Phase 1 assigns limiting-magnitude propagation to Phase 2. The admitted single-epoch implementation now consumes the exact downstream error contract from error-envelope.json:

For modeled twilight, the numerical bounds perturb the twilight response while holding the source-identified dark-sky anchor nominal. CIE MES2 adaptation and the Crumey threshold are evaluated across all four photopic/scotopic bound corners. The resulting limiting-magnitude extrema are combined conservatively with the direct-extinction target-magnitude bound to produce the final margin envelope.

The method receipt is phase2_data_pack_declared_numerical_error_envelope_v1. The solver term is propagated at exactly plus or minus one reported relative standard error; it is not relabeled as a hard maximum. The method is explicitly not a scientific-confidence interval. Input-measurement, target photometry and spectral-source, observer-population, model-form, and actual-atmosphere uncertainties remain named but unquantified.

Phase 0 Source Disposition

Every candidate source now has one of four explicit roles:

No GPL code enters the engine, no site-specific model becomes a global default, no CIE data enters the MIT wheel, and no unidentified coefficient is approved. The source gate is closed for Phase 0. Phase 1 implementation evidence, the external fixed-domain data packs, and Phase 2 engine-side single-epoch truth are complete. Phase 3 owns physical event-time solving.

This site is open source. Improve this page.