Physical Heliacal Visibility Phase 4 Jones MYSTIC Lower-Boundary Checkpoint

Historical research only. The proposed next gate in this checkpoint was canceled by the 2026-08-07 Jones/Paranal quarantine decision. Nothing in this document is an active product or release dependency.

Date: 2026-08-02 Status: Jones 2,000 m lower-model-boundary gate closed; source-faithful profile selected for spectral research; spectral grid and runtime model not admitted

Outcome

The Jones/ESO geometry is now resolved without retaining the corrected-v2 observer-bottom approximation as the candidate model.

The source-faithful research profile is:

jones_2000m_ground_observer_2640m_v1

It has:

• a 2.000 km lower atmosphere and ground boundary;
• an explicit 2.640 km observer level in both the atmosphere and O4 files;
• libRadtran zout 0.64, because numeric zout is height above the surface;
• 806.2488072631218 hPa at the 2.000 km surface, which preserves 744 hPa at the 2.640 km observer after the bound AFGL pressure scaling;
• 259.19991781804873 DU above the 2.000 km surface, which preserves the same 4.914335e11 cm-3 observer-level ozone number density as the 258 DU observer-bottom control; and
• the Jones exponential aerosol column down to 2.000 km while preserving every corrected-v2 layer boundary and optical property above 2.640 km.

The selection is controlled by the source semantics (sm_h = 2.64 and sm_hmin = 2.0), not by numerical closeness to the observer-bottom control.

Frozen Comparison

The experiment compared two profiles at 550 nm:

Profile	Surface	Observer	zout	Observer pressure	Role
observer_bottom_control_v2	2.640 km	2.640 km	0.00 km	744 hPa	Exact corrected-v2 holdout regression control
jones_2000m_ground_observer_2640m_v1	2.000 km	2.640 km	0.64 km	744 hPa	Source-faithful lower-boundary candidate

The candidate aerosol receipt is:

• observer-to-infinite-top AOD550 normalization: 0.0294;
• observer-to-20 km AOD550: 0.02939998466958803;
• 2 km-to-20 km AOD550: 0.05011536771007698; and
• restored below-observer AOD550: 0.020715383040488946.

The candidate adds boundaries at 2.000, 2.250, 2.500, and 2.640 km, then uses every corrected-v2 boundary above 2.640 km unchanged. Physical aerosol files remain bound to inclusive lower boundaries. The 20-120 km region remains an explicit null gap.

Execution

The final matrix used the three corrected-v2 reserved holdout geometries for both profiles, plus one exact candidate repeat:

• seven serial MYSTIC runs;
• 1,000,000 photons per run;
• fixed seed 271828183;
• fully spherical one-dimensional MYSTIC; and
• a frozen maximum relative Monte Carlo standard error of 0.005.

The control reproduced the committed corrected-v2 holdout radiance and standard-error values exactly. The candidate repeat was byte-identical for mc.rad.spc, mc.rad.std.spc, mc3.rad, mc3.rad.std, and randomseed.

MYSTIC uses its internal altitude-level index in the altitude-plane output filename. The source-bound surface control therefore emits mc0.*, while the 2.640 km candidate observer emits mc3.*. This is an output identity detail, not a change in the evaluated altitude.

Measurements

Case	Control radiance	Candidate radiance	Candidate/control	Relative change	Candidate minus control
Low-altitude oblique	3.09175e-08	3.26044e-08	1.0545613326	+5.4561%	-0.05768 mag
Interior cross-axis	7.88475e-09	8.07704e-09	1.0243875836	+2.4388%	-0.02616 mag
Phase-boundary waxing	9.54726e-09	9.60728e-09	1.0062866205	+0.6287%	-0.00680 mag

The maximum relative Monte Carlo standard error was 0.002984007960629856, below the frozen 0.005 ceiling.

These differences are measurements only. No candidate/control difference threshold was prefrozen, and the numerical difference was not used to choose the profile.

Independent Validation and Reproducibility

The independent validator does not import the new lower-boundary builder. It:

• verifies every corrected-v2 lineage receipt;
• reopens the exact ESO and libRadtran archives and verifies the governing configuration and source members;
• independently reconstructs the 2.000 and 2.640 km atmosphere and O4 rows;
• independently reconstructs every candidate aerosol layer and Legendre representation;
• reparses every MYSTIC input and output;
• verifies the exact corrected-v2 control regression;
• recomputes all paired diagnostics and checks; and
• enforces the closed runtime, network, redistribution, and admission boundaries.

Two final artifacts were built serially from the same frozen contract. Their entire directory trees were byte-for-byte identical:

• manifest: 71,981 bytes, SHA-256 4223b52c001f5f6e427130bdc4523c1031f617e904987c823a3cd18883911736;
• compact checkpoint: 14,529 bytes, SHA-256 6c3ef3b750ff235719a3b9014e05bb5a5a7f6c022decc2f83a0f4f85f4906b31.

The atmosphere-only probe initially left a time-varying, scientifically unused randomseed file. The final builder removes that incidental file before receipting the probe. The two final full-tree matches demonstrate that it no longer contaminates reproducibility.

Pre-Authority Attempts

Five pre-authority attempts are retained in the frozen specification history:

1. a profile probe rejected before radiance execution because its source declaration was incomplete;
2. a candidate inventory gate that discovered mc3.* instead of mc0.* at nonzero zout;
3. an inventory diagnostic that confirmed that profile-specific behavior;
4. a scientifically successful artifact superseded by zero-byte-receipt and probe-inventory validator hardening; and
5. an independently validated artifact superseded because the incidental probe randomseed receipt was time varying.

None is authoritative. The first three were atomically discarded; the two retained external artifacts are explicitly superseded by the identical final artifacts.

Boundary

This checkpoint does not:

• admit a spectral wavelength grid or interpolation surface;
• admit a runtime Jones model or production data pack;
• change Moira engine calculations or public APIs;
• add a runtime, network, or automatic-download dependency;
• redistribute ESO or libRadtran source bytes; or
• modify the legacy krisciunas_schaefer_1991 path.

Repository Receipts

• scripts/visibility_reference_lab/phase4_jones_mystic_lower_boundary_spec.json
• scripts/build_visibility_phase4_jones_mystic_lower_boundary.py
• scripts/validate_visibility_phase4_jones_mystic_lower_boundary.py
• tests/artifacts/visibility_reference_lab/phase4_jones_mystic_lower_boundary_checkpoint_2026-08-02.json
• tests/unit/test_visibility_phase4_jones_mystic_lower_boundary.py

The full generated artifacts remain external because they contain derived evidence built from caller-supplied GPL/source inputs. The repository stores only the compact checksummed receipt.

Next Gate

The next authorized work is:

design_jones_2000m_ground_spectral_admission_matrix

That design must freeze its wavelength grid, interpolation policy, numerical thresholds, and untouched spectral holdouts before executing admission holdouts. Phase 4 remains open, and no runtime implementation is authorized by this checkpoint.