36 Zero-Dispersion Common Delay in Ultra-Long Fiber Links and Its Association with Crustal Tensors

Home ／ Appendix-Prediction and Falsification

This chapter follows the publication template for the falsification program. It uses plain language, avoids equations, and preserves the fixed structure. Under unified time/frequency standards and end-to-end link calibration, we target ultra-long fiber links—intercontinental/subsea/national backbones (metro–provincial–cross-border DWDM/OTN and subsea dark/business channels). After multi-wavelength (WDM), bi-directional (round-trip/counter-propagating), and multi-link-type de-dispersion and alignment, we test for a frequency-independent, direction-independent, time-window-stable common delay, and probe its monotonic or threshold-like association with crustal tensors proxied by GNSS strain-rate/stress, InSAR deformation, borehole strainmeters, solid-Earth tides and loading. If common delays trace temperature/load/device states/dispersion slope/queueing timestamps—or fail cross-operator/link/team replication—the claim is disfavored.

I. One-Sentence Goal

Determine whether ultra-long fiber networks exhibit a non-dispersive common delay that co-occurs at zero lag across wavelengths/directions/link types and tracks crustal-tensor strength, rather than standard link physics.

II. What to Measure

• Common-delay index (text grades):
  Per link segment/corridor (terrestrial trunk, subsea, mountain–basin transitions, fault crossings), within a unified window, extract residual constant delay / slow-drift slope (after per-wavelength dispersion embedding removal and round-trip asymmetry correction). Grade strong / medium / weak; uplift / depression, indexed by window–wavelength–direction.
• Non-dispersion consistency:
  Across multiple wavelengths/sub-bands and traffic modes (dark channels / live services / idle mux spans), require common-delay direction/strength to be insensitive to carrier / band edge / dispersion slope. λ-dependent flips/scalings imply dispersion/link origins.
• Zero-lag co-occurrence (aligned):
  Under a single external timescale, evaluate zero-lag peaks and side-lobe contrast across wavelengths / counter-prop directions / link types, and test co-evolution with environmental strength.
• Environmental association and stratification:
  Co-register crustal-tensor proxies (GNSS strain-rate, InSAR time-series deformation, borehole strain, solid-Earth/ocean/atmospheric loading, nearby seismicity/slow slip) with link corridors. Tag common-delay responses as enhance / plateau / uncorrelated, stratified by fault proximity, land–sea transitions, terrain gradients.
• Route-overlap gradients:
  For co-linear / partially overlapping / disjoint links, compare spatial coherence of common delay; significance confined to co-linear sections supports the claim.
• Seasonal/event reproducibility:
  Check sign and ordering under seasonal water/thermal loading and during events (earthquakes, slow slip, strong tides, extreme weather, engineering works), and compare to prediction cards.

III. How to Do It

1. Networks and facilities:
   • Link types: metro rings; national/continental DWDM/OTN; subsea dark channels (with controllable optical power) and service-carrying channels; spares/bypasses.
   • Transfer modes: two-way time/frequency transfer, phase-stabilized optical-frequency transfer, White Rabbit/hardware timestamp links, round-trip ranging-style delay monitors.
   • Auxiliary observations: continuous GNSS (strain-rate/3-D displacement), InSAR time series, borehole/rock strain, strong-motion/broadband seismometers, Distributed Temperature Sensing (DTS)/pipeline temperature, barometry/tide/ground deformation, and EDFA/ROADM/mux–demux state logs.
2. Unified calibration and de-systematics:
   • Link end: publish bandpass kernels, band-edge hold-outs, round-trip asymmetry, EDFA gain/saturation, dispersion slope, polarization-mode dispersion (PMD), protection transients; standardize to a common kernel and window function.
   • Timescale: a single external atomic time/1PPS with cross-site closures; keep an independent reference return path to monitor timescale drift/queue jitter.
   • Thermo-mechanical fields: enable DTS/tilt/vibration where possible; regress temperature→delay / mechanical→delay, and publish residual caps.
3. Sequence construction and comparison:
   • Co-window/co-segment grid: per segment × wavelength × direction × window, tabulate common-delay and zero-lag indices (text grades) and attach environment-strength quantiles.
   • Differencing/orthogonality tests: build contrasts for temperature/power/traffic/device-switching and require the common delay to be nearly orthogonal.
   • Spatial stitching: along overlapping/disjoint routes, test spatial continuity and break responses of the common delay.
4. Forward prediction, blinding, arbitration:
   • Environment-forward team: using only GNSS/InSAR/strainmeter/tide & load inputs and corridor masks, issue prediction cards (direction/strength of common term; non-dispersion; zero-lag significance; threshold-like behavior).
   • Measurement teams (independent pipelines): run ≥ 2 cleaning paths (time-domain first / phase-domain first) and two alignment strategies without sharing residuals.
   • Arbitration: align predictions and summaries; score hit / wrong / null by link type / corridor / season / method family.
5. Cross-consistency:
   • Reuse non-dispersion/zero-lag/environment stratification from Chapter 35 (global clock scale co-drift).
   • Align link de-systematics and common-kernel methods with Chapter 24 (absolute cross-environment light-speed comparison).
   • Cross-check with regional seismic/deformation centers for event–corridor validation.

IV. Positive/Negative Controls and Artifact Removal

1. Positive controls (support “zero-dispersion common delay ↔ crustal tensors”):
   • Across multi-wavelength, bi-directional, multi-link settings, common delay shows same-sign, stable behavior with zero-lag after alignment.
   • Non-dispersion holds: no λ / band-edge / dispersion-slope flips/scalings; weak dependence on traffic power/management.
   • Monotonic/threshold association with tensor proxies; strong on co-linear corridors, weakened/absent on disjoint routes.
   • Prediction-card hit rates above chance and replication across operators/teams/facilities.
2. Negative controls (against the association):
   • Common terms follow λ / band-edge / PMD / EDFA dynamics / power / temperature laws.
   • Significance confined to one wavelength/direction/path, or fragile to kernel/alignment/window choices.
   • Label swaps/route shuffles/time reversals/parameter randomization still yield “detections.”
   • Round-trip cancellation/reference return reproduces or erases the signal, or Sagnac/diurnal alone suffices.

V. Systematics and Safeguards (Three Items)

• Thermo-mechanical dispersion/length changes: may appear as λ-dependent or slow drift. Safeguard: multi-wavelength slope embedding removal, band-edge hold-outs, DTS constraints, bi-directional cancellation, and dark/spare-fiber controls.
• Equipment/network events (EDFA/ROADM/mux, PMD): state changes and saturation produce false steps. Safeguard: event-log masks, inject–recover tests, multi-vendor/multi-path redundancy, and counter-prop cross-checks.
• Timestamp/queue jitter (non-hardware links): bursts of delay from switching. Safeguard: prefer hardware timestamp/White Rabbit, use independent reference return, peak-hour hold-outs, and publish jitter ceilings.

VI. Execution and Transparency

Pre-register corridor/segment/wavelength lists, unified calibration/de-systematics workflows, criteria for common delay / non-dispersion / zero-lag, environment variables and bins, positive/negative controls, exclusions, and arbitration scoring. Define held-out units for seacable vs terrestrial contrasts, fault crossings, urban ducts, and event windows. Enable cross-team/operator replication by sharing counter/phase raw data, DTS & power/equipment logs, and scripts; run down-sampling/noise/kernel-variant/alignment-perturbation tests. Publicly release prediction cards, segment-level common-delay tables, zero-lag/non-dispersion summaries, DTS/EDFA/route logs, and key intermediates, plus text-based corridor–environment overlays.

VII. Pass/Fail Criteria

1. Support (passes):
   • In ≥ 2 independent pipelines, ≥ 2 link classes (land and subsea), ≥ 2 operators, observe zero-lag, non-dispersive common delays that show monotonic/threshold association with tensor proxies.
   • Robust to wavelength/direction/kernel/alignment/event masking, and replicates in held-out units.
   • Prediction-card hits exceed chance.
2. Refutation (fails):
   • Results follow dispersion/link laws or are dominated by temperature/equipment/traffic; replication fails across links/teams.
   • High parameter fragility or disappearance in held-out units.
   • Arbitration hits near chance, indistinguishable from calibration/method artifacts.