Why civil dawn, sunrise and polar unknown are different states
Separate geometric twilight from apparent sunrise and avoid turning a missing event or uncertain historical offset into a false conclusion.
Keep apparent sunrise separate from centre twilight
The public San Jose question used −6° civil dawn as sunrise. The correction was to use SearchRiseSet for the apparent upper limb. This tool retains that distinction, including the mature fixed refraction and observer dip; it does not replace upper-limb rise with a geometric centre at 0° or a guessed −0.833° threshold.
| Request | Geometry |
|---|---|
| upper-limb | Apparent upper limb |
| civil | Civil twilight (centre −6°) |
| nautical | Nautical twilight (centre −12°) |
| astronomical | Astronomical twilight (centre −18°) |
A missing event needs both directions and the whole window
A missing event alone does not establish polar day or night. Both 370-day searches and the complete window must support the state; uncertainty stays unknown.
Both rise and set directions are examined even when you requested only one. Diagnostic events are marked requestedDirection=false. Both 370-day next-event searches are distinct and fully reported. Any required out-of-year probe, uncertain near-tangent interval, non-progress or ordinary engine failure remains unknown instead of becoming polar day or night.
- Include the window start and exclude the window end.
- Keep all original entities, repeated IDs and invalid values.
- Times are approximate at minute scale. Weather, refraction and the visible horizon can change an observed crossing.
Historical offsets are source facts, not a device preference
The Boston discussion reports a historical DST correction, but no complete independently verified 1917 offset calendar is provided. Use UTC when offsets are unknown, or supply your own explicit complete map. Unused declared intervals remain in the report and original.
Any input, source, locale or setting change cancels the current run and clears its old result and links immediately. After cancellation, retry the same original and parameters in a fresh worker; reduce a scope that exceeds the joint budgets.
Mature resolution and consumed intervals
This calculation adopts the fixed mature model’s single-crossing bracket and Nyquist assumptions. After a resolved event, the complete interval from its original returned time to the original positive bracket endpoint is recorded as that same event and consumed before continuing. The original 0.1-second numerical tolerance and sub-second pair exclusion remain; this is not a mathematical certificate or an observation precision guarantee. Crossings near altitude extrema, tangencies and sub-second pairs are outside the completeness guarantee; observed unsupported or touching evidence remains unknown.
Any unexplained sub-second interval, non-progress, inconsistent bracket or overlap with the requested window boundary remains unknown. Every original prune and solver exit stays in the report; the full consumed interval is retained in every search JSON in the CSV, timeline metadata, full copy and report. settings.json also retains the complete model, every original search and consumed interval, with its own signed-zero path sidecar.
- Review the actual negative/positive endpoint values and event/search identities.
- Keep the full original input, settings and all six files together.
References
- Civil dawn versus sunrise correction
Published coordinates and date context are known; web minute values are not second-precision observations.
- Polar day and night discussion
The question’s actual date is missing; separately declared dates are synthetic.
- Historical offset correction
The author reports correcting DST; this does not independently establish a 1917 offset table.
Tools in this category
Expand a tool to see its steps, options and supported formats, then open its workspace.
Solar Horizon CalendarSearch every declared solar window and retain unknown, polar and invalid states with complete evidence.
UTC windows include their start and exclude their end. Civil time uses your complete offset intervals; no historical time zone is guessed.
Steps
- Select the original JSON file or paste a complete document with all observers and windows.
- Declare the UTC endpoints and requested upper-limb or centre twilight definitions; supply the full offset map when civil time is used.
- Inspect every request, event and search state, then save all six files and the complete report.
Available options
- Active source
- One original JSON file · Pasted JSON
- Time presentation
- UTC; preserve any supplied offset map · UTC and complete declared offsets
Capabilities and limits
- Strict UTF-8 JSON ≤8 MiB; 1–1,000 observers and 1–10,000 complete windows.
- Sun only; 1900–2100. All internal probes beyond the supported years give out-of-range/unknown.
- 4,000,000 actual examined interval days, 100,000 event records and 100,000,000 combined ephemeris evaluations, including both 370-day lookaheads.
- All six physical files together ≤64 MiB; full text and copy each ≤64 MiB, typed ≤128 MiB, aggregate ≤320 MiB and wire ≤1 GiB. All allocations jointly reserve the original 256 MiB memory before allocation.
- One 120-second deadline includes capture, read, calculation, encoding, cleanup and first publication. Joint capacity and browser acceptance remain unmeasured.
- Times are approximate at minute scale. Weather, refraction and the visible horizon can change an observed crossing.
- This calculation adopts the fixed mature model’s single-crossing bracket and Nyquist assumptions. After a resolved event, the complete interval from its original returned time to the original positive bracket endpoint is recorded as that same event and consumed before continuing. The original 0.1-second numerical tolerance and sub-second pair exclusion remain; this is not a mathematical certificate or an observation precision guarantee. Crossings near altitude extrema, tangencies and sub-second pairs are outside the completeness guarantee; observed unsupported or touching evidence remains unknown.