BEACON

Methodology

Matching

Key = normalized IAU designation. 1979 XB and 1979XB are the same object. We keep each source’s raw values; no averaging.

Close-approach comparisons use a shared inclusive window of 365 days from the reconcile reference date. JPL CAD is requested with the same horizon; ESA upcoming encounters outside that window are excluded from coverage and field comparisons.

Cumulative IP and Palermo compare source values as published. JPL riskWindowYears and ESA riskYears are shown side-by-side. Unequal aggregation windows are informational context— they do not create a separate risk divergence and do not suppress the cumulative thresholds.

Relative velocity: ESA publishes DD.D km/s. Before the 1% check, JPL is rounded to that 0.1 km/s precision. Displayed source values stay raw.

Counts

significantDivergences — IP, Palermo, or Torino past threshold.

totalFieldDivergences — everything flagged, including distance / v_rel / date.

/api/debug/divergences

Thresholds

FieldKindCutNotes
missDistanceAuorbital3% relJPL vs ESA miss distance; flag above 3% relative.
relativeVelocityorbital1% rel (ESA 0.1 km/s)Relative speed; round JPL to ESA's 0.1 km/s reporting precision, then flag above 1%. Raw values unchanged.
closeApproachDateorbitaldateCalendar day only (drop time).
cumulativeImpactProbabilityrisk>2× (floor 1e-7)Ratio >2×; ignore if both <1e-7.
palermoScaleCumulativerisk|Δ| > 0.5Absolute Palermo gap > 0.5.
torinoScaleMaxriskexactAny Torino mismatch counts.

miss 3% · v_rel 1% · IP ×2 · floor 1e-7 · PS |Δ| 0.5

Caveats