01_GAUSS · SAMPLE EVIDENCE REPORT

Sample evidence report.

Synthetic · screening evidence

The shape of what a pilot delivers, filled with synthetic data. It is not a mission record. Your report covers your own telemetry and inference logs.

Scope: One asset group, one historical replay, one near-live feed
Window: Six hours around a G3 geomagnetic storm with an S2 proton event (synthetic)
Inputs: CCSDS packets with an XTCE dictionary, an onboard-AI inference log, public NOAA space-weather data
Acceptance outcome: Every inference produced from stale or corrupted inputs is flagged before anyone acts on it

01_RECORD_STATES

What held up, and what didn't.

State	Records	Meaning
Nominal	1,240	Fresh inputs, provenance complete, inside expected bounds
Degraded	96	Usable with lowered confidence; produced under elevated proton levels
Abstain	18	No output given: the inputs could not be trusted
Unknown	41	Input missing or older than the freshness limit; never shown as nominal
Blocked	7	Packet integrity check failed; record excluded

Synthetic counts, for illustration.

02_DECISIONS

Flagged 96 inferences made during the proton event, with the conditions at the time attached
Withheld 18 outputs and used the deterministic fallback, marked as such
Kept 41 records unknown through a 12-minute link gap during the geomagnetic disturbance
Excluded 7 records that failed packet integrity checks

03_UNKNOWNS_AND_LIMITS

No hardware evidence for the target processor, so no radiation-effect figures are given
Model calibration was not assessed in this sample
Simulation results would be labelled as simulation; this whole page is synthetic
Advisory only: decisions stay with the operators

04_PROVENANCE

Each exported decision carries its source, time, freshness and uncertainty. A sample row:

record	source	received	age	state	policy	reason
pkt-0412	CCSDS APID 0x2A1	03:14:07Z	2 s	degraded	degrade	bit-error rate above window limit during the S2 proton event