white border + ! = ensemble mean already exceeds the FoM limit
no data — voids the score for that day
not applicable this phase
italic blue = advisory level above the mission altitude requirement: analysed for the ceiling, excluded from the exceedance combination
struck through = non-traversable, climb arrested at a subjacent level
red hatch on the criteria score = hour disqualified by the altitude requirement, whatever the score reads
white dot = T-72h empirical escalation applied
▲ under an hour = operative slot: the best hour in an elective window (launch / recovery)
·blue hour = D+1, the recovery window falls on the night after launch
dotted border = deterministic backfill. The ensemble could not supply this
cell — either the run ended before this hour, or the API publishes no spread for that
variable — so the value came from a single deterministic model run instead. One run has
no spread of its own, so its uncertainty was invented from a lead-time error-growth
curve rather than measured from disagreeing members. The forecast value may be perfectly
good; the confidence attached to it is an assumption, and the growth coefficients behind
it are untuned placeholders. Treat the probability as indicative, and do not let a
dotted cell carry a marginal GO on its own.
ENSEMBLE every scoring input from real ensemble spread
MIXED some hours backfilled
DETERMINISTIC all hours backfilled — treat probabilities as indicative