Scan departure times

A departure scan solves the same course from several start times and ranks the results. It answers "is this start any good", which for a delivery or a passage is often the more valuable question than "which way do I go".

Setting one up

Under Also try later starts, choose four or eight starts at six-hour intervals. The scheduled departure remains the baseline; the later starts show what waiting would buy or cost.

Each start is a separate search over the same forecast, so a scan does not spend extra forecasts from your allowance. It does cost time: an eight-start scan is eight searches back to back, and that is a real wait. Fidelity is selected automatically from your plan.

Scans are solved under the consensus model, and the panel says which one that was.

Reading it

The ranking is by elapsed time, but the ranking alone is not the interesting part.

A flat spread means departure time barely matters. Take the scheduled start and stop thinking about it.

A sharp minimum means one window is much better than its neighbours. Look at what changes around it — usually a front arriving, or a tide turning — and decide how much you trust the model's timing of it, because a two-hour error in that timing moves the minimum.

A steady trend means you are racing the weather, not a window. Leaving earlier is simply better, up to whatever the trend is chasing.

Plan limits

Anonymous gets one start, Free gets three, paid tiers get up to 24. A request over the limit is rejected with plan_limit_scan_count rather than being quietly trimmed, so you know you did not get what you asked for.

When not to bother

For a fixed-start race, the scan tells you nothing you can act on. It is for deliveries, passages and pursuit starts.