计划发射, window, T-0: how to read a launch schedule without guessing

计划发射 means "no earlier than"

Every time on a launch schedule is a 计划发射 — No Earlier Than. The earliest moment the vehicle could leave the pad, given everything currently known. It is a floor, not an appointment. A date can slip forward (rarely) or backward (constantly), and a 计划发射 on a board is always provisional.

When a launch slips, the new 计划发射 is usually announced within hours. When it is scrubbed at T-30 seconds, the vehicle is generally recycled to a backup opportunity — sometimes the same day, sometimes days later, depending on the orbit and the range.

Windows: instantaneous versus hours long

A launch window is the period during which the vehicle may lift off and still reach the intended orbit. It can be seconds long or several hours:

• Instantaneous window — the vehicle must launch at a specific second. This happens when the target is in a precise position, for example a rendezvous with a station or a planetary trajectory. Miss it and you wait for the next planetary alignment, which can be months.
• Long window — anything from a few minutes to a couple of hours. The vehicle can hit the target orbit across a range of times, so weather or a minor technical issue does not necessarily force a scrub.

A schedule that shows the same 开始 and end time is telling you it is an instantaneous window.

T-minus, holds and scrubs

T-0 is liftoff. Everything else is measured from it: T-10 minutes, T-3 seconds. The countdown is not a straight line — it contains planned holds, pauses where the team checks something before continuing. A hold is not a problem; it is scheduled.

An unplanned hold is. If the team cannot resolve it inside the window, the launch is scrubbed, and the next attempt becomes a new 计划发射.

Why "90% probability" is not about the rocket

The probability figure on a launch record is almost always weather probability — the chance that conditions satisfy the launch criteria at the opening of the window. It says nothing about vehicle reliability. A 90% weather forecast can still be followed by a scrub for a valve, and a 40% forecast sometimes launches on the one clear minute.

The criteria themselves are stricter than "is it raining": ground winds, upper-level winds, the chance of triggered lightning, cloud thickness, and for some missions the recovery conditions hundreds of kilometres downrange.

Why so many entries are TBD

Because the schedule is public-facing and the planning is not finished. A launch can appear on a board months ahead with a placeholder date and no fixed time, then firm up two weeks out. Manufacturers and range authorities publish updates on their own timelines, so a launch board is a snapshot of other people's uncertainty.

That is why our board prints the age of its snapshot on the page. A schedule that does not tell you when it was fetched is asking you to trust a number that may be a week old.

Why the countdown should use your own clock

Time zones are the single most common source of missed launches. A 计划发射 published as 2026-10-30 18:21 UTC is 02:21 the next morning in Beijing, 19:21 the previous evening in London, and 13:21 the same afternoon in Chicago. Every one of those is the same instant, and at least one person will read the wrong one.

The fix is boring and effective: keep the canonical time in UTC, and compute everything else against the clock on the device that is showing it. Our launch pages tick their countdown from your local clock, so a device with a wrong timezone will be visibly wrong rather than quietly wrong.

How to actually follow one

Pick a launch with a fixed window, open its mission brief, add it to your calendar with the .ics button, and then check the schedule again the day before — the 计划发射 will probably have moved by minutes, and occasionally by days. That is not the board being unreliable; that is how launch scheduling works.