Skip to content

TAF Decoder

Paste a raw TAF and read the forecast in plain English: the period it covers, and every FM, BECMG, TEMPO and PROB group broken out on its own.

Nothing is fetched and nothing is uploaded. The forecast is read in this page, group by group, and every group the decoder has no rule for is listed rather than skipped.

How a forecast is put together

  1. Station, then the day and time the forecast was issued, in UTC
  2. The period it covers: day and hour of the start, day and hour of the end
  3. The base forecast — wind, visibility, weather and cloud, in METAR grammar
  4. FM, a clean break: everything after it replaces what came before
  5. BECMG, a change that settles somewhere inside a window
  6. TEMPO, conditions that come and go inside a window
  7. PROB30 and PROB40, a stated chance rather than a forecast
decoded
Station
EPWA
Issued
day 19 at 11:00 UTC
Valid
from day 19 at 12:00 UTC to day 20 at 18:00 UTC

Base conditions

Wind
from 270 degrees true at 12 kt
Visibility
10 km or more
Cloud
scattered (3 to 4 oktas) at 3000 ft

Temporarily, from day 19 at 18:00 UTC to day 19 at 22:00 UTC

Visibility
4 km
Weather
light rain showers
Cloud
broken (5 to 7 oktas) at 1200 ft

From day 20 at 06:00 UTC

Wind
from 300 degrees true at 15 kt, gusting to 25 kt
Visibility
10 km or more
Cloud
broken (5 to 7 oktas) at 2000 ft

A TAF is written in almost the same code as a METAR, and that is exactly what makes it easy to misread. The groups look identical — wind, visibility, weather, cloud — but a METAR describes one moment that has already happened, while a TAF describes a period that has not. Everything hangs off that: the validity period after the issue time, and the change groups that carve it up. FM starts a new set of prevailing conditions and replaces everything before it. BECMG says the change works its way in across a window. TEMPO says the conditions come and go inside one. PROB30 and PROB40 say the forecaster is not sure. This page walks those in order, prints each period on its own, and lists any group it has no published rule for instead of quietly skipping it. Nothing is fetched and nothing is uploaded.

How it is calculated

TAF CCCC YYGGggZ Y1Y1G1G1/Y2Y2G2G2 dddffKT VVVV w'w' NsNsNshshshs [FMYYGGgg] [BECMG|TEMPO|PROBnn YYGG/YeYeGeGe]

After the station and the issue time comes the pair of four-figure groups that says what the forecast covers: day and hour of the start, then day and hour of the end. An end at midnight is written 24 rather than 00, which is why a forecast can run to hour 24 of a day that also has an hour 00. Older forecasts, and some outside the United States, still carry the six-figure form — day, start hour, end hour — and this page reads both. What follows is the base forecast, and then any number of change groups. FM carries a date, an hour and a minute and everything after it supersedes what came before; BECMG, TEMPO and PROB carry a window with a start and an end and describe only what differs inside it. The rest of the groups are the METAR grammar unchanged, so the same reader handles them.

Source: NWS/FAA Key to Aerodrome Forecast (TAF) and Aviation Routine Weather Report (METAR), worked example TAF KPIT 091730Z 0918/1024 — 0918/1024 is the "Valid period, either 24 hours or 30 hours. The first two digits of EACH four digit number indicate the date of the valid period, the final two digits indicate the time (valid from 18Z on the 9th to 24Z on the 10th)"

Questions people ask

What is the difference between FM, BECMG and TEMPO?
They divide the forecast in three different ways. FM is a clean break at a stated minute: everything after it replaces the conditions before it, and the handbook requires every element to be repeated even if it has not changed. BECMG is a transition that settles somewhere inside a window and then holds. TEMPO is a fluctuation: conditions that come and go, each spell under an hour, adding up to less than half the window. Only FM redefines the prevailing weather; the other two describe departures from it.
What does PROB30 actually mean?
That the forecaster puts the chance of that condition at about thirty percent in the stated window. Only 30 and 40 are used — a lower chance is not worth forecasting, and above 40 the condition belongs in the prevailing group or in a TEMPO. You will also see PROB30 TEMPO together, which means a thirty percent chance of a fluctuating condition, and that is decoded here as both.
Why does the validity end at hour 24?
Because midnight closing a forecast is written 24, not 00, so that the day it belongs to is unambiguous. A forecast running to 0924 ends at midnight at the end of the 9th; one starting at 1000 starts at midnight beginning the 10th. This page prints the hour the way the code carries it rather than silently rolling it over to the next day.
I pasted a METAR and it complained.
On purpose, and it says so rather than trying. A METAR has no validity period, and after the time it goes straight into the observed wind, then carries a temperature and an altimeter setting that a TAF never has. When this page sees that shape it tells you it is looking at an observation and sends you to the METAR decoder, which is a more useful answer than a decode of the wrong grammar.
What will it not decode?
Temperature forecast groups such as TX15/0918Z and TN04/1006Z, icing and turbulence groups from military forecasts, wind speeds given in metres per second or kilometres per hour, and everything after RMK. Those are listed by name rather than interpreted. The reason is the same one that runs through the rest of this site: a decoder that guesses at a group is indistinguishable, from the outside, from one that read it correctly.
Is the wind in a TAF true or magnetic?
True. The forecast wind is a three-digit direction relative to true north, rounded to ten degrees — the same convention as a METAR. The wind read out by a tower, an ATIS or an AWOS broadcast is magnetic instead, and runway numbers are magnetic too. It matters the moment you turn a forecast wind into a crosswind for a particular runway, which is a different page on this site.
Does it fetch the current forecast for an airport?
No. There is no network call on this page at all. You bring the forecast — from an aviation weather service, a briefing pack or a printout — and it is decoded in your browser. That also means it works on an old forecast, on a made-up one, and with no connection.

Related tools

Found a problem, or want more?

A number that disagrees with its source is a defect, not a rounding preference.

What did you enter, what did the tool show, and what did you expect instead? If you have a source that disagrees with ours, a link to it is the most useful thing you can send.

Write to us

Opens your mail app with the page and tool already filled in.