TAF decoder: from raw forecast to plain English
A TAF decoder splits a terminal aerodrome forecast into its base conditions and change groups, then rewrites each period in plain language with real dates and times. The decoder on this page applies the metar-taf-parser library to the pasted text and shows issue time, validity, TX and TN temperatures, and each BECMG, TEMPO, FM and PROB group. It does not display the US wind shear group. It is a reading aid only.
Decoding runs in your browser with the open-source metar-taf-parser library (MIT). Nothing is sent to a server. Not for operational use.
Key points
- The decoder prints the station, the issue time, the validity period, the base conditions and then one block per change group.
- Times are shown as day of month and hour in UTC; an end hour of 24 is shown as 24:00.
- FM groups are shown with minutes, and PROB, TEMPO and BECMG groups with their own validity periods.
- The wind shear group of US TAFs (WS followed by height and wind) is not displayed by this decoder.
- The word TAF is added automatically when it is missing, and a trailing equals sign is removed.

What the decoder does
The decoder on this page takes a raw terminal aerodrome forecast, separates the base forecast from its change groups and writes each period in plain English. The pasted text is cleaned first: spaces at both ends are trimmed, repeated spaces are collapsed, a final equals sign is removed, and the word TAF is added at the start when it is absent. The string is then passed to the open-source metar-taf-parser library, version 9.1.3, released under the MIT licence. The table is a presentation of what the library returns; no forecast rule is written specifically for this site.
When the library cannot read the text, it raises an error, and the page shows that the weather string is invalid. The decoder does not try to repair a damaged forecast.
The rows of the table
The rows follow the order of the forecast. A row is shown only if the forecast contains the corresponding element.
| Row | What it shows |
|---|---|
| Station | The ICAO location indicator |
| AMD | A note when the forecast is an amendment |
| Issued | Day of month and time of issue in UTC |
| Validity | Start and end as day and hour in UTC |
| Base conditions | Wind, visibility, weather and clouds of the first period |
| TX and TN | Maximum and minimum temperature with the day and hour |
| Change groups | One block each for PROB, TEMPO, BECMG and FM |
| Remarks | Remark text, when present |
The structure of the forecast itself is described on the how to read a TAF page and the validity rules in the AIM.
How the groups are interpreted
The validity group has the form DDHH/DDHH. In the example 1906/2012, the decoder prints day 19 at 06:00 to day 20 at 12:00 UTC. The AIM states that a TAF is valid for 24 or 30 hours and that the time groups use the day of the month and the hour.
A FM group has the form FMDDHHMM, which is why the decoder prints its start with minutes. The AIM gives the example that FM100000 is 0000Z on the 10th. FM marks a rapid change, usually under one hour, with a complete new set of conditions. The FM group page explains the consequences.
BECMG and TEMPO carry their own validity of the form DDHH/DDHH. BECMG is a gradual change, with only the changing elements listed, and TEMPO is a temporary fluctuation. The BECMG page and the TEMPO page cover the differences. PROB30 and PROB40 give a probability of 30 or 40 percent and are used for thunderstorms or precipitation, as explained on the PROB page.
Inside each block, wind, visibility, weather and cloud text are produced with the same logic as in the METAR decoder. A visibility such as P6SM is shown as "6 SM (9.7 km) or more". A calm wind is 00000KT, which the AIM uses when the wind is 3 kt or less.
A worked example
The forecast below is constructed for this page and is not a real bulletin.
TAF EGLL 190500Z 1906/2012 24010KT 9999 SCT025 BECMG 1910/1912 27015G25KT TEMPO 1914/1918 4000 SHRA BKN012 PROB30 1918/1922 TSRA FM200200 31008KT CAVOK
The decoder returns these elements.
- Station EGLL, issued on day 19 at 05:00 UTC, valid from day 19 06:00 to day 20 12:00 UTC.
- Base conditions: wind from 240 degrees at 10 kt, visibility 10 km or more, scattered clouds at 2,500 ft.
- BECMG from day 19 10:00 to day 19 12:00 UTC: wind from 270 degrees at 15 kt, gusting 25 kt.
- TEMPO from day 19 14:00 to day 19 18:00 UTC: visibility 4000 m, showers of rain, broken clouds at 1,200 ft.
- PROB30 from day 19 18:00 to day 19 22:00 UTC: thunderstorm with rain.
- FM at day 20 02:00 UTC: wind from 310 degrees at 8 kt and CAVOK.
The decoder repeats the CAVOK label in the FM block, a cosmetic quirk of the page and not of the forecast.
A US example with temperatures
A US-style forecast exercises the SM unit, the cloud type and the temperature groups.
TAF KJFK 191720Z 1918/2024 28012G22KT P6SM FEW050 SCT250 FM200000 31008KT P6SM SKC TEMPO 2004/2008 3SM BR PROB30 2010/2014 1SM TSRA SCT008CB
The decoder shows validity from day 19 18:00 to day 20 24:00 UTC, visibility 6 SM or more, TEMPO visibility 3 SM (4.8 km), and a PROB30 block with scattered clouds at 800 ft CB. A group such as TX20/1921Z TN05/2012Z is shown as a maximum of 20 degrees Celsius on day 19 at 21:00 UTC and a minimum of 5 degrees Celsius on day 20 at 12:00 UTC. The TX and TN page explains the groups.
How to read the result
The result is only a translation. A TAF must be read as a timeline: the base forecast applies from the start of validity until a change group takes over. An FM group replaces everything before it, while BECMG and TEMPO carry only the elements that change, which means the other elements continue from the previous period. The decoder prints the blocks in order but does not merge them into a single state for each hour. Comparing a forecast with the latest observation is covered on the TAF versus METAR page.
Limits of automatic decoding
- Wind shear group: a US TAF can carry a group such as
WS015/30045KT. The parser accepts it, but this page does not display it. The group is silently left out of the table. - Remarks: only remarks that the library recognizes are described.
- Date context: the decoder shows the day of month without a month, because the TAF has none.
- Currency: the decoder does not check whether the forecast has been amended, corrected or cancelled.
- Differences: national practice varies, for example in the use of BECMG and PROB40, and the AIM states that the National Weather Service does not use BECMG in the TAF.
A schematic report
TAF LXXX 190500Z 1906/2006 24010KT 9999 SCT025 BECMG 1912/1914 30015KT TEMPO 1915/1918 3000 SHRA BKN012
The decoder returns the station LXXX, issue on day 19 at 05:00 UTC, validity from day 19 06:00 to day 20 06:00 UTC, base wind from 240 degrees at 10 kt, visibility 10 km or more and scattered clouds at 2,500 ft. The BECMG block shows wind from 300 degrees at 15 kt from 12:00 to 14:00 UTC, and the TEMPO block shows visibility 3000 m, showers of rain and broken clouds at 1,200 ft between 15:00 and 18:00 UTC. The station code and the forecast are schematic.
In real reports
Real forecasts are rarely as short as the examples. A busy airport can have several FM groups, a BECMG followed by a TEMPO, and a PROB group inside the same day. Forecasts from the United States use statute miles, P6SM and the SKC code, while forecasts from most other countries use meters, CAVOK and NSC. Amended forecasts carry AMD after the type, and the validity can be 24 or 30 hours. The strings in this article are constructed to exercise the decoder, and the decoder on this page accepts any string pasted in.
Method and limits
The behavior of the decoder was checked by running its code on the strings above and reading the rows returned, including a forecast with a wind shear group, which was decoded without an error and without a row for that group. The parsing is the work of metar-taf-parser 9.1.3, whose project page states the MIT licence. The coding rules for the groups follow the Aeronautical Information Manual (chapter 7, section 1), which gives the 24 or 30 hour validity, the FM, BECMG, TEMPO and PROB definitions, and the US wind shear group, and the WMO Users' Handbook (WMO-No. 782) for international practice. The label wording, the date format and the conversion of statute miles to kilometers (1.609) belong to the page code and are not a standard. The site is a learning resource and is not an operational weather source.
Source code of this tool (MIT)
// Tafrob METAR/TAF decoder - MIT licence (c) Tafrob.
// Parsing is done by the open-source metar-taf-parser library (MIT, https://github.com/aeharding/metar-taf-parser).
import { parseMetar, parseTAF } from 'metar-taf-parser';
import fr from 'metar-taf-parser/locale/fr.js';
import en from 'metar-taf-parser/locale/en.js';
type Lang = 'en' | 'fr';
export type Row = [string, string];
const W: Record<string, [string, string]> = {
'-': ['light', 'faible'], '+': ['heavy', 'fort'], VC: ['in the vicinity', 'au voisinage'],
MI: ['shallow', 'mince'], BC: ['patches of', 'bancs de'], PR: ['partial', 'partiel'], DR: ['low drifting', 'chasse-basse de'],
BL: ['blowing', 'chasse-haute de'], SH: ['showers of', 'averses de'], TS: ['thunderstorm', 'orage'], FZ: ['freezing', 'se congelant'],
DZ: ['drizzle', 'bruine'], RA: ['rain', 'pluie'], SN: ['snow', 'neige'], SG: ['snow grains', 'neige en grains'], IC: ['ice crystals', 'cristaux de glace'],
PL: ['ice pellets', 'granules de glace'], GR: ['hail', 'grêle'], GS: ['small hail / snow pellets', 'grésil'], UP: ['unknown precipitation', 'précipitation inconnue'],
BR: ['mist', 'brume'], FG: ['fog', 'brouillard'], FU: ['smoke', 'fumée'], VA: ['volcanic ash', 'cendres volcaniques'], DU: ['widespread dust', 'poussière'],
SA: ['sand', 'sable'], HZ: ['haze', 'brume sèche'], PY: ['spray', 'embruns'], PO: ['dust/sand whirls', 'tourbillons de poussière'], SQ: ['squalls', 'grains'],
FC: ['funnel cloud', 'trombe'], SS: ['sandstorm', 'tempête de sable'], DS: ['duststorm', 'tempête de poussière'], NSW: ['no significant weather', 'fin du temps significatif'],
};
const Q: Record<string, [string, string]> = {
SKC: ['sky clear', 'ciel clair'], FEW: ['few (1-2 oktas)', 'peu (1 à 2 octas)'], SCT: ['scattered (3-4 oktas)', 'épars (3 à 4 octas)'],
BKN: ['broken (5-7 oktas)', 'fragmenté (5 à 7 octas)'], OVC: ['overcast (8 oktas)', 'couvert (8 octas)'], NSC: ['no significant cloud', 'pas de nuage significatif'],
};
const L = {
en: { type: 'Report type', station: 'Station', time: 'Observation time', wind: 'Wind', vis: 'Visibility', wx: 'Weather', clouds: 'Clouds', vv: 'Vertical visibility', temp: 'Temperature / dew point', alt: 'Pressure', rvr: 'Runway visual range', rmk: 'Remarks', trend: 'Trend', validity: 'Validity', issued: 'Issued', calm: 'calm', variable: 'variable', from: 'from', gust: 'gusting', varies: 'direction varying between', cavok: 'CAVOK: visibility 10 km or more, no cloud below 5,000 ft or minimum sector altitude, no CB/TCU, no significant weather', ft: 'ft', base: 'at', nosig: 'NOSIG: no significant change expected in the next 2 hours', tx: 'Max temperature', tn: 'Min temperature', prob: 'probability', day: 'day', auto: 'AUTO: fully automated report', utc: 'UTC', minvis: 'minimum visibility', ndv: 'no directional variation available', more: 'or more', less: 'less than', amd: 'AMD: amended forecast', ws: 'Low-level wind shear' },
fr: { type: 'Type de message', station: 'Station', time: 'Heure d’observation', wind: 'Vent', vis: 'Visibilité', wx: 'Temps présent', clouds: 'Nuages', vv: 'Visibilité verticale', temp: 'Température / point de rosée', alt: 'Pression', rvr: 'Portée visuelle de piste', rmk: 'Remarques', trend: 'Tendance', validity: 'Validité', issued: 'Émis', calm: 'calme', variable: 'variable', from: 'du', gust: 'rafales à', varies: 'direction variant entre', cavok: 'CAVOK : visibilité de 10 km ou plus, pas de nuage sous 5 000 ft ou l’altitude minimale de secteur, ni CB ni TCU, pas de temps significatif', ft: 'ft', base: 'à', nosig: 'NOSIG : pas de changement significatif prévu dans les 2 heures', tx: 'Température maximale', tn: 'Température minimale', prob: 'probabilité', day: 'jour', auto: 'AUTO : message entièrement automatique', utc: 'UTC', minvis: 'visibilité minimale', ndv: 'pas de variation directionnelle disponible', more: 'ou plus', less: 'moins de', amd: 'AMD : prévision amendée', ws: 'Cisaillement de basse couche' },
};
const TR: Record<string, [string, string]> = { U: ['rising', 'en hausse'], D: ['falling', 'en baisse'], N: ['no change', 'stable'] };
const p2 = (n?: number) => String(n ?? 0).padStart(2, '0');
function wind(w: any, l: Lang) {
const t = L[l];
if (!w) return '';
if (w.speed === 0) return t.calm;
const unit = w.unit === 'KT' ? (l === 'en' ? 'kt' : 'kt') : w.unit === 'MPS' ? 'm/s' : 'km/h';
let s = w.degrees === undefined ? `${t.variable}, ${w.speed} ${unit}` : `${t.from} ${p2(w.degrees).padStart(3, '0')}° ${l === 'en' ? 'at' : 'à'} ${w.speed} ${unit}`;
if (w.gust) s += `, ${t.gust} ${w.gust} ${unit}`;
if (w.minVariation !== undefined) s += `; ${t.varies} ${w.minVariation}° ${l === 'en' ? 'and' : 'et'} ${w.maxVariation}°`;
return s;
}
function vis(v: any, l: Lang) {
const t = L[l];
if (!v) return '';
const val = v.unit === 'SM' ? `${v.value} SM (${(v.value * 1.609).toFixed(1)} km)` : v.value >= 9999 ? `10 km ${t.more}` : `${v.value} m`;
let s = v.indicator === 'P' && v.value < 9999 ? `${val} ${t.more}` : v.indicator === 'M' ? `${t.less} ${val}` : val;
if (v.min) s += `; ${t.minvis} ${v.min.value} m (${v.min.direction})`;
if (v.ndv) s += ` (${t.ndv})`;
return s;
}
function weather(ws: any[], l: Lang) {
const i = l === 'en' ? 0 : 1;
return ws
.map((w) => {
const desc = w.descriptive === 'TS' && w.phenomenons.length ? (l === 'en' ? 'thunderstorm with' : 'orage avec') : w.descriptive && W[w.descriptive]?.[i];
const core = [desc, ...w.phenomenons.map((p: string) => W[p]?.[i] ?? p)].filter(Boolean).join(' ');
if (!w.intensity) return core;
if (w.intensity === 'VC') return `${core} ${W.VC[i]}`;
return l === 'en' ? `${W[w.intensity][0]} ${core}` : `${core} (${W[w.intensity][1]})`;
})
.join('; ');
}
function clouds(cs: any[], l: Lang) {
const i = l === 'en' ? 0 : 1;
return cs
.map((c) => `${Q[c.quantity]?.[i] ?? c.quantity}${c.height !== undefined ? ` ${L[l].base} ${c.height.toLocaleString(l === 'en' ? 'en-US' : 'fr-FR')} ft` : ''}${c.type ? ` ${c.type}` : ''}`)
.join('; ');
}
function common(o: any, l: Lang, rows: Row[]) {
const t = L[l];
if (o.wind) rows.push([t.wind, wind(o.wind, l)]);
if (o.cavok) rows.push(['CAVOK', t.cavok]);
if (o.visibility && !o.cavok) rows.push([t.vis, vis(o.visibility, l)]); // CAVOK already implies 10 km or more
if (o.windShear) rows.push([t.ws, `${o.windShear.height.toLocaleString(l === 'en' ? 'en-US' : 'fr-FR')} ft ${l === 'en' ? 'AGL' : 'sol'}: ${wind(o.windShear, l)}`]);
if (o.weatherConditions?.length) rows.push([t.wx, weather(o.weatherConditions, l)]);
if (o.verticalVisibility !== undefined) rows.push([t.vv, `${o.verticalVisibility} ft`]);
if (o.clouds?.length) rows.push([t.clouds, clouds(o.clouds, l)]);
}
export function decodeMetar(raw: string, l: Lang): Row[] {
const m: any = parseMetar(raw.trim().replace(/\s+/g, ' ').replace(/=$/, ''), { locale: l === 'fr' ? fr : en });
const t = L[l];
const rows: Row[] = [];
rows.push([t.type, m.type ?? 'METAR']);
rows.push([t.station, m.station]);
if (m.day) rows.push([t.time, `${t.day} ${m.day}, ${p2(m.hour)}:${p2(m.minute)} ${t.utc}`]);
if (m.auto) rows.push(['AUTO', t.auto]);
common(m, l, rows);
for (const r of m.runwaysInfo ?? []) {
if (r.minRange !== undefined) rows.push([`${t.rvr} ${r.name}`, `${r.indicator === 'P' ? '> ' : r.indicator === 'M' ? '< ' : ''}${r.minRange}${r.maxRange ? `-${r.maxRange}` : ''} ${r.unit ?? 'm'}${r.trend ? ` (${TR[r.trend]?.[l === 'en' ? 0 : 1] ?? r.trend})` : ''}`]);
}
if (m.temperature !== undefined) rows.push([t.temp, `${m.temperature} °C / ${m.dewPoint ?? '?'} °C`]);
if (m.altimeter) rows.push([t.alt, m.altimeter.unit === 'inHg' ? `${m.altimeter.value.toFixed(2)} inHg (${(m.altimeter.value * 33.8639).toFixed(0)} hPa)` : `${m.altimeter.value} hPa (${(m.altimeter.value / 33.8639).toFixed(2)} inHg)`]);
if (m.nosig) rows.push(['NOSIG', t.nosig]);
for (const tr of m.trends ?? []) rows.push([`${t.trend} ${tr.type}`, trendText(tr, l)]);
for (const r of m.remarks ?? []) rows.push([`${t.rmk}: ${r.raw ?? ''}`, r.description ?? r.raw ?? '']);
return rows;
}
function trendText(tr: any, l: Lang) {
const rows: Row[] = [];
common(tr, l, rows);
return rows.map(([k, v]) => `${k}${l === 'fr' ? ' :' : ':'} ${v}`).join(' | ') || tr.raw;
}
export function decodeTaf(raw: string, l: Lang): Row[] {
let txt = raw.trim().replace(/\s+/g, ' ').replace(/=$/, '');
if (!/^TAF\b/.test(txt)) txt = 'TAF ' + txt;
const x: any = parseTAF(txt, { locale: l === 'fr' ? fr : en });
const t = L[l];
const rows: Row[] = [];
rows.push([t.station, x.station]);
if (x.amendment) rows.push(['AMD', t.amd]);
if (x.day) rows.push([t.issued, `${t.day} ${x.day}, ${p2(x.hour)}:${p2(x.minute)} ${t.utc}`]);
const v = x.validity;
if (v) rows.push([t.validity, `${t.day} ${v.startDay} ${p2(v.startHour)}:00 → ${t.day} ${v.endDay} ${p2(v.endHour)}:00 ${t.utc}`]);
common(x, l, rows);
if (x.maxTemperature) rows.push([t.tx, `${x.maxTemperature.temperature} °C, ${t.day} ${x.maxTemperature.day} ${p2(x.maxTemperature.hour)}:00 ${t.utc}`]);
if (x.minTemperature) rows.push([t.tn, `${x.minTemperature.temperature} °C, ${t.day} ${x.minTemperature.day} ${p2(x.minTemperature.hour)}:00 ${t.utc}`]);
for (const tr of x.trends ?? []) {
const vv = tr.validity;
const when = tr.type === 'FM'
? `${t.day} ${vv.startDay} ${p2(vv.startHour)}:${p2(vv.startMinutes)} ${t.utc}`
: `${t.day} ${vv.startDay} ${p2(vv.startHour)}:00 → ${t.day} ${vv.endDay} ${p2(vv.endHour)}:00 ${t.utc}`;
const label = `${tr.probability ? `PROB${tr.probability} ` : ''}${tr.type === 'PROB' ? '' : tr.type}`.trim();
rows.push([`${label} (${when})`, trendText(tr, l)]);
}
for (const r of x.remarks ?? []) rows.push([`${t.rmk}: ${r.raw ?? ''}`, r.description ?? r.raw ?? '']);
return rows;
}
Training and reference only. For flight planning, always use official briefing sources (your national weather service, AIS or approved provider).
Frequently asked questions
How do I read a TAF step by step?
Read the header first: station, issue time and validity. Then read the base forecast, which is complete, and apply each change group in order. FM replaces the conditions, BECMG describes a gradual change, TEMPO a temporary fluctuation and PROB a probability. The decoder on this page prints the same sequence as separate blocks.
What does the decoder do when the word TAF is missing?
It adds the word TAF at the start before parsing, removes a final equals sign and collapses repeated spaces. This lets a forecast copied from a bulletin be decoded without editing. A string that still does not follow the TAF structure is rejected with an invalid weather string message.
Why is the wind shear group missing in the output?
The parser accepts the US wind shear group, but the table of this page does not print it. A TAF with a group such as WS015/30045KT is decoded without an error, and the group is silently left out. The original TAF must be read to see it.
How are the times of a TAF shown?
The decoder shows day of month and hour in UTC, for example from day 19 06:00 to day 20 12:00. A period that ends at hour 24 is shown as 24:00 of the same day. A month is never shown, because the TAF itself carries only the day and the hour.
Can the decoder replace an official briefing?
No. It is an educational tool that translates codes. A forecast can be amended, corrected or cancelled, and a briefing from an authorized source also gives NOTAMs and other products. The decoder does not check whether the TAF is current.
Sources
- GitHub, Alexander Harding. metar-taf-parser (source code and documentation)
- Federal Aviation Administration (2025). Aeronautical Information Manual, chapter 7, section 1
- Federal Aviation Administration (2024). Aviation Weather Handbook (FAA-H-8083-28A)
- World Meteorological Organization (2025). Aerodrome Reports and Forecasts, A Users' Handbook to the Codes (WMO-No. 782)