TAF Traker keeps every Terminal Aerodrome Forecast (TAF) issued for about 800 airports in the continental US,
watches what actually happens at each one every five minutes, and grades every forecast against it. Its one big question is:
is the weather worse than the TAF said it would be?
Experimental, not for flight planning. TAF Traker is a study of forecast accuracy, not a weather briefing.
Weather data is shown as published by NOAA and the National Weather Service, and everything derived from it here can be wrong.
Always get an official briefing. Terms & disclaimer
Open the map. Dots are airports colored by their current flight category; rings mark airports where conditions are worse than the TAF in effect.
Hover a ringed airport for a quick summary, then click it for the full comparison and a 6-hour TAF-vs-actual strip.
Click Open airport detail. The top chart shows every forecast version stacked over what actually happened.
Click any hour in that chart to see how the forecast for that hour changed as it got closer.
Visit Performance to see which airports' TAFs tend to run optimistic.
All times are UTC ("Z") unless a label says local. Press / on any page to jump to the airport search.
The basics
Flight categories
Everything on the site is organised around the FAA flight categories, worked out from the ceiling (the lowest broken,
overcast or obscured layer) and the visibility. The worse of the two decides.
Category
Ceiling
Visibility
VFR
above 3,000 ft
and
more than 5 SM
MVFR
1,000 – 3,000 ft
or
3 – 5 SM
IFR
500 – 999 ft
or
1 – less than 3 SM
LIFR
below 500 ft
or
less than 1 SM
The colors are the standard aviation ones and are used the same way everywhere: green VFR, blue MVFR, red IFR, magenta LIFR.
Gray means no current data.
What a TAF says
A TAF covers 24 or 30 hours and changes over time through change groups. TAF Traker decodes all of them:
From that minute, a complete new forecast replaces everything before it. Above: from 03Z the 29th, IFR in mist with an 800 ft overcast.
TEMPO
Temporary fluctuations during the window. Only the elements listed change; the rest stay as the prevailing forecast. Above: between 05Z and 09Z, expect occasional ½ SM fog with a 200 ft vertical visibility (LIFR).
PROB30 / PROB40
A 30 or 40% chance of the listed conditions during the window. Treated like TEMPO.
BECMG
Conditions becoming different gradually during the window, then holding. Either state may occur during the change.
AMD / COR
An amended (unscheduled update) or corrected TAF. Routine TAFs come out four times a day; amendments whenever the forecaster changes their mind.
"The TAF in effect"
Each airport usually has several TAF versions covering the same hour, because new ones are issued every few hours. The
TAF in effect at any moment is the most recent one already issued and valid at that time: the one a pilot would have
been reading. The map and the main verdicts always use the TAF in effect. Airport pages also show and grade every older version.
Where "actual" comes from
5-minute ASOS
Automated observations every 5 minutes from most airports (via NOAA's MADIS). Usually 15–25 minutes behind real time. This is what most grading uses.
METAR
The routine hourly report, usually issued just before the hour. Arrives within a minute or two.
SPECI
A special report issued between METARs when conditions change significantly. Often the freshest sign that things just got worse.
"Current conditions" are whichever of these is newest. About 220 airports (mostly military fields and some smaller
airports) have no 5-minute feed and rely on METAR/SPECI alone.
MOS: the forecast before the TAF
GFS MOS is NOAA's statistical model guidance for each airport, issued four times a day and reaching about 72 hours
out, much further than a TAF. TAF Traker uses it for the run-up: the forecasts that existed before the first TAF reached
a given hour. MOS gives ceiling and visibility as categories (ranges) rather than exact values.
Lead time
Lead time is how far ahead a forecast was issued: a TAF issued at 12Z forecasting 18Z has a 6-hour lead. Forecasts
generally get more accurate as lead time shrinks, and several charts break accuracy down by lead time.
The six verdicts
Every observation is graded against the TAF in effect at that minute and gets exactly one verdict. They're ordered from
fine to most concerning, and the rings on the map use the same colors.
As forecast
Same flight category as the prevailing forecast. Small differences inside the category are fine.
Better
A better category than forecast: a cautious TAF. Not a safety concern, but it shows up in the performance stats.
Lower clouds / vis
Same category, but meaningfully worse than forecast: a ceiling or a scattered-or-thicker layer well below what was
forecast (below 5,000 ft), or visibility well below forecast. Example: BKN040 observed when the TAF said SCT045. Still VFR,
but there's now a ceiling at 4,000 ft that wasn't forecast.
Inside TEMPO
Worse than the prevailing forecast, but within what a TEMPO or PROB group allowed for at that time. The forecaster
flagged the possibility; it happened.
Miss
One flight category worse than anything the TAF allowed, TEMPO and PROB groups included.
Example: MVFR observed, TAF said VFR with no TEMPO.
Bust
Two or more categories worse than anything the TAF allowed. Example: LIFR fog observed, TAF said VFR.
These are the ones that matter most.
"Worse than anything forecast" means miss + bust. "Worse in any way" means lower clouds/vis + inside TEMPO + miss + bust.
Wind verdicts (graded separately)
Wind gets its own verdict, independent of flight category. An airport can be as forecast for category and a wind miss at the same time.
Small differences don't count; only wind or gusts well above anything the TAF allowed:
Wind miss
Sustained wind 10+ kt above the highest forecast speed (and at least 15 kt), or gusts 10+ kt above the highest forecast gust (and at least 20 kt). Example: 20 kt observed, TAF said 10 kt.
Wind bust
The same by 20+ kt (sustained at least 25 kt, or gusts at least 30 kt). Example: 15G35KT observed, TAF said 10 kt.
Inside TEMPO
Above the prevailing forecast wind by those margins, but within a TEMPO or PROB group's wind.
Not counted
7 kt vs a forecast 5 kt, lighter wind than forecast, and direction changes.
The map
Dots, rings and arrows
Dot color: the current flight category from the newest observation (5-minute ASOS, METAR or SPECI).
Faded dot: the newest observation is more than 75 minutes old. Gray: no recent observation.
Ring: the verdict against the TAF in effect (see the six verdicts). Misses glow and busts pulse so they catch your eye. No ring means as forecast or better.
Orange ▼ arrow: a sharp downward trend: the forecast was lowered (the TAF in effect now says worse for the next 12 hours than the TAF from 6+ hours ago), or the actual weather dropped fast in the last 3 hours. Teal ▲: the same, improving. Hover to see which.
Labels (airport codes) appear as you zoom in; misses and busts are labeled earlier.
Category, Wind or Look-back
The tabs at the top of the panel choose what the rings, list, counts and regions compare. Dot colors always show the current
flight category.
Category (the default): actual flight category, ceiling and visibility against the TAF in effect.
Wind: observed wind and gusts against the TAF's wind. The list shows wind against wind, and regions outline airports windier than forecast.
Look-back: the weather now against the forecast as it stood some hours ago. Pick TAF (3 to 24 hours ago)
or GFS MOS (6 to 60 hours ago). For a TAF it's each airport's newest TAF issued by then; for MOS it's the newest run
that had been published by then (a run comes out about 4 hours after its cycle time). Rings show worse (amber, orange, red as
usual) and, only in this view, teal for better than that forecast. It answers "how did yesterday's forecast for right
now pan out?" across the whole country at once.
Highlight regions
When at least three airports within 180 km of each other share something, the map can outline the area and label it. The
Highlight regions chips in the panel choose which outlines to draw; the number on each chip is how many regions there are now.
Click inside an outline for the list of airports.
Misses (solid, glowing): airports below forecast at the same time, at least one a miss or bust ("9 below forecast").
It usually means something larger-scale the forecasts missed: a stratus deck moving in faster than expected, a fog bank, a line of showers.
Forecast lowered (dashed coral): the TAFs there now forecast worse for the next 12 hours than the TAFs from 6+ hours ago.
This shows a region trending worse even where today's TAFs are still verifying, which is often the first sign of trouble.
Forecast raised (dashed teal): the opposite, forecasts improving.
In Look-back, the chips become Worse than forecast and Better than forecast.
The left panel
Live / Replay and the time being shown.
Conditions now: how many airports are in each category. Click a category to hide or show those airports on the map.
Verdict counts for the active tab, one row each; the faint bar behind a row shows its size. Click a row to list those airports.
Highlight regions: which region outlines to draw (above).
Trends (Category tab): how many forecasts got worse or better than the TAF from 6+ hours ago, and how many airports' actual weather dropped or improved sharply in the last 3 hours. Click a number to list them.
List: Below forecast (the default) lists every airport with a ring, worst first, and dims the others on the
map; Better lists airports doing better than forecast; Trends lists airports with a trend arrow; All lists everything. Click an entry to fly there and open it.
On a phone the panel is a bottom sheet; tap the arrow to open it.
Hover and click
Hovering an airport shows the 5-minute ASOS and latest METAR/SPECI side by side, what the TAF in effect says now (plus any
active TEMPO/PROB group), the verdict, trends, and the raw METAR.
Clicking opens the full card:
The verdict and the reasons for it, plus a Recent miss badge if there was a miss or bust in the last 3 hours, and how many of the last 6 hours had one.
Actual and TAF side by side: ceiling, visibility, sky, weather and wind. Values worse than forecast are orange. TEMPO/PROB alternatives are listed under the TAF.
Observations: the latest 5-minute ASOS and the raw METAR/SPECI with their ages.
Last 6 hours: a strip with the TAF in effect (top), what actually happened (middle; a thin colored line under the TAF
row marks TEMPO/PROB groups) and the ceiling trace (bottom; guide lines at 1,000 and 3,000 ft).
The raw TAF, and Open airport detail.
The card always fits on screen: if it would run off an edge the map pans to make room, and a tall card scrolls inside itself.
If a card says "Graded against the 1720Z TAF, in effect when this was observed", a newer TAF was issued after the latest observation. The verdict is fair to the forecast that was actually in effect.
Layers
Base map: Dark (default), Light, Imagery (satellite photo) or Terrain.
Radar: NWS MRMS base reflectivity, the national radar mosaic. Opacity slider underneath.
Satellite: GOES Infrared (cloud-top temperature; works day and night), Visible (daylight only; best for fog and low cloud) or Water vapor. Opacity slider underneath.
Miss heat (a heat map weighted by verdict), Trend arrows and Airport labels can each be turned off. Region outlines are chosen in the panel.
Your layer choices are remembered in this browser. The map's position is kept in the page address, so you can bookmark or share a view.
Replaying the last 24 hours
The bar along the bottom is a time scrubber. Drag it back to any 5-minute step in the last 24 hours, and the airports, verdicts,
clusters, radar and satellite all show that moment. ▶ plays forward in 30-minute steps; LIVE jumps back to now.
The bars above the slider show, hour by hour, how many airports were worse than their TAF: bust (pink), miss (orange) and lower
cloud or visibility (yellow); in the Wind tab, wind busts and misses. Its caption names the peak. Hover a bar for that hour's
numbers, and click it to jump the map there. It's the fastest way to spot when things went wrong.
Replay radar and satellite come from TAF Traker's own archive (the last 30 days, a frame every 5 minutes for radar and every 10 for satellite).
If there's no frame near the chosen time, that layer is hidden rather than showing the wrong time.
Airport pages
Open one from the map, the search box (/), or any table. The address is /airport/KXXX.
Header and time window
Actual now: current category, ceiling and visibility, labeled with its source, plus the latest raw METAR/SPECI.
TAF: what the TAF in effect says right now, with any active TEMPO/PROB groups.
Verdict: the current verdict, its reasons and any trends.
Window: 12 h, 24 h, 3 days or 7 days ending now. Set Ending to look at the past; the arrows step back and
forward half a window; Now returns to live.
Ahead (live only): how far the charts run into the future, 12 h, 24 h or 3 days, shaded and labeled FORECAST.
The TAF covers the first day or so; the latest GFS MOS run keeps going to about 72 hours.
Full screen: every chart has a ⛶ button in its top-right corner that blows it up to fill the screen
(charts redraw at the bigger size; the waterfall's rows and the cloud-layers chart benefit most). Press Esc or the button again to come back.
What's coming
A table of the hours ahead (live only): what the TAF in effect says for each step (category, ceiling, visibility), any TEMPO/PROB
group with a worse category, the TAF's wind, and the latest GFS MOS run's category and wind. Columns past the end of the TAF say
beyond TAF; MOS carries on. Where the two disagree, the TAF is the forecaster's call and MOS is the raw model guidance.
How the forecast evolved
Illustration. The 1720Z TAF kept VFR; actual conditions fell to IFR before the amendment at 1905Z caught up, and the pink cell in the bottom strip is the hour graded as a bust.
Each row is one forecast: GFS MOS runs at the top (purple labels), then every TAF version, oldest to newest. Amendments have amber labels.
Bright means that TAF was the one in effect at that time. Faded means it was valid but superseded by a newer one, or not yet in effect. So the bright blocks step down diagonally through the TAF rows.
Hatched lower half: a TEMPO or PROB group allowed for a worse category, in the hatch color.
▶ marks when that TAF was issued.
MOS rows show only the run-up, the stretch before the first TAF reached that time.
ACTUAL: what happened, from the 5-minute ASOS where available. METAR/SPECI: the hourly and special reports underneath; white ticks mark SPECIs, and a burst of ticks means conditions were changing fast.
vs TAF: the category verdict for each hour against the TAF in effect (worst observation in the hour; dark gray = as forecast).
vs TAF wind underneath is the wind verdict for the same hours.
In live mode the latest MOS run continues past NOW, so you can compare it with the current TAF for the hours ahead.
Hover anywhere for every forecast's value at that moment. Click to pick that hour for the chart below. Click a row label to highlight that TAF and open it in the history list.
How to read it: read down a column to see how the forecast for that time changed version by version.
Read across the bright blocks to see what pilots were actually being told. Compare both with the ACTUAL row.
How the forecast for one hour converged
Illustration. MOS said VFR, then MVFR, days out; early TAFs said VFR; a TAF 6 hours ahead had MVFR with a TEMPO for IFR; the last TAF before the hour finally said IFR, matching the actual.
Shows ceiling and visibility for one hour: the worst hour in the window by default, or whichever hour you click above.
The line above the charts sums it up: what actually happened, how far ahead the TAFs had it (or what the last TAF said if none did), and when MOS first showed it.
Left to right is how far ahead each forecast was issued, out to 72 h on the left, the hour itself on the right. The shaded stretch on the left is before the first TAF, when only MOS existed.
Diamonds are MOS runs, with a bar for the range of their category. They fade once TAFs took over.
Circles are TAF versions; the fill is the forecast category. A dashed line down to a hollow circle is a TEMPO/PROB group's value.
The white line, labelled ACTUAL, is what was observed during that hour; the pale band is its range.
Stems join each forecast to the actual: orange when the forecast was too optimistic (a better category than happened), teal when too cautious. No stem means the category was right.
Good forecasts converge on the white line early and stay there. Late catches only reach it in the last few hours.
Flip-flops zig-zag. Busts never reach it.
Ceiling and visibility charts
Bright white line, labelled ACTUAL: what was observed (the 5-minute ASOS where there is one). White rings: METARs. Amber rings: SPECIs. For ceiling, spells with no ceiling run along the "none" row at the top.
Thick cyan, labelled TAF: the TAF in effect. Cyan hatching down to a dashed cyan edge: the range a TEMPO/PROB group allowed. Actual inside the hatching was covered.
Dotted purple, labelled MOS: GFS MOS (the newest run before the TAFs took over, and the latest run past the TAF's end), with a faint band for its category range. Faint gray: every other TAF version.
Orange or red wash with a tick along the top: an hour graded a miss or bust.
NOW splits observed (left) from forecast (right), and shaded bands are the flight-category limits. Ceiling uses a log scale up to 12,000 ft; visibility is stretched at the low end, and P6SM is plotted at 7.
All the charts share one time axis and crosshair: hover one and the same moment lights up on all of them.
Every chart card has a How to read button with a drawn example and what to look for.
Wind
Observed sustained wind (pale line) and gusts (orange dots) against the TAF in effect (thick cyan), its forecast gusts (dotted cyan),
TEMPO/PROB winds (dashed lavender), older TAF versions (faint gray) and GFS MOS (dotted purple: the newest run out at each hour,
then the latest run ahead of now). Hours graded a wind miss or bust are shaded orange or red.
Underneath, three rows of wind barbs: Actual, TAF and MOS, about once an hour. The staff points to where the wind blows
from; each long feather is 10 kt, a short one 5 kt, a triangle 50 kt, and a circle is calm. G25 under a barb
is a gust. Actual barbs turn orange in wind-miss hours. Hover a barb for its exact direction and speed.
Cloud layers
Every reported cloud layer is drawn at its height. FEW and SCT layers are pale gray, more opaque as coverage increases.
BKN and OVC layers (ceilings) are colored by the flight category that height means, so a red bar is an IFR ceiling. Cyan outlines are the layers
the TAF in effect forecast, and dashed lavender outlines are TEMPO/PROB layers. Horizontal lines mark 500, 1,000 and 3,000 ft.
This is where "lower clouds than forecast" becomes obvious: solid bars sitting below the cyan outlines.
ASOS ceilometers only see up to 12,000 ft. Higher clouds are reported as clear, so the chart stops there.
Scorecard and accuracy by lead time
Scorecard: for the window, the share of observations in the same category as the TAF in effect, worse than anything forecast, and worse in any way, plus a bar with all six verdicts and the GFS MOS run-up's numbers.
Accuracy by lead time: the same breakdown for every version grouped by lead time: TAFs in 6-hour bands, MOS in 12-hour bands. The two numbers on the right are same-category % and worse-than-forecast %.
Forecast revisions
Each new TAF compared with the one before it, for the hours both covered. Each row shows the hours whose prevailing
category changed: ▼ Lowered or ▲ Raised, from what to what, how many hours, and how far ahead. Rows outlined in
orange are big downgrades (two or more categories, or down into IFR/LIFR). Click a row to show it in the charts.
A run of ▼ rows from successive TAFs means forecasters kept lowering the outlook. That's the "it's getting worse than we
thought" signal, and the time to treat the latest forecast with extra caution too.
TAF history and observations
TAF history, newest first. Tags show IN EFFECT, AMD and COR. The colored ribbon is the prevailing category across the TAF's whole validity. The score covers only observations after it was issued. Click to expand the raw TAF (one change group per line; TEMPO/PROB indented) and a decoded table.
Observations: every 5-minute ASOS, METAR and SPECI with category, ceiling, visibility, sky, weather, wind, temperature/dewpoint and the raw report. METAR/SPECI only hides the 5-minute rows.
Performance
Long-run statistics for every airport. Pick a period (24 h, 7, 30 or 90 days, all, or exact UTC dates) and a
minimum sample, so airports with only a few graded hours don't top the rankings by luck. Unless a chart says otherwise,
it uses the TAF in effect.
Each dot is an airport. The lines are the medians; orange dots are the ten most optimistic airports.
Spot on
Rarely worse and rarely better than forecast. The forecasts match reality.
Optimistic
Conditions are often worse than anything forecast. The quadrant to watch.
Cautious
Conditions are often better than forecast. Forecasters there hedge toward bad weather. Safer, but it can cost unnecessary alternates or fuel.
Erratic
Often wrong in both directions. The forecasts there are hard to rely on either way.
Everything else on the page
KPI tiles: airports graded and observations counted; same category; worse than anything forecast; busts; lower clouds/vis; better than forecast.
The second row is the same for wind: wind misses, wind busts, inside TEMPO, and hours with a wind miss.
Category / Wind switch on the map: color airports by category misses or wind misses.
Most wind misses and Wind above forecast, by lead time: wind's own leaderboard and lead-time curve.
Where TAFs miss: a map of each airport's worse-than-forecast rate (darker brown = rarely, bright orange = often; bigger = more data).
Most reliable in bad weather: the airports whose TAFs got the category right most often in the hours that mattered, when either the forecast or the weather was below VFR. Raw accuracy would just reward sunny airports, so this ranking ignores the easy VFR-forecast-VFR hours.
Most optimistic / Most cautious: the highest worse-than-forecast and better-than-forecast rates.
By lead time: same-category and worse-than-forecast rates against how far ahead the forecast was issued, TAF (cyan) vs MOS (purple).
Forecast vs. observed category: for every station-hour, what the TAF said (rows) against what happened (columns), as a share of each row. The diagonal (blue) is correct; below it (orange) is worse than forecast; above it (teal) is better.
Day by day: the national worse-than-forecast and same-category rates per day.
Time of day: worse-than-forecast and lower-cloud rates by UTC hour (last 31 days at most). For the continental US, 10–14Z is around dawn, when fog and low stratus peak.
All airports: sortable and filterable. Miss hours is the share of hours with at least one miss or bust; Non-VFR h and Non-VFR right are the bad-weather hours and how often the category was right in them; MOS same cat is the GFS MOS run-up's accuracy for comparison.
Clicking any airport opens its page over the same period (up to 7 days).
Scenarios
Worked examples of what you'll see and how to read it. Airports, times and values are illustrative.
Bust
1. Dawn fog the TAF didn't see
What you see: at 11Z a single pulsing pink ring at an inland airport. Hovering shows LIFR · VV002 · 1/4SM FG
against a TAF saying VFR · SCT220 · P6SM. The trend line reads "▼ Down from VFR to LIFR in 3 h".
On the airport page: the ACTUAL row turns magenta around 09Z; the bright TAF row stays green; the METAR/SPECI row has
a cluster of SPECI ticks as visibility dropped. The convergence chart for 10Z shows every forecast, MOS and TAF, on the green
VFR line far above the observed band.
How to read it: no forecast in the chain saw it coming. Radiation fog after a clear, calm night with a small
temperature–dewpoint spread is a classic way for TAFs to bust. Check the observations table: temperature and dewpoint converging overnight is the giveaway.
Watch for: whether an AMD followed, and how quickly. The revisions list and the AMD rows in the waterfall show that.
Lower clouds / vis
2. Same category, lower ceiling
What you see: a yellow ring on a green dot. The card says "Ceiling 4,000 ft vs forecast none". The TAF said SCT045; the airport reports BKN040.
How to read it: both are VFR, so on paper the forecast was right, but a ceiling now exists at 4,000 ft that wasn't forecast.
For a VFR flight over terrain, or one that needs to climb through, that matters. The cloud-layers chart shows the solid bar sitting under the cyan SCT outline.
Watch for: yellow rings that turn orange. A lowering ceiling that's still VFR is often the first step toward MVFR.
If several yellow rings sit near each other, check satellite for a deck spreading in.
Inside TEMPO
3. The TEMPO group did its job
What you see: a lavender ring on a red IFR dot. The TAF's prevailing forecast is MVFR, with TEMPO 2818/2822 2SM -SHRA BKN008.
How to read it: the shower brought IFR exactly as the TEMPO group allowed. The forecaster flagged it, so this isn't a miss.
In the waterfall that stretch has a red-hatched lower half on the bright TAF row; on the ceiling chart the actual line dips to the dashed cyan TEMPO level.
Watch for: the TEMPO window ending while conditions stay down. Once the window passes, the same IFR becomes a miss.
▼ Forecast lowered
4. Each new TAF is worse than the last
What you see: an orange ▼ on an airport that's VFR right now. Hovering says "21Z forecast lowered to IFR (was VFR in the 1120Z TAF)".
On the airport page: the revisions list shows ▼ rows from three successive TAFs, each pulling the low ceilings earlier or lower.
Reading down the waterfall at 21Z, the cells go green → blue → red.
How to read it: forecasters are steadily losing confidence in the better outlook. Nothing has gone wrong yet, which is exactly
the point: this flag warns before the weather arrives. Treat the latest TAF as the floor of what to expect, not the ceiling.
The inverse (▲ Forecast raised) means successive TAFs keep improving the outlook. It's usually good news, but check the actual trend too.
▼ Actual deteriorating
5. A SPECI beats the 5-minute data
What you see: the hover shows two observation rows: ASOS IFR · BKN005 · 19m and SPECI LIFR · BKN004 · 6m.
The dot is magenta, so the map is using the newer SPECI. The verdict is a bust against a TAF calling for MVFR.
How to read it: 5-minute ASOS data arrives about 15–25 minutes late, but a SPECI arrives within a minute or two. When they disagree,
the newer one is the better picture of right now. A burst of SPECIs (white ticks on the airport page) means conditions are changing quickly.
Watch for: "Down from MVFR to LIFR in 3 h" alongside a forecast that hasn't been amended. That combination means reality is moving faster than the forecast.
Region
6. A whole region below forecast
What you see: an orange outline over the Florida peninsula labeled "10 below forecast", with a mix of orange and yellow rings.
The histogram along the bottom shows misses climbing for the last few hours.
How to read it: when many airports miss together, the cause is usually one weather system that the forecasts mistimed, not
individual forecasters. Turn on Visible satellite (daytime) or Infrared and Radar, then press ▶ to replay the last few hours:
you'll often see a stratus deck or a shower band advancing faster than the TAFs assumed.
Watch for: airports at the edge of the region that are still green. Anything in the system's path is a candidate to be next.
Better
7. A cautious TAF
What you see: no ring, a green dot, and a card saying "Better than forecast (VFR vs IFR)". The TAF expected low cloud that never formed.
How to read it: not a safety issue, so there's no ring. If an airport does this constantly it lands in the Cautious quadrant
on the Performance page; coastal fields with a fog forecast that often doesn't verify are typical.
Miss → AMD
8. The amendment arrives after the miss
What you see: an orange ring and a card note: "Graded against the 1720Z TAF, in effect when this was observed". The TAF column already shows an amendment from 1905Z.
How to read it: the forecaster has caught up, but the latest observation was taken before the amendment, so it's graded fairly
against the forecast pilots had at the time. On the airport page you'll see the miss hours in the vs TAF strip end where the amber
AMD row turns bright. That gap is how long the forecast lagged reality.
MOS vs TAF
9. Did the forecaster beat the model?
Where: an airport page, the convergence chart and Accuracy by lead time.
How to read it: purple diamonds are where the model guidance stood days out; cyan circles are what forecasters published once TAFs
began. If the diamonds were already on the white line and the early TAF circles moved away from it, the forecasters overrode good guidance.
If MOS was off and the TAFs pulled toward the truth, the forecasters added value. Across many hours, the Performance page's lead-time charts show the same comparison nationally.
MOS forecasts category ranges, so a diamond's bar is a range rather than a precise value.
Replay
10. "What happened this morning?"
On the map, look at the histogram above the time bar and find the tallest stack of pink and orange.
Drag the slider there. The panel switches to REPLAY and shows that time; airports, rings and clusters change to match.
Turn on radar and satellite, then press ▶ to watch it unfold in 30-minute steps.
Click an airport to see its card as it was then, then open its page and set Ending to that time for the full picture.
Press LIVE to return to now.
Analysis
11. Which TAFs should I trust least?
On Performance, choose 30 days and a minimum sample of 7 days so small samples drop out.
Look at the Optimistic quadrant and the Most optimistic list. These airports' weather is most often worse than anything their TAFs allowed.
Cross-check Most reliable in bad weather: an airport can have a high miss rate but still be good when it counts, or the reverse.
Sort the table by Bust: two-category misses are rarer but matter more than single-category ones.
Open an airport from the table. Its page opens over the same period; the vs TAF strip and revisions show when and how it misses (dawn fog? afternoon storms?).
Small samples mislead. Two weeks of fair weather can make any airport look perfect, so prefer longer periods and the bad-weather ranking.
Wind miss
13. Gusts the TAF didn't mention
What you see: in the map's Wind view, a group of orange rings behind a cold front, outlined as "5 airports windier than forecast".
Hovering one shows Wind 18G32 kt vs TAF 12 kt. In the Flight category view the same airports may look fine: VFR as forecast.
How to read it: the front arrived with stronger mixing than forecast. On the airport page the wind chart shows orange gust dots
far above the cyan TAF line, and the vs TAF wind strip lights up for those hours. Check whether an amendment added gusts, and how late.
Why it's separate: wind problems rarely change the flight category, but they matter for crosswinds and turbulence near the
ground. Keeping them separate means a gusty VFR day doesn't hide in the category numbers, and a foggy calm day doesn't inflate the wind ones.
Quirks
12. Gray dots, faded dots and missing TAFs
Faded dot: no observation for over 75 minutes. Common at fields that only report METARs and have closed for the night.
Gray dot: no usable recent observation at all (station outage, or reports missing visibility and sky).
"No TAF in effect": many smaller and military airports issue TAFs only during operating hours. The airport keeps its current-conditions dot but gets no verdict overnight.
Military fields often look extreme in the rankings in both directions. Their observations are METAR-only and their TAF style differs; treat their stats with care.
How grading works
Every observation is compared with the TAF in effect at that minute: the prevailing forecast from the FM/BECMG chain, plus any TEMPO, PROB or BECMG transition active then.
Which observations: where an airport has 5-minute ASOS, that series is graded. At METAR-only airports, the reports are spread onto the same 5-minute grid (each slot takes the latest report, up to 75 minutes old), so a burst of SPECIs during bad weather doesn't count extra.
Categories follow the FAA limits above. Forecast ceilings above 12,000 ft count as "no ceiling", because ASOS can't see that high to check them.
Lower clouds / vis needs all of: same category as forecast, nothing already covered by a TEMPO/PROB group, and one of:
a ceiling at or below 5,000 ft that is at least 500 ft and 20% below the forecast ceiling, or where none was forecast;
a scattered-or-thicker layer at or below 5,000 ft at least 1,000 ft and 30% below the lowest forecast layer;
visibility of 5 SM or less, at least 1 SM or 40% below forecast.
Fair to the forecast: a TAF is never graded on observations taken before it was issued.
Every version is graded, not just the one in effect, which is what makes lead-time and convergence analysis possible. MOS runs are graded only for the run-up, before the first TAF covered that hour.
Hours and days: an hour's verdict is its worst observation. Daily totals feed the Performance page.
Trends:
Forecast lowered/raised compares the TAF in effect with the one in effect 6+ hours earlier, over the next 12 hours. It's flagged when any hour moves 2+ categories, or 1 category into or out of IFR/LIFR.
Actual deteriorating/improving compares now with 2–3 hours ago: a move of 2+ categories, 1 category into or out of IFR/LIFR, or a ceiling that more than halved to 3,000 ft or below.
Regions: at least 3 airports within 180 km, each lower clouds/vis, miss or bust, including at least one miss or bust. Wind regions use wind misses and busts only.
Wind is graded the same way, with its own verdict (see wind verdicts), against the TAF's wind
including TEMPO/PROB winds. MOS only forecasts sustained wind, so only sustained wind is graded against MOS.
Limits & data timing
Data
Source
Updated
Typical delay
TAFs
NWS Aviation Weather Center
checked every 10 min
minutes
5-minute ASOS
NOAA MADIS
every 5 min
15–25 min
METAR / SPECI
NWS Aviation Weather Center
every 5 min
1–2 min
GFS MOS
NOAA NCEP
4 runs a day
~4 h after run time
Radar
NWS MRMS mosaic
live tiles; archived every 5 min
a few min
Satellite
GOES via NOAA nowCOAST
live tiles; archived every 10 min
5–15 min
The map refreshes every 2 minutes; verdicts and trends are recalculated every 5 minutes.
ASOS ceilometers only see to 12,000 ft. Above that the site shows "no ceiling".
Replay radar and satellite cover the last 30 days. Archived frames are about 1.5 km per pixel, softer than the live layers when zoomed far in.
History before TAF Traker started recording was rebuilt from public archives, which sample TAFs hourly, so a short-lived amendment may be missing from early history.
Upstream outages show up as gaps. The Status page shows how fresh every feed is.
How far ahead of the forecast time a forecast was issued.
METAR
Routine (usually hourly) airport weather report.
MRMS
Multi-Radar Multi-Sensor: NOAA's national radar mosaic.
P6SM
Visibility more than 6 statute miles.
PROB30 / PROB40
30% / 40% probability of the listed conditions.
SPECI
Special weather report issued between METARs when conditions change significantly.
TAF
Terminal Aerodrome Forecast: the official forecast for an airport and the area within 5 SM of it.
TAF in effect
The latest TAF already issued and valid at a given moment.
TEMPO
Temporary fluctuations expected during a window.
VV
Vertical visibility into an obscuration such as fog; counts as a ceiling.
Z / UTC
Coordinated Universal Time. All times on this site are UTC unless marked local.
FAQ
Is this an official source? Can I use it to plan a flight?
No. TAF Traker is an experimental study of forecast accuracy. Use an official briefing for any flight. See the terms.
Why does an airport show a miss when the TAF had a TEMPO for it?
Most likely the TEMPO window had already ended, or the observed conditions were worse than even the TEMPO group allowed. The card's reasons say which, and a miss counted against a TEMPO says "(TEMPO/PROB IFR)".
Why didn't the dot change when I know the weather just changed?
5-minute ASOS data reaches the site 15–25 minutes late, and the map refreshes every 2 minutes. A SPECI usually arrives sooner and will update the dot once it's in.
Why are TEMPO hours not counted as misses?
Because the forecaster warned about them. "Inside TEMPO" is tracked separately, so you can still see how often TEMPO conditions actually happen.
Why does MOS only appear before the TAF period?
MOS is used to show the run-up: what the guidance said before any TAF covered that time. Once TAFs exist, the comparison that matters is between TAF versions.
Why do some airports have no verdict overnight?
Part-time TAFs. Many airports only publish TAFs while the tower or base is open. Their dot still shows current conditions.
Why is the radar missing when I replay?
Replay uses archived frames for the last 30 days; if there's no frame within a few minutes of the chosen time, the layer is hidden rather than showing the wrong time.
Why do high clouds never show as a ceiling?
ASOS ceilometers only measure up to 12,000 ft. Anything higher is reported as clear, so TAF Traker treats forecast ceilings above that as "no ceiling" when grading.
TAF Trakerbeta