Graphical Forecasts for Aviation: How to Use the GFA
What the GFA shows, how its observation and forecast halves differ, how to read the time slider for a real leg, and where the text products still win.
The Graphical Forecasts for Aviation is the Aviation Weather Center’s interactive map, and it does two jobs on one screen. Scrub the time slider backward and it replays what the weather has actually done over the last 18 hours: station plots, radar, satellite, PIREPs, active advisories. Scrub it forward and it draws hourly model grids of ceiling and visibility, clouds, precipitation, thunderstorms, winds, temperature, turbulence and icing, out 18 hours and up to FL480.
Hold onto that structure. The GFA isn’t a document you read front to back; you query it, one hour and one altitude at a time. Do that and it answers a question no text bulletin could: what the sky will look like where you’ll be when you get there. Look only at the current hour and all you have is a pretty picture of right now. It belongs in the middle of a full preflight weather briefing, after the observations and before the decision.
What the GFA replaced
On October 10, 2017 the National Weather Service stopped producing the six Area Forecasts covering the contiguous United States: FAUS41 (Boston), FAUS42 (Miami), FAUS43 (Chicago), FAUS44 (Dallas), FAUS45 (Salt Lake) and FAUS46 (San Francisco). Service Change Notice 17-89 states that the FAA and NWS moved those bulletins, “currently used as flight planning and pilot weather briefing aids, to digital and graphical alternatives.” FAs for Alaska, Hawaii, the Caribbean and the Gulf of Mexico were unaffected at that time.
The old FA smeared a huge region across a long valid period in a block of contractions, so a pilot flying 200 NM got the same paragraph as one flying 600 NM the other way. The GFA answers by grid point and by hour instead. That is a real gain in resolution and a real increase in the amount of clicking.
Prog charts did not go away with the FA. They live under the Products menu on the same page, alongside TAFs, G-AIRMETs, SIGMETs, Center Weather Advisories and the regional forecast discussions.
The two halves of the page
Everything on the GFA falls on one side or the other of the present moment, and the two sides have different pedigrees. Confusing them is the fastest way to over-trust a grid.
| Observations & Warnings | Forecasts | |
|---|---|---|
| Time range | 18 hours back to now | Now to 18 hours ahead |
| Step | Hourly (15 minutes for the gridded flight-category analysis) | Hourly |
| Where it comes from | METARs, PIREPs, MRMS radar mosaic, satellite, forecaster-issued advisories | Model grids: NDFD, the FAA/NCAR Forecast Icing Product on hourly RAP data, Graphical Turbulence Guidance |
| Altitudes | Surface, plus PIREP and advisory levels | Surface to FL480 |
| What it answers | What the weather has been doing | What the models think it will do |
The observation side is the stronger half, because most of it is measurement rather than prediction. Station plots carry temperature, dewpoint, wind and gust, altimeter, present weather, ceiling and visibility around each airport symbol, and clicking one gives you the decoded and raw METAR. To work through an unfamiliar report field by field, our METAR decoder puts the plain-English reading first, and the glossary covers the contractions.
The radar layer is the MRMS mosaic, which AWC describes as combining over 140 NEXRAD sites into one image, looped over five frames ending at the time in the slider. Four products are selectable: Lowest Angle (the default), Composite, Composite 0–4 km, and Echo Top. That third one earns its keep. AWC says it “gives a better view of all precipitation for low altitude aviators,” without the overstatement a full composite inherits from cores at 25,000 ft. Why these products mislead in opposite directions is covered in our guide to reading radar like a pilot.
Advisories overlay both halves, filtered to the level you have selected, so a G-AIRMET Zulu polygon appears on the icing tab at the altitudes it covers rather than everywhere at once. What those polygons mean is a separate subject, handled in AIRMETs and SIGMETs explained.
The time slider is the product
The single most common way to waste the GFA is to open it, look at the current hour, and close it. Hour zero is the least useful frame on the page, because you already have METARs for right now.
Work the slider against your actual clock. A 1400Z departure on a three-hour leg means the arrival picture lives at 1700Z, so that is where the slider goes first. Then step 1500Z, 1600Z, 1700Z to watch the trend along the way, and one frame past your ETA, because 1800Z is when you’d be flying a diversion if the destination went down. Four clicks, and you have the shape of the day rather than a snapshot of the moment you happened to look.
One mechanic catches people out. Per the AWC help, all layers are synchronized to the slider, but any pop-up box already open must be refreshed to reload data for the adjusted time. A stale pop-up over a stepped-forward map is a quiet way to read the wrong hour. The valid time itself sits above the slider in UTC, with an offset showing how far from now you are.
The layers worth turning on
Icing. The severity grid comes from the Forecast Icing Product, built by the FAA and NCAR on hourly RAP data. AWC defines the severity bands by how long an airfoil would take to accrete a quarter inch of ice: trace is about an hour, light 15 minutes to an hour, moderate 5 to 15 minutes, and heavy under 5 minutes. Select MAX to find the worst severity anywhere in the column, then walk the altitude slider to see whether a usable level exists underneath it. The freezing level and supercooled large drop layers hide in the same dropdown and both belong in a winter brief.
Turbulence. The Graphical Turbulence Guidance grid is expressed in eddy dissipation rate, which is an atmospheric measurement rather than an aircraft-specific one, and AWC publishes the thresholds by aircraft weight class. For light aircraft under 15,500 lb, EDR×100 values of roughly 13, 16, 36 and 64 mark light, moderate, severe and extreme. For a medium jet the same four bands sit at 15, 20, 44 and 79. The gap matters: an EDR the airlines would file under light chop is already moderate turbulence in a normally aspirated single.
Low Altitude mode. Inherited from the old Helicopter Emergency Medical Services tool, this switches temperature, wind, turbulence and icing to 500 ft increments from the surface to 2,000 ft AGL and 1,000 ft increments up to 5,000 ft AGL. Forecasts run six hours ahead and observations two hours back in 15-minute steps. For anyone operating low, it is the most useful thing on the page.
The cross section. Enter a route in the Flight Path box, pick temperature, wind speed, turbulence or icing, set a top, and the GFA draws a vertical slice along the whole route rather than at a single flight level. This is the answer to “where does the ice start and does it end before the ridge,” and it takes about ten seconds.
Winds. Available at the surface, in 3,000 ft increments to FL300 and 6,000 ft increments from there to FL480, plus a MAX layer showing the strongest wind at any level. Faster than reading the coded text for picking a cruise altitude, though the winds aloft forecast is still where the temperatures that drive an icing plan are easiest to read.
Where the grids mislead
The forecast layers are model output dressed as certainty, and the AWC help is honest about the limits. Three of them change how you read the map.
The grids are not the forecast of record. The note under the clouds layer says it directly: the color-coded grid and grid circles are model-derived data, and users should check TAFs for forecast conditions. The same caution applies to the ceiling and visibility layer. A model grid painting MVFR over an airport that has a TAF is a hint; the TAF is the forecast.
The last two forecast hours may be a copy. As of March 2026, AWC states that the high-resolution model driving the CONUS icing and turbulence products only carries updated guidance through forecast hour 16, so hours 17 and 18 are identical to hour 16. If your planning window reaches the far edge of the slider, those frames are not telling you anything new.
Absence of a hazard on the map is not absence of the hazard. Weather outside the outlined GFA domain is masked out, which the help notes “does not indicate the absence of such phenomena in these areas.” Separately, AWC warns of visible seams where turbulence and icing solutions of different horizontal resolutions are pieced together to cover the whole domain, most often where the CONUS and Alaska domains meet the open Pacific, the Atlantic and the Gulf. A discontinuity at a domain edge is an artifact, not a front.
The observation side has its own version of this. The gridded flight-category analysis is an interpolation between reporting stations, and AWC says it “may not represent actual conditions occurring.” Over sparse terrain that is a smooth-looking guess between two airports 80 NM apart.
What the GFA doesn’t carry
There are no NOTAMs on the GFA, and no TFRs. Check the layer list and you’ll find METARs, PIREPs, TAFs, fronts, CWAs, NWS warnings, convective forecasts, SIGMETs and G-AIRMETs, and nothing at all from the NOTAM system. Since 14 CFR 91.103 requires the pilot in command to “become familiar with all available information concerning that flight,” a session on the GFA is a component of preflight action rather than the whole of it. You still have to go get the NOTAMs for the route. The FAA’s guidance on building a compliant self-brief is AC 91-92, Pilot’s Guide to a Preflight Briefing.
Note also that the official description lags the tool. AIM paragraph 7-1-4 still describes the GFA as covering the continental United States from 14 hours in the past to 15 hours in the future. The live page covers the US including Alaska and Hawaii, the Gulf and Caribbean, and parts of the Atlantic and Pacific, at 18 hours either side. Where the AIM and the help page disagree on a detail of the tool, the help page describes the software actually running.
The AIM does carry one item the web page buries. Two static graphics, the Aviation Surface Forecast and the Aviation Cloud Forecast, live under Decision Support Imagery at aviationweather.gov/graphics. They update every three hours and give snapshots at 3, 6, 9, 12, 15 and 18 hours. They load on hangar wifi that the full map will not.
A four-minute routine
The whole pass is quick if you run it in order. Open the observation half and scrub back six hours to see whether conditions are improving or decaying. Switch to forecasts, set the slider to your ETA, and step through ceiling and visibility, then clouds, then icing at your planned altitude with the freezing level on. Run the cross section along the route. Then close the map and go read the TAFs, the advisories and the NOTAMs, because the GFA has just told you which of them to read carefully.
Common questions
- How far ahead does the GFA forecast?
- Eighteen hours, in hourly steps, for the main forecast layers. Low Altitude mode runs six hours ahead in finer altitude slices, and the extended-range temperature and wind graphics reach four days. One caveat from the Aviation Weather Center: as of March 2026 the high-resolution model behind the CONUS icing and turbulence layers only updates through forecast hour 16, so hours 17 and 18 are copies of hour 16.
- Does the GFA count as a weather briefing?
- No. It carries no NOTAMs and no TFRs, and 14 CFR 91.103 requires the pilot in command to become familiar with all available information concerning that flight. A self-brief that satisfies the rule can be assembled without calling Flight Service, and AIM 7-1-5 says so outright while pointing pilots at AC 91-92. The GFA is one screen inside that process, not the process.
- Why does the GFA ceiling grid disagree with the TAF?
- Because they come from different places. The ceiling and visibility layer is a model-derived grid; a TAF is written by a human forecaster for a specific airport. The Aviation Weather Center says so directly in the GFA help: the color-coded grids are model-derived data, and users should check TAFs for forecast conditions. Where they conflict at an airport with a TAF, the TAF wins.