Area Forecast replacement – what does it really say?

The old Area Forecast (FA) is going away, and I for one say good riddance. This textual forecast product has plenty of valuable information, but its ALL CAP, coded format is a leftover from a previous era, making it a pain to use. Besides, almost everything in the FA can be found elsewhere these days.

Everything, that is, except cloud top forecasts.

That’s a problem, because cloud tops are an important factor for safe and comfortable flying. Especially in the winter, when ice is a constant concern, knowing if the tops are at 5,000 or 25,000 matters a lot. It’s the difference between flying and canceling for some piston pilots.

The FAA has touted the experimental Graphical Forecasts for Aviation tool as a potential replacement for the Area Forecast, but it’s very much a work in progress. On the good side, it is presented as an interactive, layered map instead of an endless page of text. That’s a major step forward – seeing how radar, satellite, observations and SIGMETs work together is much easier with a color image than with pages of text.

In my opinion, though, the forecast cloud top product is not very accurate for GA pilots, at least not yet. Here’s an example. After going to the GFA tool at, tap, Clouds at the top of the map, then Tops on the left side. I did this recently when a line of storms was moving through Indiana, and saw this.

FullSizeRender 16

The forecast map suggests relatively low tops – overcast ceilings about 2000 ft. and tops about 10,000 ft. – with just “CI ABV.” No problem in a turbocharged Cirrus or a TBM, right? Just blast through that layer and you’ll be on top.

Not quite. Look at the radar map in ForeFlight; it tells a very different story. And sure enough, tapping on one of those PIREP symbols reveals tops all that way to FL370. Not a great place to be flying a Cirrus (or anything really).

FullSizeRender 12

The cloud top forecast also doesn’t line up with the textual Area Forecast, which is still around (for now). On this day, it showed tops in Indiana as “layered to FL220.”

FullSizeRender 15

What’s going on here? For one, the cloud top product seems to be focused in particular on flights below 18,000 ft. But the forecast is also based on a numerical model (RAP), not the skeptical eye of a local forecaster. These models are good at a lot of things, but not estimating tops.

Here’s the Skew-t plot for a station in Indiana. It’s a geeky chart, but you can see the temperature and dewpoint lines converge at the 900mb level (about 2500 ft.) and they diverge again at 700mb (roughly 10,000 ft.). Those numbers match the data shown on the cloud top chart.

FullSizeRender 14

While such an observation is a helpful tool for understanding what’s happening in the atmosphere, it’s hardly a precise cloud top predictor.

At this point, it’s best to rely on other tools to determine where the tops really are. One obvious tool is PIREPs. While these aren’t always accurate for icing or turbulence (moderate to a 172 may be light for a Boeing), they are valuable for cloud bases and tops. Seek them out, whether from online weather sources or from ATC, and give them yourself when you’re flying.

Another good option is the color satellite image. This doesn’t show specific cloud altitudes, but it gives you an idea of the relative height – yellow and orange are higher.

FullSizeRender 17

The CIP/FIP icing forecast is another great tool, allowing you to see likely icing conditions at different altitudes and times in the future. On this same day, it showed icing at FL190, in clear disagreement with the GFA tool.

FullSizeRender 11

And of course the old advice to understand the big picture is probably best. A look at the surface analysis this day makes it clear that the rain and clouds in this area are the result of a fast-moving cold front, with a low pressure area centered on Lake Erie. Knowing that setup, it’s obvious that clouds around the front will be higher than 10,000 ft.

As always with weather, use all available forecast tools to add details, but don’t become a slave to one particular product.

Am I reading the forecast wrong? Anybody have another idea?


  • VADs (velocity altitude/azimuth diagrams(?)) are awesome for cloud tops, is a text graphic representing wind at various altitudes above a single point…

    It’s derived from Doppler radar returns which only register on clouds… No info at a given altitude, no clouds! Shows tops, bottoms, layers, etc.

    Unfortunately I’ve only ever found them at FBO briefing consoles with a commercial weather product subscription… Don’t know where to find them otherwise, they seem to be kinda obscure

    • Thanks, Scott that’s a helpful presentation! I learned to use VADs in Michigan, where the main question was how thick layers were in winter when flying an arrow around… they worked well for that, and it’s good to know how to weigh their info in other conditions!

  • John – The RAP model isn’t going to show you convective cloud tops (which is what these were) unless you lift a parcel of air. You can’t do that with the app you were using. The RAOB out of Wilmington valid at 00Z does show the potential of convective cloud tops of roughly 32,000 feet. At 00Z, Wilmington is to the east of the convection, but nonetheless does depict the tops pretty accurately.

    Keep in mind that the GFA will never really be able to predict the convective cloud tops accurately. It wasn’t designed to do that. It’s better to use other products like the one I provide in my Wx Brief Roadmap here: which shows the tops to be between 10 and 12 km or 33,000 to 39,000 feet. Here’s a forecast that’s also in my Roadmap that provides tops at 5,000 ft intervals – Clearly shows the tops are forecast to be between 30,000 and 35,000 feet.

    The CIP analysis you used is a tool that combines a RAP 3 hr forecast, METARs, satellite, NEXRAD, lightning and our pilot reports. So it’s going to do a better job than what you’ll get with the GFA.

    • Great explanation Scott. My main concern is that I’ve heard a lot of pilots blindly follow the GFA as a cloud top predictor. It’s just not that simple.

      • Yep, not that simple at all. When I do online training with my students I usually spend about an hour training pilots on how to find stratus cloud tops. Pretty easy. Then it takes about 6 hours (or more) to teach them about cumuliform tops.

  • So, let me understand-the textual Area Forecast that is/was used for cloud tops for the ga community is being replaced by the Glitzy GFA which cannot be relied upon for cloud top forecasts? I respect Scott’s expertise, but how does the average airman find this info that borders on the exotic? Will the FSS be able to tell me flight tops or will they they also be in the dark? Pretty confusing!

    • Ken – it’s still in the experimental stage right now. It’s automated so there’s no amendment criteria either. I’ve submitted many comments about the tool in hopes that it’ll drive some change. They’ve made a few changes recently, but they are a long way away from replacing the forecaster’s expertise. Just yesterday I was looking at the tool and it needs to react better to a changing environment. Here’s a depiction from that morning at 11Z over Charlotte. And here’s the ceiling depiction in ForeFlight at that same time. This clearly shows ceilings in Charlotte were low IFR. Radiational cooling was taking the ceilings down to this level which started around 09Z. It needs to be run hourly, but updated using the latest observations/trends on a 15 minute basis.

  • Scott:
    Does the Model Analysis and Guidance (MAG) system provide a way to ascertain cloud tops? There is a MAG Users Manual that shows you how to navigate through the program, but the data provided is of a scientific nature. I have not been able to find a document/manual that gives you guidelines for translating the data into more practical piloting terms, i.e. cloud tops, ceiling, icing levels, instability, etc. Do you know of any?

    • Patrick – No the NCEP MAG site doesn’t provide a tops analysis or forecast. It’s hard to find a depiction that has a good resolution. Most of the forecast model output is at a 5K ft vertical resolution (tops could be from surface to 5K, 5K to 10K or 10K to 15K, etc). And if there’s a broken or overcast cirrus deck up there, that’s what you’ll see on the forecast, not the real tops you are likely looking for that may be at 8,500 ft, for example. Here’s a link to the RAP model tops forecast ( which is the same model used in the GFA tool that John discussed in this article. I provide a tops forecast out to 18 hours in my Internet Wx Brief Roadmap for my subscribers. It’s a much better depiction than the RAP link I provided, but still only has a 5K vertical resolution.

  • A lot of experimental efforts flow through the Aviation Weather Center, and cooperative efforts between FAA and NOAA. A lot of the work never is seen by the public, or if not protected behind passwords, isn’t publicized. However, even those developments are incorporated into future products. Most of the experimental products do have a feedback tab: Use it and let them know what you think.

    Tops analysis from a modeler’s viewpoint is a bit tricky. Models tend to have a limited number of vertical levels through the atmosphere. Of course, the “required” levels are in there, and depending on the nature of the forecast, other, specialized levels might be included (think, low-level high vertical resolution for wind energy forecasting). As computing power becomes more readily available (models are computationally expensive, if done right), more vertical levels (better vertical resolution) might come into play, or greater horizontal resolution might be chosen instead. The point of this is, as the models start to see better vertical resolution, dependent on the cloud physics options incorporated into the models, we might see better estimates of cloud tops and thickness. I’ve already seen some models that do a credible job of cirrus and stratus depiction, but these aren’t ready for prime time yet.

    The Skew-T shown is one way to evaluate this, but care must be taken to not put too much faith in vertical profiles derived from model output. Although TAMDAR data (vertical profile data derived from commercial air traffic) do go into the models used for weather prediction today, we’re still feeling our way along which of the profile observations can augment the gold-standard radiosonde balloon launches that usually only occur at 00z and 12z, yep, twice a day, and from a relatively small number of sites across the CONUS. I tend to only trust 00z and 12z long-term models (NAM,GFS), and 00-01z and 12-13z RAP, and never beyond 6 hours when I’m briefing for a flight.

  • OH, That was easy! Cloud tops are the last thing I do after a weather brief. The current FA is simple and to the point.

Leave a Reply

Your email address will not be published. Required fields are marked *