It turns out that there a person sitting there deciding what the weather is going to do. Which seems obvious, except that I had somehow never asked msyelf who “someone” was.

Generally speaking, I’m writing about things that have already happened, so when it comes to the relevant weather, I’ll reach for the Meteorological Aerodrome Report (METAR), which gives the observed weather at a weather station for a specific time. But if you are flying, you do want to know as much about the immediate future as possible, and as such, you’ll be relying a lot on the Aerodrome Forecast for your destination. The forecast used to be called a Terminal Aerodrome Forecast (or Terminal Area Forecast in the US, because they have to be different) and that acronym is still used: TAF.

A TAF generally offers a weather prediction over the next 24 to 30 hours, though there’s some variation. TAFs valid for 12 hours or more are issued or updated every six hours or when the forecast stops matching reality. In the UK, smaller aerodromes have 9-hour TAFs which are updated every three hours.

So a METAR offers the actual weather, while a TAF looks ahead for expected conditions. But the interesting difference between them is that a METAR reports measurements, while a TAF is written by humans, using human judgement about what will happen. This is why TAFs are not available at every METAR station.

Now that’s about everything I know, so I was thrilled to see that the Civil Aviation Authority of New Zealand published an article by TAF-writer Alwyn Bakker! That is not his real title, his real title is MetService aviation meteorologist. But the reason that I am excited is that he has written an article about how one writes a TAF and CAA New Zealand has graciously allowed me to reprint it!

I’m also interested to see Bakker mentioning the QNH in the TAF. It turns out that this is a New Zealand-specific requirement for domestic TAFs:

[Domestic TAFs] differ from International TAFs in that visibilities above 10km are included, as are cloud groups above 1500 ft. Winds at 2000 ft AGL and forecast QNH at the aerodrome are added at the end of the TAF.

The QNH forecast is for the maximum and minimum QNH values expected during the validity period of the TAF, or during a specified validity period. The forecast is provided as a guide to the range of pressure expected during a TAF or specified validity period. The main purpose is to provide a check on the actual QNH passed to an aircraft to ensure that any errors in transmission do not result in an incorrect altimeter setting. The QNH forecasts, appended to Domestic TAF, shall not be used for altimeter setting.

Writing an accurate, helpful TAF

by MetService aviation meteorologist Alwyn Bakker

Meteorologists start writing a TAF by looking at coarse model data (10 to 20km resolution) and satellite imagery. We’re working out where features such as fronts and lows are, and how they’re likely to move over the forecast period.

Next, we use finer resolution models (800m to 4km resolution) to investigate how those features may affect the airports we’re looking after.

Finally, we use a mix of local knowledge, experience, historical data, and rules of thumb to refine our thinking further, until we end up with a story we’re happy with.

Along the way, we compare notes with our fellow meteorologists. If conditions are conducive to thunderstorms, we’ll discuss timings and probabilities with the specialist thunderstorm forecaster. The severe weather forecaster might have warnings and/or watches out, which may affect our TAFs, especially if significant wind gusts are expected. Fog and low cloud questions go to the lead forecaster, but just as often they’ll want our input, so it’s best for us to come prepared with what we know.

When writing the TAF, it’s good practice for us to keep a ‘stock’ TAF in mind and consider deviations from that. For example, Queenstown Airport will typically start a summer day with a light north-easterly katabatic (cool air flowing down from nearby mountains). A westerly lake breeze will then develop around midday as the land warms up, then that lake breeze will die off in the evening and let the katabatic redevelop overnight.

If it’s going to be a particularly warm day, a heat low will form over southern Canterbury, resulting in a brisk southerly from late afternoon, potentially carrying through until well after dark.

If nothing else is expected to happen, then that’s my Queenstown TAF done. But if there’s a front moving over the region, I need to think about when it will show up, how long it will affect the airport for, and how much moisture it’s carrying.

For something moving quickly and not bringing too much moisture, I’d consider occasional showers (TEMPO 7000 SHRA). For something slower and particularly moist, it might be better to describe it as a band of rain (BECMG 6000 RA), with intervals of heavier rain and lower cloud bases (TEMPO 3000 +RA BKN008).

If there are thunderstorms expected, I’ll add those in with a 30 percent chance of them occasionally affecting the airport (PROB30 TEMPO 2000 TSRA FEW020CB). If I’m confident the airport will see those thunderstorms, I’ll upgrade that to a 40 percent chance (PROB40), or even a straight TEMPO if I’m really confident.

The QNH (atmospheric pressure, corrected to mean sea level) and the 2000ft winds are generated automatically from model data. I’ll tweak the 2000ft winds if needed, but I won’t touch the QNH, as I can’t add any value there.

Keeping it useful

To prevent TAFs from becoming unusable essays, we include only significant changes from the weather at the start of the TAF period. For example, we mention a change in wind direction if the change is at least 60° and the wind speed before or after is at least 10KT (so going from 330° 5KT to 180° 5KT doesn’t count).

For cloud bases, we worry only about cloud layers that are overcast or broken (covering at least five ‘eighths’ of the sky), and when the cloud base changes between 500ft, 1000ft, and 1500ft (so broken cloud with base lifting from 600ft to 900ft doesn’t meet the ‘significant’ criteria, but the base lifting to 1200ft does).

Visibility has more thresholds but follows similar rules. There are also rules about types of weather phenomena, gusts, timings, and probabilities. If you want to know more, Annex 3 to the Convention on International Civil Aviation has all the information, in exhaustive detail.

It’s worth noting that in New Zealand we write our domestic TAFs a little differently to those for our international airports, where complying with Annex 3 is a must.

In the ‘main line’ of a domestic TAF, information about non-significant weather might be spelled out more clearly (“30km visibility and clear sky”, rather than the Annex 3 term “CAVOK” (Ceiling and Visibility OK)). But from there on, the significant changes are the same, no matter how the airport is categorised.

Keeping it up-to-date

The final part of writing TAFs is monitoring and, if necessary, amending it. We have an in-house program that keeps track of incoming airport weather observations (METARs) and data from our automatic weather stations at airports. It flags anything that’s telling a different story to the TAFs, using those thresholds for significant changes.

Sometimes it’s a one-off fluctuation, and the TAF shouldn’t be amended as the fluctuation likely wouldn’t last long enough, such as shallow mist forming around dawn and dissipating within half an hour.

Other times, it gives us a heads-up of when reality is not lining up with the forecast. For example, that a feature is bringing heavier rain than the models were predicting, or a wind change is just not showing up on time, or some fog has rolled in from elsewhere. If so, we’ll look at the current conditions (using METARs, pilot reports, ATIS reports, webcams, even phoning ATC in a pinch) and work out what’s going to happen next, then issue an amendment.

It must be correct and it must be detailed

Meteorologists walk a tightrope between accuracy and precision. I can write you a perfectly accurate forecast that will be useless because it’s not precise enough (“It will rain somewhere in Aotearoa this year”). Or I could write a perfectly precise forecast that will be useless because it is so staggeringly unlikely to be right (“Lightning will strike the Sky Tower at 11.43am on Christmas Day 2026”), or because you’ll have to wade through pages of minute-to-minute information to find out whether you can safely land or not.

By sticking to what really matters, a TAF does its job – it tells the aviation system what to watch out for, and when. It gives pilots, planners, and controllers a shared picture of upcoming weather risk without getting lost in unnecessary detail.

That clarity helps everyone plan ahead, make timely calls when conditions change, and keep operations moving safely and smoothly.

More information

See CAA’s meteorology section.

GAP booklet: VFR MET (PDF 4.3 MB)

You can order for free the VFR MET Minima Card or the recently updated Weather Card from order publications.

I hope you found Bakker’s article as interesting as I did! I love the examples and I have a new respect for all those TAF-writers out there. And a shout-out to CAA New Zealand’s Vector magazine filled with interesting explanations and good advice.