Weather Social Media 101: How to decode those cryptic Tweets from weather geeks

So, you thought you'd stay informed on the crazy weather this winter by friending or following a meteorologist on Twitter.

And then comes the snow or storm forecasts, and Tweets become filled with cryptic terms and maps that look like some sort of clandestine secret agent communications, complete with funny looking acronyms and random numbers that don't seem to make sense. It's like trying to learn chemistry from an instructor that only speaks Pig Latin.

You won't earn 5 credits at your local college, but here's how you can translate some of those posts into English.

What does "Z" mean?

Many times notes will include something like the "0Z" or "12Z" model or any clock number and a Z. 'Z' is the time when the forecast model was started (or "initialized") and stands for 'Zulu'. That's a military/government designation for what you might better recognize as GMT/Greenwich Mean Time or UTC -- basically the standard for using time internationally. GMT is actually the old name -- the French bought the naming rights and it's now "Universal Time Coordinated" (UTC)). So Z = GMT = UTC. It's all the same.

Zulu time is 8 hours ahead of PST, and 7 hours ahead of PDT. Some forecast models come out four times a day -- at 0Z, 06Z, 12Z and 18Z; others just twice a day; some hourly. You can do the math for the Seattle times, but it's easier to remember this way:

  1. 06Z - late night model run (10-11pm)
  2. 12Z - early-morning model run (4-5am)
  3. 18Z - early afternoon model run (10-11am)
  4. 00Z - late evening model run (4-5pm)

In addition, the actual maps denoted in the model run will usually denote their forecasted time in Z. So if you look in the model run and the 24 hour forecast might say it's for "12Z on March 15" or "0Z on March 17". This is where it helps to memorize the most frequently used time forecasts: 12Z means that morning; 0Z means that evening.

BUT! 0Z charts take an extra brain step to process because "0" = midnight so it'll show the next day's date on the map, but you have to subtract the date back for Seattle. For example, if you see a forecast for "0Z March 17", that is actually a forecast for the evening of the *16th* here. Remember, midnight March 17 in France is 4 p.m. on March 16th for Seattle/Pacific Standard Time.


In addition to the time, you'll see lots of chatter about the models themselves. There are quite a few weather forecast models at meteorologists' disposal, with some more popular than others.

The WRF (Weather Research & Forecasting) model is one of the American forecast models names, which used to be the GFS (Global Forecasting System) because NOAA apparently can't stand keeping a model name the same for too long (just ask defunct "NGM" and "MRF" too.) There is also the NAM - North American Model. These are global models that run their computations across the globe but at a relatively coarse resolution.

Luckily, here in the Northwest, we have the awesome University of Washington Atmospheric Sciences Department, which runs their own model based on a localized version of the WRF -- basically it takes the mathematical equations from the WRF and applies it to higher resolution, but focuses just on the Northwest. Local meteorologists use that model extensively.

Also popular is the "ECMWF" -- also known as "Euro" for "European Model". It has been getting a lot of positive press over the past few years for having better skill than the American models because the Euro uses far more computing resources. Only caveat? It's a private consortium and the model data is not available for free to the public. You have pay for it. There are also the UKMET (British) and GEM (Canadian) models.

And there is a newer model called the "HRRR" -- "High Resolution Rapid Refresh" model. This model is proving very valuable because instead of updating every 6-12 hours like the global models, it updates every hour and provides an 18-hour forecast, so it's more for short-term forecasting and you'll see it frequently used the day before and day of an expected storm. Bonus: it runs at a very high resolution -- just 3 km, which means each forecast point is just 3 kilometers apart and it can better pick up local nuances in the terrain and topography.

Those models all spend their resources running one model run. But there are also a group of models called "ensemble" models, which take a big model run and then run it several times (usually at somewhat lower resolutions to save time) with minor tweaks in the data and math to account for gaps. More on those below.

What do the pretty colors mean?

I thought I'd showcase some of the more popularly-tweeted images that I and others use. It's just scratching the surface of what's out there for meteorologists to show and Tweet, but you're probably already getting early stages of carpal tunnel syndrome scrolling this far...

(NOTE: These are all archived maps from previous events -- not current forecasts!)

Let's start with the most popular: The snow maps!

UW model Snow Map: The upper left corner of the charts give a name of what time frame it's showing, and the upper right corner shows the time the model was created ("Init" -- initialized) and "Valid" is when the forecast is for.

In this case, this chart is for total snowfall over a 48 hour period as of the model that was generated on 0Z Sunday Feb. 5 (remember --that's 4 p.m. SATURDAY Feb. 4 for our time) valid as of 0Z Tuesday Feb. 17 (The UW model helpfully translates that into PST for us) . So this is showing predicted snowfall totals over 48 hours, ending at 4pm Monday Feb. 6., more specifically the snow predicted from 4 p.m. Saturday to 4 p.m. Monday. The legend on the right is predicted snow in inches.

This particular model run was predicting 5.9 to 8" inches over south Seattle (the purples) and about 4" in North Seattle ( the blues). It ended up a bit over forecasted there. It's not the first time that's happened...

Euro ("ECMWF") Snow Map: Same general idea -name of model in top left corner, initialized time just underneath, and valid time across the top. This was a 24-hour total snow forecast between Feb. 3 and Feb. 4. Again the legend is on the right in inches.

The versions I typically use are from WeatherBell, Inc., which is a weather model site that is a vendor of the European models. The little numbers on there correspond to specific forecasts at official NWS stations -- the white '4' in Seattle is Boeing Field; the black '4' is Sea-Tac. White vs black text color is just based on which color the map thinks will be easier to see based on the background color.

HRRR Snow Map: Basically quite similar to the Euro map. Just the HRRR is a short-term map. This map shows total accumulated snow over 17 hours. Note the Valid Time on these particular HRRR maps is also translated into in Eastern Time, in addition to "Z" time. So for ease you can just subtract 3 hours from that time stamp on the upper right. (The vendor I use for these HRRR maps is based on the East Coast. )

HRRR Radar Loop: The HRRR model also allows a feature of a "Predicted Radar". This is showing future predicted radar blobs in the state -- greens/yellows for rain; blues for snow. Valid time is in the upper right corner in EST again. The legend on the right is radar DBZ -- a measure of precipitation intensity.

UW Rainfall Chart: During heavy rain events, I'll sometimes use the predicted rainfall charts, which work in similar ways to the snow maps, only it predicts rain. The legend on the right is slightly tricky, it's in hundredths of an inch. So "2" means it's predicting 0.02 inches. "32" means 0.32 inches; "128" means 1.28 inches; and 512 = 5.12 inches. This particular map is predicting between 1.28 and 2.56 inches of rain in Seattle, but almost none near Port Townsend in the Olympic Rain Shadow. (It was pretty spot on!)

UW Wind Gusts: And during wind storms, these maps are helpful to predict forecasted wind speeds. There are versions that show "sustained" winds and then other versions that show predicted peak gusts. The legend on the right shows predicted peak gusts during the model time frame, and the UW maps use knots for units, which is 1.15 mph. So 50 kts = 58 mph.

Yes, I'm getting the salt out for the old wound -- this was a map generated two days before the wind storm that wasn't on Oct. 15. Indeed, it was predicting 60 knots (69mph!) in the city of Seattle.

HRRR Wind Gusts: Works the same way as the UW map, just a different color palette. This too uses knots. The Euro wind map, which looks very similar to the HRRR palette, will translate into mph so be sure to note on the map which unit it shows.

UW / Euro Sea Level Pressure Maps: These are useful in day-to-day forecasting but during storms they can give an idea of intensity .The black lines are isobars and the closer they are together near your area, the windier it will be. When you see a lot tightly wrapped around a low in your region, strong winds are possible.

Ensembles: Grasping this one might earn credits for 301 Weather Social Media. Ensembles are relatively new but of growing importance. Main forecast models have to make some assumptions in their calculations since we can't compute every square inch of the planet and atmosphere. To cover the gaps, ensembles take a main forecast model (The "deterministic") and then run it in several other iterations with some tweaks to the data to get multiple forecasts. If these forecasts come in pretty close together, you have pretty high confidence the model is on the right track. If they ensuing results are all over the place, it means there is greater uncertainty in the calculations.

In my Tweeted example below, there are two ensembles shown. The one on the left runs off the GFS; the one on the right is off the Euro model. On the left you see several different predictions of snow in Seattle during that early February snow event -- some as high as 8 inches; some as low as 1-2". The blue line is the deterministic; the black line is the average of all the models -- in this case about 3-4". Sea-Tac officially got 7 inches.

The Euro one runs 51(!) different runs off its main model and paints out predicted snow (or whatever event you want predicted; works for wind, temperature, rain,etc. too) as time increases from left to right (predicted time stamps in the middle row). The color palette is on the right -- this matches the same color palette on the Euro snow total maps I showed above. On the bottom is the total snow predicted (additive). Blue bars are the deterministic model; green is the average of the ensembles. If the two closely match, as they did here, it means high confidence. Sure enough, the Euro predicted 7 inches -- spot on!

These are just two. There are other ensembles out there as well.

So if you've made this far and haven't had to schedule that doctor appointment for that sore scrolling finger of yours, I hope this helps empower you to wade through meteorologists' social media posts. And next time you see something like this:

You'll know what we mean!

close video ad
Unmutetoggle ad audio on off