Saturday, November 9, 2013

An Open Letter to Twin Cities TV Forecasters

Dear Beloved Twin Cities Television Forecasters,

It’s snow time, and a winter-obsessed populace turns its blue eyes to you. If you’ll follow a few of our pointers, we’ll get through this season with a healthy appreciation for meteorology and a minimal amount of misunderstandings.

Ditch the Snow Meter
It’s certainly tempting to use all the tools that WSI and other weather providers give you to wow us. But remember, we’re generally lazy, literal simpletons who glob on to whatever we see on the screen (especially if we’re not listening closely). If we see you point to a graphic showing 7.5 inches of snow in Mankato, 11.6 in Maple Grove, and 5.1 in Rochester, we’ll take it to the bank.

Here’s the problem. The snow meter suggests a level of preciseness that the science can’t yet support. It’s hard enough to get snowfall forecasts right to within an inch or two; why make it more complicated by displaying forecast amounts to the tenth of an inch?

Go Barry ZeVan on Us
Sometimes, old school is the best way to go. There’s nothing wrong with hand-drawn general estimates of snowfall amounts presented in colored gradation bands (just proof them to make sure no bands are missing, something that happens surprisingly often). Yes, this may be more time consuming than simply slapping up a pre-programmed depiction, but it’s worth it. Knock yourself out by using fancy colors but keep the presentation simple.

Keep Model Data Behind the Green Screen
These days there’s no shortage of weather models, whether proprietary or public. And access to the information is equally unprecedented. But please keep the model data to yourself. And never, NEVER, display the one outlier model that shows gargantuan amounts designed to make us drool. We’ll take it the bank and that’s no good for anyone. Ultimately, model data is your domain and should be kept under control of those who know how to interpret it.

(Major Exception: Weather Bloggers)
It makes more sense for weather bloggers to show model data, though even they might want to be careful as well. Just as true foodies might like to hide out in the chef’s kitchen, weather enthusiasts might enjoy a little inside baseball when it comes to weather. Weather bloggers also have the luxury you don’t: unlimited time. They can educate us to their heart’s delight and update essentially continuously.

Use Snowfall Probabilities to Communicate Uncertainty
There’s no such thing as a sure bet in snowfall forecasts, so why not communicate uncertainty? Tell us there’s a 20% chance of 4+ inches of snow, a 60” chance of 2-4 inches, and a 20% chance of less than 2 inches. This may threaten your inner macho, but it’s all about communication. When it comes down to it, it’s your communication abilities that we judge, almost as much or more than your actual knowledge of meteorology. The Capital Weather Gang, a weather blog produced by the Washington Post, sets the standard for communicating uncertainty.

Stand Up to the News Director Who Wants to Play Up A Storm
OK, this may tread on job security, but if you’re honest with yourself you know you’ll be doing the right thing. We know that snow sells almost as much as sex. You’ve probably been asked at some point to “play up the snow possibility” either explicitly or subtly. When the anchor leads with, “Hey Eddie, big snow coming our way, eh?” nip the hype in the bud with, “Ginger, this one will barely cover the grass!” Viewers will appreciate the straightforward approach.

Post-Storm Evaluation
Here’s where you can earn gobs of credibility. If the storm didn’t pan out as forecast, as often happens, come out and say so directly. A simple phrase like, “It was less than we expected and here’s why,” is all that’s needed. Better yet, if you communicated storm uncertainty with pre-storm probabilities, you can say, “the concern we had that thunderstorms in Iowa would steal the system’s energy materialized.”

By following these tips, you can become a truly trusted resource in this mother of all Minnesota seasons.

P.S. – Go easy on the non-weather tweets or consider setting up a separate Twitter account for non-weather tweets. The more extraneous stuff you throw in, the harder it is for us to sift through the important stuff.

If you're a meteorologist or consumer, what are your thoughts? Agree? Disagree?


  1. Very nice Bill.....lets hope they listen....I agree with ditching the snow meter, especially when the snow meter doesnt even come close to what their actually saying. Well said about owning the forecast after the event. It hasnt been a stellar start for the NWS so far in regards to snow, but I understand forecasting snow is not easy business but in my retail business we rely on accurate forecasts so that the public is prepared in worst case scenarios.

  2. So how long do we need to endure this boring weather. Is there any news worthy snows in sight?

  3. An excellent post. I agree with every point. There is so much skepticism out there regarding snow forecasts, particularly with a certain good neighbor, that most people I talk to have very little confidence the predictions.

  4. What a timely post, the best practice's that you describe may come into play around the 19th of November if the Euro and GGEM are correct. This would be about 45 days after the historic snowfall in the Dakotas, and may be evidence of Lezak Recurring Cycle. This describes it...

    Its way to early to talk about snowfall amounts up here, we may miss it entirely, or this system could be a figment of the models imagination. But my best guess is that this could be real for the Upper Mississippi Valley to the Ohio Valley. Especially if this high pressure system that will bring our cold air in for Monday and Tuesday will settle over the SEUS

  5. Excellent comments! Agree on each point made.

    The snow meters are the derived model outputs and those decimal settings ARE controllable by the meteorologist using them. The software has settings you can control -- units, decimal points, range, etc. The perception of precision is the thought behind it and is pretty silly with the meteorologists' all knowing this, but it "looks good" to the director/GM at the station.

    Showing models on-air is a recent trend due to software improvements of tv graphics systems. Several reasons used: easy to create as no hand drawing, allows multiple graphics, and that 'precision' mode where the kinks and imperfections in a contour is believed. It is a crutch and can become lazy tv meteorology especially at a local market if no time/short-staffed to draw up the maps.

    Probabilities is a trend trying to be introduced by the NWS, but not terribly popular by TV side for variety of reasons. Hopefully, it will start to catch on and done in a way viewers can easily understand. Stay tuned.

    Post-mortuem analysis -- simply awful by TV meteorology for variety of reasons. If a blown forecast that is not something you can hype or promote for the station. Also, it takes time and effort to go through the details of the data and most TV meteorologists don't have the time, and a few (not all) don't really care as they are onto the next storm.

    News directors, producers, marketing depts, etc., for local and national stations have definitely impacted weather coverage. Obviously, the info is out there but at times mightily overdone ad nauseam. Like anything in today's society, events of any news type are overdone or overreacted in response. Just how it is ....

    1. Thanks for the additional elaboration. It gives great context!

    2. LOL great post I named one forecaster in the metro area "The Boy Who Cried Wolf" always playing up the treat of snow. I do however make my living plowing snow and consume every bit of forecasting I can get.
      ... I heard we are in for a big storm next Monday :-)

    3. @Bob, I've met a number of local TV forecasters and I don't believe they purposely play up snow, though I do believe there's often pressure on them to do so.

  6. And never, ever, name a winter storm.

  7. I wonder, too, about the comparisons between past "major" storms (like the Halloween Blizzard or something similar). Is that really helpful? For most of us, all we hear is that it's going to happen again, rather than the broader comparisons of weather patterns. So, I'd suggest doing away with the, "Looks eerily similar to the Halloween Blizzard of '91" when in fact only 2 inches of snow is predicted. That's a confusing message to me.

  8. Well done Bill.

    Nice comments everyone.

    I've got my eye on next Sunday & Monday. I love that large trough that will be pulling into the Plains. The temperature contrast that will be available for this next storm will be impressive. Fun to watch.

  9. #1, you're right. as much as meteorologists like myself want to be that precise, it doesn't happen that easily. besides, I know many a winter-precip client in the past that wishes we could have done that.

    #2, that would mean the forecaster would have to think about their amounts, and occasionally show they can blend the models in their head or on paper. some can do that well, some can't.

    #3, unfortunately, the weather channel brought this upon us several years ago, and forgot to explain what actually goes into making the models as well as why there are several of them out there. I know personally it can take a bit of time to do, but that would be a good subject idea for local mets to blog about when the weather is calm. whether they actually do that is another matter.

    3a: yea, I know I can tend to get a bit weenie-ish at times on my own personal blog. but I try to at least explain things a bit and show all the possibilities when relevant, so that the reader can tell why i'm leaning a certain way. some mets are better than others in that respect.

    4: probabilistic forecasting is where I know a lot of mets would like to go, myself included, but the public has a hard enough time comprehending probability of precipitation, much less probabilistic snowfalls. that being said though, I have seen a temperature product and a presentation of that product that I think could work done by a company called EarthRisk with their "Apollo 101" graphics. The way they present their temperature forecasting with the error bars and discussions is actually quite good. If some station went along with that format for the snowfall forecasts and such, I think we could get the probabilistic forecast mainstreamed, and that station would look better for it. But it would take some doing to make that happen (like convincing news directors and a blog entry/video instruction on how it works).

    5: the hype, well if it wasn't for the ratings game that stations play, especially several back east, we wouldn't have this issue (look at the nor-easter that was supposed to happen later this week, only to become just a snow-squall event for the great lakes). but you are right there. just have to give the qualifiers.

    6: post-storm analysis, there is enough of a hard-time sometimes for mets to look over their own work on a day to day basis, because this can take time, and places don't necessarily pay for you to be sure you're hitting the forecast as they should. trust me, i know this from the inside.
    the other part here is plain ego, whether we believe that or not. because existentially, when we forecast, we are trying to read the mind of god and nature (albeit unsucessfully at times). so when we're on a good streak and all of sudden something goes bust, we don't like to admit we screwed up, especially when you have news directors on their backs saying your wrong forecast caused X.x drop in the ratings. but you are right in that, it's better to admit you're wrong and be straight up, than act like a politician and spin things like crazy. i have been on both sides of the street on the radio, and let's say I honestly think the public is smarter than most people think, if you explain things properly the first time.

    but that's just my side of things. I am sure others can and will differ.

  10. Bill, in a past post (maybe a couple years back?) I posted about my frustration with mets who use the word "normal" instead of "average." I won't repost the whole thing here - can't remember all I said anyway. In general, though, the words carry with them two very different meanings and usually a lot more hype is drawn from straying from what's "normal" rather than what strays from "average" (which is actually very normal). So, saying, "We're 2.3 degrees above NORMAL for these past 3 weeks of weather!!!" (Stop the madness!) is ridiculously sensationalistic and leads the average weather viewer to think things are way out of whack weather-wise. If we use the word average, the same stat isn't actually that big of deal at all because averages are the averaging out of extremes. So it's actually "normal" to be all over the map, so to speak. But of course, the latter doesn't make for very exciting weather news. :)

  11. Great post Bill. Lots of great points made in this article. I've never understood the whole playing up of a storm when it in reality won't be as big of a deal. If anything, it would hurt the ratings of a station rather than help them because people won't find their forecasts reliable and will go somewhere else to get something more accurate. However, weather is something that impacts every individual out there. We all commute in it, some work in it, so it is something that everybody cares about. Any chances at a disruptive snow of any way, shape, or form will likely lead a newscast because it's information people want. People want it to be accurate though. The more accurate you are, the more likely people will be coming back to watch you later. As "fun" as it is sometimes to see the ridiculous amounts of snow that a storm could put out according to one of the many models out there it is pretty ridiculous showing them on air unless you think those totals will come to fruition. As for next week, as much as snow lovers are hoping for a Euro type solution to pan out, there is pretty much zero model agreement. If anything, the Canadian model is closer to the GFS and the Euro is on its own. Euro didn't do too hot with the storm that was supposed to form off the east coast where the GFS had a better handle on it. Euro was fairly consistent with that storm like it is with this one. However the GFS was pretty consistent was its east coast solution, where now it's kind of all over the place with its solutions. Too many unknowns in the long term to really side with one model or another. If the GFS starts creeping towards a more Euro type solution and the Euro holds its ground then we could be in for quite the storm next week. It's also possible that we will see nothing but flurries as well. I posted this in the previous blog post, but will post it here too. Several of you may be on Twitter, so if you want to follow along my Twitter handle is @dewppler. I mostly tweet weather and sports stuff, with the occasional other randomness. Feel free to follow! The name is Storm Chaser.

  12. And while we're piling on, no comedy routines, please. Yes, definitely be personable and engaging. It helps us absorb the forecast. But don't go overboard. The goal isn't a YouTube blooper reel here.

    1. unless they're looking to sell the blooper reel for 99 cents a clip, 9.95 for the complete set.

  13. Very nicely put. Views | clicks | mentions play far too large are roll these days, forecasters forget what they're meant to do, and who they are meant to serve. Putting an emphasis on UNCERTAINTY should be better appreciated, something we strive to communicate on my Denver based blog Weather5280 (

  14. Look at this MET in Mankato put out on twitter today...
    The person this post is all about. full disclosure im sure but two weeks out? no certainty even on next weeks storm but yet he needs to puts this out why? This is the same guy who last year put out a tweet that school will be cancelled the next day in the Mankato area because of snow, and school was not cancelled. guess some will never learn. Great post Bill!