FINAL RESULTS ARE IN!!!
If this is the first time you’re seeing this page, I suggest you read up on how this contest was conducted. You can check that out down below the Final Results image in this article. Just scroll down.
In short – this contest went as expected. All 4 models, and utilizing two different snow output methods from those models, showed more snow than what we actually had to shovel.
The official snow measurement for the Scranton/Wilkes-Barre International Airport was 37.8″.
Weather models showed anywhere from 44.3″ to 74.4″ of snow.
As a brief reminder, this is what the 00z run of the models showed looking out 168 hours (7 days) into the future.
The 10:1 method off of the Canadian GEM proved to be the most accurate this winter when looking out 7 days into the future. It had the 44.3″ total
The Canadian GEM was also the closest to the real total using the Kuchera method as well. It had 53.8″ as the total using that method.
On the other end of the spectrum, using the Kuchera method off of the GFS or EURO was a complete disaster. Both of them showed over 70″ of snow for us this snow season.
I found it interesting that the German ICON model proved to be the second most accurate on the whole. This model doesn’t get nearly enough love. Now don’t get me wrong. It overshot the landing by a wide margin on several occasions, but at the end of the season, it looks second best on paper.
Also of interest is just how poorly the EURO model performed this winter. A certain TV station touted during one storm that “it’s usually the most accurate” as the basis for one of their forecasts this winter. Now in that particular case, that ended up being correct. However – I would point your attention to my findings this snow season as to why simply hanging your hat on the EURO is an incredibly poor idea.
In fact, the 10:1 method off the EURO showed more than 1-1/2 feet of snow than what actually fell. The Kuchera method showed about 3 feet of snow more than actually fell.
I hope this contest will help you think more critically when seeing snowfall forecasts in the future. Just because models will spit out snowfall maps several weeks into the future doesn’t mean that people should share them. Unfortunately – some people on social media use this to fear monger for likes and clicks that directly translate into dollars in their pocket. These same people are never held accountable and you can find them sharing more scary looking maps mere moments after the lastest hyped “storm” doesn’t pan out.
I also hope that this contest helps you think about the myriad of weather models that are out there. Why is the account you’re seeing only sharing the snowfall map with the biggest numbers? Or… the smallest numbers? There are also some people in Northeast PA that do that here on the interwebs as well as TV.
While weather will always be an inexact science, let “Weather Model Madness” be your guiding light away from the HYPE! Know that while my forecasts might not always be exactly right, I will always dive much deeper into what I think will happen far beyond just “what the models are saying”.
Final results table below.
Welcome to my weather snowfall contest I’ve dubbed “Weather Model Madness”.
If you’ve followed me for any length of time, you know how much I’ve railed against sharing raw weather model snowfall maps well into the future. Anyone with an internet connection can share weather model output showing snow maps 1 week, 2 weeks, or even LONGER into the future. As I’m sure you’ve seen, some of these maps are really EXTREME & usually come with phrases like “BIG SNOWSTORM COMING IN TWO WEEKS” or “NOT A FORECAST BUT WOW!!!! LOOK AT WHAT THE ______ MODEL IS SHOWING”.
Weather model snowfall maps far into the future are often very inaccurate. They will make a flurry look like a blizzard a week or more away. Of course, the opposite can also be true… but the scary maps are what gets shared for likes and clicks.
I’ve long said that we “shovel” a lot more modeled snow than actual snow. This snow season, I’m going to try and prove it.
Here’s how this will work.
Every Sunday – I will use the 00z model runs of the GFS, EURO, Canadian, and ICON weather models. 00z is currently the 7PM run time of those models. Once we “spring forward”, it becomes the 8PM model run time. This refers to when the models start crunching numbers, not when the data becomes available. That takes several hours to happen. The 00z model suite is what most morning meteorologists, like myself, look at first when analyzing model data early in the morning. Of course – I don’t start with model data when making a forecast. I start with current conditions 
.
I will look at a time period of 00z Sunday to the following 00z Sunday. This actually equates to 7/8PM Saturday to the following 7/8PM Saturday. The bottom line though is it’s a one week snowfall forecast total.
In the event there’s an issue with that particular model run… sometimes the super computers crash or there’s some other kind of outage, I will use the previous 12z model run for that model.
You’ll notice that the Canadian model map looks different. That’s because it comes from a different source. For whatever reason, the “State Level” zoom for the Canadian model doesn’t exist on Pivotal Weather. To try and keep the zoom levels consistent for you, I’m pulling the Canadian numbers from Weather Bell. The maps look different, but the snowfall numbers are the same.
I will be using both the 10:1 (raw) maps as well as the Kuchera maps.
10:1 (raw) maps use a strict 10″ of snow to 1″ of liquid equivalent for every single storm.
Kuchera maps attempt to create a more accurate snow to liquid ratio. This is done by taking the warmest temperature in the lowest ~18,000 feet of atmosphere and making a snow to liquid ratio based off of that.
Obviously, there’s a lot of areas on the map, but not every single area has a reliable snowfall measurement. For the purpose of this case study, I will be comparing the numbers specifically with what’s measured at KAVP (Wilkes-Barre/Scranton Airport). That measurement is reliably taken because KAVP is an official National Weather Service climate site.
Also – to be clear, when I share these maps, I will simply be sharing them objectively. I won’t be weighing in on what I think of them.
Hope you enjoy following along for a snow season of WEATHER MODEL MADNESS!