Oh look! The GFS model is showing a big snowstorm in our region just in time for Thanksgiving!
But wait, there is more!
I am already seeing online posts and videos about a storm system on November 21st.
These are being posted for likes and shares. Click-bait is what some people call it.
Be careful about sharing long range snow forecast maps and videos.
If I were to post an hour 300 heavy snowfall map, then it would get hundreds if not thousands of shares and likes.
If I post a thread saying the hour 300 snowfall maps are crap, then I might get a dozen shares.
I don’t understand (and will never understand) the fascination with liking and sharing fear-mongering snow and severe weather posts. But, it is a thing!
You will see me discuss things in the long range, but with a heavy dose of wording that we aren’t making a forecast yet. We aren’t saying this is going to happen.
You will usually see me say “I am watching this storm system around this date”. And, leave it at that.
We do not trust models past a few days. Once you get to day six, seven, eight or more then they are almost worthless (not 100% worthless, but close).
You might as well throw darts.
We also use forecast ensembles. What are ensembles?
We have several long-range models that help us forecast the weather. Let’s talk about two of them.
The GFS and the EC models. You may have heard these names before.
The GFS is an American made model. The EC is a European made model. Both do quite well in the medium and long range.
Remember, we don’t forecast based on models. Models are tools that help us make a forecast. They are not gospel. All models are wrong. Some are more wrong than others.
Typically, a good forecast is made up of several models. Combining them with your forecast experience.
Okay, back to the GFS and EC models.
They run those models four times each day. Every six hours.
Each time they run those models, they come up with a slightly different solution. Sometimes, the two models completely disagree. Sometimes they agree.
At this time, both models are showing a storm system around November 20th to 22nd. Give or take.
One shows all rain. One shows heavy rain and snow. Could it snow? Yes. Is it likely? We have no way of knowing that this far out. It’s just not possible.
Nine times out of ten the hour 300 snow and rain maps are wrong. I would be wrong 90% of the time.
Now, let’s circle back around to ensembles.
The GFS and EC ensembles.
An ensemble weather forecast is a set of forecasts that present the range of future weather possibilities. Multiple simulations are run, each with a slight variation of its initial conditions and with slightly perturbed weather models.
They run the GFS 30 times. They run the EC more than 50 times. They do this four times a day.
Each time they run the model, they slightly change the beginning variables. The general idea is that the more times the ensembles agree, the higher the probability of the outcome. I will post some ensemble examples in the comment section below.
For example, if 30 of the 50 EC ensemble runs shows a snow event on the 21st of November, then confidence increases that we will have some type of snow event in or near our region on the 21st.
It doesn’t mean it will happen; it means confidence is higher that it might happen.
If 48 of the ensemble runs show no precipitation, then I am certainly not going to forecast snow on the 21st of November. Especially not ten days out! It would be irresponsible.
As of today, only two of eighty ensemble members are showing snow on the 21st of November. Yet, I am seeing people post snow maps and videos for the 21st of November calling for a winter storm in the Ohio Valley.
Again, could it happen? Anything is possible. Should you be forecasting it to happen and getting people worked up and excited? No.
This will go on all winter. We talk about this every winter. Many of you that follow my pages are aware of this. You have gotten quite good at messaging me and asking me about something you see on another page.
You have learned not to trust hour 300 snow maps! Hour 300 tornado maps. Insert hour 300 whatever maps.
I have no issue with discussing long range forecasts. As long as you understand the caveats of long range forecasting.
Confidence in a long range forecast is always going to be lower than a short range forecast.
Be careful about sharing long range snow forecast maps and videos.
I will post some more maps below. Showing you the ensembles.
Two out of thirty of these ensembles are showing significant accumulating snow. The majority are not forecasting snow.
None of these ensembles are showing a significant snow event.
None of these ensemble members are showing a significant snow event. Could it change? Yes. It could. It is too early to know.
And that concludes my soapbox message on long range snow forecasting!
