GRADING THE WINTER STORM FORECAST

Model Snowfall Graph

If you have watched me on TV or followed me on social media/this blog for a while, you know that I don’t hide from busted forecasts. I admit when I am wrong. I also do proudly proclaim that our forecasts are the most accurate you can find from any source for our area and I have the statistics to back that up. Accuracy is very important to us. I track it obsessively, recording our (and 4 other sources) forecasts each day and having a complicated spreadsheet tabulate the errors for each time period.

So in that spirit, how do I think we did with this storm?

Pretty well, but with plenty of room for improvement.

1) INITIAL COMMUNICATION OF THREAT: A

We started taking about the possibility of a storm during the middle of last week. By Thursday and especially Friday, we started talking about the potential for an “impactful” storm, enough to shovel, plow and cause travel disruptions. It was too early to give specifics on timing and snow accumulations but I think we did well in giving the audience a heads up that something significant was coming.

I posted this on Friday:

confidence

2) THREAT COMMUNICATION OVER WEEKEND: A

By the weekend, our confidence in some things increased. We were pretty sure of the timing. We were somewhat confident that the area would experience snow and a mix of snow, ice and rain. But the snow forecast remained very tricky and we told you that. We told you to check the forecast often!

Winter Storm Confidence Grid

3) IMPACTS: A

I think we shined here. On Monday we highlighted the threat for severe impacts on travel. We talked about the threat for power outages due to the heavy, wet nature of the snow. I did not see many other sources talking about this. There were thousands without power in Columbiana County earlier today. We advertised that a large amount of school adjustments would happen, impacting thousands of families.

Weather ImpactsWinter Storm Impacts

My only knock on this is that the ice threat was overplayed to a degree. There was some freezing rain southeast of Youngstown but not enough to have big impacts.

4) SNOW AMOUNTS: B-/C+

This was always going to be a very difficult forecast. Would mixed precipitation play a big factor? Would the deepest moisture reside over the Valley? These are questions that were nearly impossible to answer with a decent degree of confidence until the day of the storm.

That said, there were signs that this would be very BIG. The European model, generally considered to be superior to it’s US counterpart (the GFS) locked onto the right idea over the weekend. Other models were all over the place (Quick aside: The European is an already excellent model and is being upgraded in March, which should make it THAT much better than the GFS). I regret not respecting the Euro enough to take it’s high snow totals more seriously.

Our initial map, issued Sunday afternoon: 

1st snow

A decent first effort but obviously underdone for 75% of the area.

The next major update to the map came at midday Monday. 

updatedsnow.jpg

Good step, just not far enough. Played down the mixed precip, introduced an 8-12″ area and moved the 4-8″ area farther east a little.

By mid-afternoon Monday, the storm was really “showing its hand” and the hi-resolution, short-range modeling was locking on to the fact that the mild air was not going to much of a factor and the Valley would reside in the “bullseye” of highest moisture content.

The final map, issued a couple of hours before the snow started: 

finalsnow

This was pretty good. The 8-12″ area could have come a little farther south and east but not by much.

OVERALL GRADE: A-/B+

I preached that people should focus on the IMPACTS rather than if their yard would get 7″, 9″, 12″ or whatever. The impacts were high and we communicated that risk well ahead of time. The snow forecast ended up being very good at the end but I regret not pulling the trigger on that final map 12 hours earlier. Some of the data showed it but not enough of it did. But, when the European is “locked in:” and not waffling back and forth, it needs to be a big red flag…even if the other data contradicts it. That’s the main lesson here from a purely forecasting standpoint.

Thanks for reading!

Eric

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s