💊 • Wx Forecaster since ‘08 ❄️ • NCSU Alum 🐺 #GoPack • Diver 🤿
Raleigh, NC
Joined June 2009
- Tweets19.3K
- Following206
- Followers4.5K
- Likes5.9K
More: Digging into time periods and SSW events producing more frequent snow events:
Red: more snow events 1958-1999
Blue: more snow events 2000-2025
I did break it down by ENSO further and neutral caught my eye, beating both…and neutral has a pretty even split between the two eras as well. It could allude to a possibility that extremes in both ENSO states can overwhelm any notable SSWE benefit.
Here’s some era comparisons, close to an even split of SSWEs before and after 1999. A decent drop in snow day frequency change as a whole, but some areas like the apps have done well.
We arrive at the same answer when the model only learns from earlier winters. That doesn't prove SSW has no influence, but it isn't a reliable snow signal on its own.
Early-winter SSWs look snowier (+1.40% vs +0.37%), but mostly a consequence of low sample size (only 6 “early” SSW events)
The real test: does knowing an SSW happened help predict snow days? Not here. Error was 2.29 without it and 2.34 with it.
Every winter, SSW gets sold as a possible snow trigger. I checked all 41 since 1958 against 12,430 stations. First 6 weeks vs. matched winters without one: some regions got more snow days, others fewer. The overall average: +0.52%. 🧵
“Can't imagine why anyone would even look at the GFS on a daily basis”
