January 8, 2024 Arctic Blast
After two blizzards, the bottom drops out
Too much going on over the next 10-days to fit into one post. So, updating on the intensity and progress of the forecast Arctic blast mainly peaking on Monday and Tuesday of next week across the central U.S. including Texas, and then into the Great Lakes and Ohio Valley. I’m also constantly updating on X/Twitter (@RyanMaue).
The conventional weather models including GFS initially weakened the extreme cold of the air mass with a cut-off low from the Pacific undercutting the polar front. However, that has largely faded. And, to top it off, the A.I. trained models are more extreme. I’m not entirely sure if these models have a cold bias, but both FourCastNet (Nvidia) and GraphCast (Google) are impressive!
I spent some time adjusting my color table to highlight the bright pink for temperatures centered on 40°F below normal. I’m quite happy with my artwork.
GraphCast 12z — next 10-days
GraphCast actual temperatures are a notch colder than ECMWF and GFS especially but not necessarily ICON or CMA-GFS (China’s global model). However, I’m suspecting sub-zero temperatures in North Texas are in the extreme cold tail of the distribution, but that’s important information regardless.
FourCastNet from Nvidia: Temperature Anomaly and Actual Temperatures on Monday, January 15, 2024 at 12z. Severe cold on par with GraphCast is quite alarming. However, I’m going to believe right now that these A.I. models have a cold bias. Of course, we don’t know anything about their performance because no one created/provided a reforecast database or a post-processed version. I’m working on that now, but the Copernicus (ERA5) server is down today.
Conventional NWP — ECMWF Ensembles temperature distribution for Dallas, Texas
Hopefully everyone knows how to read a box-and-whisker plot from your TI-83 graphing calculator days.
Dallas lows are mean/median around 20°F on Monday and Tuesday with the interquartile spread (25% to 75%) from 15°F to 22°F, so few doubts about a potentially 48-hour+ period of sub-freezing temperatures in Dallas.
Looking at the ECMWF HRES for Houston — also near 20°F for next Tues/Wed as the polar front extends well into the Gulf of Mexico, at least in this model.
ICON, ECMWF, CMA-GFS 7-day Forecast Temperature Anomaly
Similar patterns through 7-days which is not unusual. Most models should have the gist of the weather pattern correct. Each conventional NWP model has different land surface, radiation, and physics schemes that yield different interpretations of cold pool formation, maintenance and intensification over snow cover.
Over the next few days, we’ll have much better details about high/lows in individual cities. It’s looking rough.
Keep reading with a 7-day free trial
Subscribe to Weather Trader to keep reading this post and get 7 days of free access to the full post archives.