Tomer Burg

Now that more observations have arrived, here's my updated hand-drawn contour analysis for snow accumulation from this week's blizzard:

The failure was the placement. The cyclone was slightly east & more positively tilted than CAMs modeled and I expected, shifting the deformation band east into E NJ & NYC. Since a sharp gradient was expected west of this band, the east shift meant much lower totals than expected near the gradient.

At a glance, the success was the general structure of snow bands - SSW-NNE deformation band from DE-NJ, weak subsidence axis in between that & heavier banding to its east, secondary intense band in SE New England.

Keeping in mind the map on the right is only a preliminary verification analysis (made while the storm was ongoing in parts of New England), I wanted to start with a quick verification of my forecast map - I'll write up a more detailed review when I have some time later this week.

Thanks Sam! Definitely pleased with how I resolved the individual features & bands in the forecast, but could've done better with the placement as I weighted too heavily on the CAMs when global models (and the RGEM) were correctly farther east with the cyclone.

Using all public storm reports from the NWS so far, I contoured a preliminary estimated snow accumulation map for today's blizzard (through 6-8pm EST tonight).

Truly a historic storm - both in terms of impacts, and the forecast challenges leading up to it.

With approximately 13” recorded in 5 hours, Providence, RI got about as much snow in 5 hours as Denver received all winter to date

4⃣ Tomorrow -- As the occluded cyclone moves off to the northeast, forcing for ascent becomes increasingly disjointed from the cyclone center. Strong low-mid level warm air advection pushing through RI & E MA leads to snow rates over 2-4"/hour, but without the pivoting snow bands expected out west.

...producing heavy snow bands in excess of 2-4"/hour moving from SE to NW through the NYC metro and merging into the dominant pivoting snow band, now likely over the western half of NJ. Thundersnow isn't out of the question with these convective bands.

3⃣ Late tonight -- As the cyclone & its associated low-level jet (LLJ) rapidly strengthen, the LLJ rotates towards the southeast, with strong offshore convection along the triple point of the cyclone rotating into the cold conveyor belt in an area of strong isentropic ascent...

2⃣ Tonight - As the surface low & upper-level trough become aligned & sfc cyclone rapidly deepens, the frontogenesis swath pivots to the SSW-NNE, with a slow-moving to near stationary snow band setting up from DE to S NJ & Philly. Heavy snow rates upwards of 2-3"/hour are expected with this band.

Detailed snow banding discussion:

1⃣ Short term -- An ongoing zonally oriented frontogenetic band associated with the mid-level warm front is slowly pushing north into DE & southern NJ, producing large aggregates and heavy wet snow. This band will continue slowly moving north & intensifying today.

Given these factors & assuming high-resolution models are better resolving the convective aspects of the cyclone development and its rapid deepening, I sided with the CAM suite which has the max snow swath extending west of the global model suite.

HRRR 10:1 ratio map was used as a starting point:

As of this afternoon, nearly all high-resolution models have converged on roughly the same cyclone location, amplitude, and frontogenesis evolution, with the NAM & HRRR having been far too amplified and northwest yesterday and correcting southeast since then.

❄️ FINAL FORECAST: BLIZZARD OF 2026 ❄️

Now that the snowstorm is underway, I went into a high level of detail trying to narrow down the snow band placement and evolution.

Forecast reasoning follows in the thread below:

NWS New York makes an excellent point in their discussion. Snow rates will start light tomorrow morning & above-freezing temps mean snow will accumulate little if at all for many.

Do not be fooled - this doesn't mean the forecast busted.

Conditions will steadily deteriorate into the evening.

NWS New York makes an excellent point in their discussion. Snow rates will start light tomorrow morning & above-freezing temps mean snow will accumulate little if at all for many.

Do not be fooled - this doesn't mean the forecast busted.

Conditions will steadily deteriorate into the evening.

Something interesting of note for modeling applications & later evaluation efforts - the NSSL MPAS 2-moment (left) vs. 3-moment (right) MP schemes actually have a non-negligible difference in how far west the snow band expands:

Keeping in mind the NAM is likely too amplified/too far NW, the main driver of peak snow totals will be a pivoting axis of strong mid-level frontogenesis, with a secondary enhancement from a rapidly strengthening low-level jet overnight into NJ, NYC metro & Long Island.

The large-scale driver is a rapidly deepening cyclone forced by a deep trough rapidly becoming negatively tilted and phasing with a weaker lead shortwave trough. Post-occlusion, the strongest forcing for ascent becomes disjointed from the cyclone and moves off to the northeast.

Given the latest data & analysis, this is my current forecast snow map for the upcoming major East Coast snowstorm. Major uncertainties remaining include how far east/west the sharp gradient over PA/DE/NJ will be, and where a >24" corridor of snow totals will be.

Even if an LLM has good knowledge of your preferences, it still doesn’t fully replace personal autonomy over decision making unless you actively hand over your decision making to it

I just had an interesting realization

I’m traveling this weekend and asked ChatGPT for tips on what to visit and what cafes & restaurants to try — it gave me a whole list and backup options — and I disregarded all of them and found even better options on my own

You might've heard of the butterfly effect - but did you know the person who coined it (Ed Lorenz) was a meteorologist?

Check out how a very slight shift in the initial position of a tropopause polar vortex (TPV) over Quebec rapidly grows over time:

It's thanks to the NCAR supercomputer that I was able to finish my PhD dissertation - without it the task would've been much harder and taken much longer. I'm sure many others in the atmospheric science community are in the same boat.

All I'll say is what the analysis showed - while latest data indicates a major snowstorm is unlikely, it is not impossible, and changes that could lead to one occurring are within historical margins of error at this lead time.

If you asked me to put in my 2 cents... in the wise words a professor once told me, "don't make a forecast you don't have to make."

I get off easy here since I'm not an operational meteorologist, but it's no doubt a communication challenge to people that do this for a living.

Now looking at backwards ESA, this map shows the sensitivity has little to do with the southern trough (#1) - but almost everything to do with trough #2 (right). The deeper the trough is, the more likely a rapidly deepening low is, and vice versa.

So what should we be keeping an eye out for?

Bringing back my favorite ensemble sensitivity analysis (ESA) tool - this shows the primary mode of variability in the EPS ensemble, meaning either we get a rapidly deepening low off the Mid Atlantic coast, or we don't.

Compare the 6z vs. 12z runs of the AIFS - the result of my last post above is a much more potent upper-level trough over the Mid Atlantic with stronger forcing for ascent, meaning that not only does precipitation extend farther north, but it's much heavier as well with a much stronger low.