Jns2183

9.25". Crazy rates. If I can hold change over enough 30min I'll be over 10" Sent from my SM-S731U using Tapatalk

I'm just happy I have additional support for my totals due to not measuring my board setup for several hours due to sleep. Definitely have had a bit of a crazy band setup over me Sent from my SM-S731U using Tapatalk

8.5" Sent from my SM-S731U using Tapatalk

Over 8" This might be one of the more insane temperatures gradients Sent from my SM-S731U using Tapatalk

It also went back and forth forth for a couple hours after first pingers 1" last 30 min 7.75' Sent from my SM-S731U using Tapatalk

1/4mile visibility Sent from my SM-S731U using Tapatalk

1/2" in 17min Finally past my NAM storm snowfall prediction Arkansas areas got 8" of sleet Sent from my SM-S731U using Tapatalk

I think it slows down approach from south but increases from the east. You look at sleet line now it's basically flat. This starts tilting it some. Depends on how fast it takes over and starts pulling down cold at mid levels. We still fully changed over but for next few hours we have prime forcing. If we can hold out 3 hours that's another 3-6" on top of what we have now. The better the forcing that longer we can hold of. Happy that forcing setup right over us. I think I have good chance at 10-12" before hint of Sleet. Sleet will still accumulate some Sent from my SM-S731U using Tapatalk

The mid-level never really transfers that is why we deal with the sleet and don't have a ccb that forms. Sent from my SM-S731U using Tapatalk

Better view of that 850-700mb frontogensis sitting over us Up to 6.75. I've almost reached NAM predicted snow total Sent from my SM-S731U using Tapatalk

Coastal low has formed off mattress. Also we have some major lift right over us. Sent from my SM-S731U using Tapatalk

Good in terms that the mixing line down there is heavily rate dependent. Also I don't think I'll be sleet at 11am like the nam last night depiction Sent from my SM-S731U using Tapatalk

I believe kmdt was at 3.2" as of 7am I've just passed 5" here on way to 6" Sent from my SM-S731U using Tapatalk

It truly is NAM vs world. Even the RRFS gives kmdt 11" with everyone else 12"-19" with most 13"-17". The nam and 3k are 7". I compared latest runs of 13 models below Sent from my SM-S731U using Tapatalk

I'm near capital city airport and it says snow mainly after 3am 2"-4" Day 6"-10" Night 1"-2" 9"-16" Sent from my SM-S731U using Tapatalk

Took the 0z observed soundings in Nashville and Greensboro n/c and compared them to 0z hour on 18z run of hrrr, nam, GFS. Subject: 00z Sounding Validation Analysis: Mid-Level (850-500 hPa) Model Performance Report This analysis evaluates the handling of the mid-tropospheric environment for the January 25, 2026, 00z cycle. Validation is performed against observed balloon soundings at Nashville (KBNA) and Greensboro (KGSO), focusing on thermal profiles, moisture depth, and vertical motion (omega). I. Nashville, TN (KBNA) - Analysis of Pre-System Moistening The 00z KBNA sounding revealed a transition toward a more saturated mid-level environment as synoptic-scale lift began to overspread the region. * Mid-Level Warmth (Thermal Profile) * HRRR Performance: Exceptional handling of the 850-500 hPa lapse rates. The HRRR achieved an RMSE of 0.57°C, accurately resolving the cooling trend at the 700 hPa level (-2.4°C observed). * NAM/GFS Performance: Both models exhibited a notable warm bias in the mid-layers. RMSE values hovered around 0.93°C, with both models over-forecasting the 700-500 hPa thickness, likely due to slightly aggressive warm air advection (WAA) timing. * Mid-Level Moisture (Saturation Depth) * HRRR Performance: Most representative of the moisture plume depth. It maintained a Dew Point RMSE of 1.05°C, correctly identifying the saturation of the mid-levels. * NAM/GFS Performance: These models were too aggressive with the "dry slot," keeping the 700-500 hPa layer too dry compared to the observed moisture. This would have resulted in an under-forecast of mid-level cloud opacity. * Mid-Level Lift (Omega/Vertical Velocity) * HRRR/NAM Alignment: Both models correctly signaled weak but consistent large-scale ascent. Observed Omega values in the mid-levels ranged between -0.1 and -0.3 µb/s. * GFS Performance: Too neutral. The GFS failed to capture the subtle dynamic lift associated with the incoming shortwave, lagging behind the high-resolution guidance. II. Greensboro, NC (KGSO) - Analysis of Subsidence and Dry Stability In contrast to Nashville, the Greensboro environment was dominated by a dry, stable mid-level regime associated with a prominent subsidence inversion. * Mid-Level Warmth (Thermal Profile) * HRRR Performance: Led the field with an RMSE of 1.16°C. It was the only model to successfully resolve the sharp gradient of the subsidence inversion near the 800-750 hPa layer. * NAM/GFS Performance: Poor handling of the stable layer. Both were significantly too warm (RMSE 1.89°C to 2.12°C), essentially "washing out" the inversion and over-predicting the mid-level temperature. * Mid-Level Moisture (Dry Layer Integrity) * HRRR Performance: Closest match to the observed dry slot with an RMSE of 1.80°C. It maintained the integrity of the dry mid-level air. * NAM Performance: Major outlier. The NAM exhibited a massive moisture bias (RMSE 5.11°C). It forecasted a nearly saturated mid-troposphere, whereas the sounding showed a significant dew point depression. This represents a significant failure in boundary layer/mid-level moisture mixing. * Mid-Level Lift (Omega/Vertical Velocity) * HRRR/GFS Alignment: Correctly identified the sinking motion (subsidence) prevalent in the post-frontal/high-pressure regime. Both showed positive Omega values (descending air) of +0.1 to +0.4 µb/s. * NAM Performance: Showed very weak, nearly neutral vertical motion. This lack of sinking air likely explains why the NAM failed to keep the mid-levels dry. III. Summary Numerical Ranking (850-500 hPa Layer) Scores are based on a normalized average of RMSE across Temperature, Dew Point, and Wind components specifically within the mid-troposphere. * HRRR (Score: 0.72) — TOP TIER * The clear winner for vertical resolution. It is the only model currently capable of resolving sharp inversions and mid-level moisture gradients with high accuracy. * GFS (Score: 1.08) — MID TIER * Reliable for general synoptic trends but lacked the resolution to handle the specific thermal structure of the inversion at KGSO and the lift at KBNA. * NAM-Nest (Score: 1.25) — LOW TIER * Significant moisture handling issues at KGSO and a persistent warm bias in stable layers. Use with caution for cloud cover and precipitation type forecasting in this regime. Technical Conclusion For practitioners focusing on mid-level dynamic features (e.g., cloud depth, precipitation efficiency, or lapse rate steepness), the HRRR is currently providing the most reliable guidance. The NAM is currently showing a high "false alarm" rate for mid-level moisture in high-stability regimes. Sent from my SM-S731U using Tapatalk

There isn't any really. I'm using a super raw method to do it by getting nohcos 24hr snowfall map based on some observations then estimated. Then getting images from models for same period. Using python to try to make sure maps are for same area because I have it map each pixel location, and RGB value and assign snow based on legend RGB value. Uploaded it all to gemeni pro and have it run that statistical analysis, pull into my notebooklm of meteorology books, papers, to help interpret, then produce nice infograph for it all. Haha, honestly it's a pain and wish I could automate it more. But worth it for first big storm to explore the concept. Take into account there are a myriad of ways to arrive at right snowfall for wrong reasons. I'm hoping it harder to arrive at large area of right snowfall for wrong reasons. Sent from my SM-S731U using Tapatalk

Scorecard from Tennessee/Missouri valley 24 hr snowfall ending 0Z 1/25 Sent from my SM-S731U using Tapatalk

It definitely has. I'm just starting what was the results of analyzing model performance in Arkansas up to 6pm today. I'll be able to start looking at states to East soon. All the models there had about the same mean absolute error of 1.8". It's was the nam though with a root mean square deviation of almost twice that, while others were close, along with a massive dry bias, and horrible snowfall distribution correlation that stood out. Below is a metaphor Thinking about these weather models is a bit like judging a free-throw shooting contest between five different players. Even if they all missed by the same average distance, the way they missed tells the real story. Imagine each model is a basketball player trying to hit a specific spot on the rim: The "Average" Score (MAE) Every player in this contest missed their target by about 1.8 inches on average. If you just looked at that one number, you’d think they were all equally "okay" at their jobs. But once you look at the game tape, one player (the NAM) stands out for all the wrong reasons. The NAM: The "Wild Card" (RMSE & Bias) The NAM was the player who didn't just miss—it missed spectacularly. The Big Misses (RMSE): While the other players missed by 1 or 2 inches consistently, the NAM would hit the backboard or miss the rim entirely on some shots. Because RMSE penalizes big mistakes much more than small ones, the NAM's "penalty score" was twice as high as the others. The "Dry" Excuse (Bias): On top of the wild misses, the NAM had a massive dry bias. In our metaphor, this player was consistently shooting way too short. If the hoop was at 10 feet, the NAM was aimlessly throwing the ball at 8 feet. The "Broken Rhythm" (Correlation) Finally, there’s the snowfall distribution correlation. In plain English, this is rhythm. A good player might miss, but if the target moves left, they move left too. The NAM had horrible correlation. It was essentially playing a different game. When the actual storm "moved left," the NAM "moved right." It didn't just get the amounts wrong; it got the entire pattern of where the snow would fall completely backward. Sent from my SM-S731U using Tapatalk

Here Sent from my SM-S731U using Tapatalk

In Arkansas it was best model for snow. The GFS, euro were close. Sent from my SM-S731U using Tapatalk

It's driving me nuts. Something is there but after how I saw it do so far, a pixel analysis correlation of .18 vs every model being 2.5x-3.5x higher because of how bad it screwed up the changeover line Sent from my SM-S731U using Tapatalk

The HRRR Consistency: The HRRR was the most balanced. It had the lowest error across both categories, making it the most reliable tool for this specific event. It handled the "Rain-Snow line" better than any other model. Sent from my SM-S731U using Tapatalk

Same as before I mean it also never backed down in Oklahoma and Arkansas as was and this was the result The NAM Split: The NAM is the most interesting case. For liquid rain, it has a staggering 0.96 correlation, meaning it mapped the storm's shape almost perfectly. However, for snow, its correlation plummeted to 0.18. This proves the NAM understood where the moisture was, but failed completely at the "transition physics"—it thought the air was too warm, keeping it as rain when it should have been snow. Every other model had snow correlation between .5 to .7 Sent from my SM-S731U using Tapatalk

I've also never seen a sleet changeover blow through at 60mph like it wants Sent from my SM-S731U using Tapatalk