Weak storm Tuesday / Wednesday, unsettled into Friday.
Moisture is available, but dynamics are scarce – this sets the stage for some surprises, but overall, not too much snow on the way through Friday.
Retrospective - Overview
Tuesday evening kicked off several days of snow which wrapped up this morning, with the highest accumulations in central and northern Colorado. This has kept us holding on to a shaky start to the season.
This morning, a few more inches of snow fell across northern Colorado until drier air and large-scale subsidence shut down snow production in the high country.
Very strong winds leave some questions about actual snow totals – for instance, the Buffalo Pass SNOTEL has picked up >3” of liquid precip, which would mean 2-3 feet of snow, but the actual snow depth gauge has bumped up by less than half of that.
For a breakdown of Tuesday night’s snowfall, and Wednesday night’s unexpected punch of deep NW flow for Summit County, check out our previous discussion.
Here’s what the SNODAS model thinks happened. In general, this model tends to underestimate a little, but it’s pretty correct in terms of overall distributions and impacts.
In general, the Park Range (Steamboat and Buffalo Pass) underperformed a little, while Summit County and parts of central Colorado overperformed. Do you want the specifics? Well, we have them in agonizing detail. Check out this new tool!
Forecast Verification and Model Comparison Tool
Last updated .
* = No reliable observations, verification uses SNODAS estimates which
may be incorrect.
"Score" accounts for the spread/range of an accurate forecast. Lower
spread = higher score. For instance, if 4" falls at a certain location,
the forecast for 2-7" gets less points than the forecast for 3-5". Points are deducted for very inaccurate forecasts.
Ah… it feels good to outwit a cluster of supercomputers. In general, models botched this storm. Only the CAIC WRF had a better than 50% verification rate (at 51%, haha), and incredibly, the National Weather Service’s model (which takes human forecaster inputs!) did terribly. Thomas came in with a cool 70% verification rate, though needed a slightly larger spread to get there – despite that, he still came away with the highest score.
We’re going to go through a spiel here since it’s the start of the season…
So what’s the deal with the scores and spreads and such? We’re trying to set up this tool so that models and forecasters which are less consistent or confident (large spreads in their forecast, e.g. 5-12”) are penalized more and need to pick up more wins to beat models with tighter spreads. Obviously, the smaller the forecast range, the less likely you are to be correct, so we award more points to models that get forecasts correct with smaller ranges. We also subtract points (well, our code does) if you’re particularly wrong.
This also keeps us from putting down 2-16” forecasts and patting ourselves on the back for an accurate forecast. There is, of course, a balance. Many readers want a small range of numbers, but at some point forecasters would be providing somewhat disingenuous of a service.
It is rare when there isn’t enough data disagreement to put down a nice 7-8” forecast for a certain resort (as opposed to 4-8”), and if you do that despite a lot of evidence that the atmosphere may not quite set up the way you expect, you’re just hiding some very tangible uncertainty which could easily rear its head.
Part of the utility of these handwritten forecast ranges, in our opinion, is that they convey the range of probable outcomes which is something a single model run can’t provide. If the forecast range is large – well sure, it’s us wanting to be correct, but it shows the level of uncertainty we have for the atmospheric setup, model disagreement, etc. We want to minimize those spreads, but we also want to minimize surprises.
In particular, Thomas’s style is to avoid low busts, and he strongly weights evidence that setups could be less productive than expected. With only two low busts for the previous storm, that was a nice strategy – and less high busts (booms) than the models, too – so not overly pessimistic.
Let’s analyze the analysis, if you’re in to that kind of stuff. We’ll be keeping track of this stuff throughout the season to see which sources are most reliable, where the biases are, and if we can continue to outwit the weather models…it’s personal. If you can’t fight the ski traffic to slay some chowder / dangerous avalanche conditions, you might as well look at a bunch of numbers.
Forecast Verification Analysis
Here are the overall results. Dry bias means the source generally underforecasted (more snow fell than forecasted), wet bias means the opposite.
Thomas Horner: 44pts / 70% verified. Dry bias. 🍻Oh yeah.🍻
NAM 12km model: 40pts / 43% verified. Slight dry bias.
National Weather Service NDFD model: 18pts / 24% verified. Extreme dry bias. 🔔Shame.🔔
So let’s talk about the NWS NDFD model. This is the model that powers weather.gov – it’s a high resolution model blend that allows NWS forecasters to modify the grids based on their own expectations. The NWS was very pessimistic about this storm, which wasn’t the right call.
Internally, there has been a push to get the NWS to switch to the NBM and stop modifying their forecast grids – with some evidence that the NBM performs better overall. It certainly won out for this storm!
The Euro did worse than the GFS – again – though some of that is due to the big swings the GFS was taking ahead of the storm which allowed it gave it some larger spreads (average spread: 4.2) for some areas. The Euro was a lot more consistent in the leadup to the storm (average spread: 2.2), which is reflected by its score only being a bit lower.
The CAIC WRF had the best verification, though it’s the only high-res mesoscale model here (the NAM is mesoscale, but we wouldn’t consider 12km to be high-res). It’s also one of the only models with dynamic snow-liquid ratio calculation (the NBM and NDFD do that as well) – for the other models we used a fixed 15:1 ratio. If you don’t like that, too bad.
That’s the statewide picture. It gets a bit more interesting when we look at individual regions.
11 areas forecasted: the Aspen resorts, Crested Butte, Cottonwood Pass, Kebler Pass, Independence Pass, Monarch, Powderhorn, Sunlight.
Overall: Dry bias, but almost everyone overforecasted Powderhorn. Thomas had the best verification, but the Euro did pretty well with much lower forecast spreads. Only Thomas really caught on to the potential for higher snow totals, most models were much too dry.
ECMWF (Euro) model: 11pts / 45% verified.
NAM 12km model: 10pts / 45% verified. Slight dry bias.
Thomas Horner: 9pts / 72% verified. Slight dry bias.
National Weather Service NDFD model: -1pts (lol) / 9% verified. Extreme dry bias.
12 areas forecasted: A-Basin, Berthoud Pass, Cameron Pass, East Portal, Echo, Eldora, Hidden Valley, Jones Pass, Keystone, Loveland, St. Mary’s Glacier, Winter Park.
Overall: Dry bias. Everyone underforecasted Berthoud Pass, and Cameron Pass came in with much less snow than many expected. Overall, everyone struggled here due to the unexpected snow production on Wednesday night.
NAM 12km model: 10pts / 42% verified. Strong wet bias.
Perhaps next time we’ll take a look at what some other forecasters are doing.
Snow in Denver: When!?
In our initial forecast we talked about potential for snow on the urban corridor – and its unlikelihood. Sure enough, Denver has gotten through another week without measurable snow. The record latest snowfall day of November 21st is now rapidly approaching, without any strong signal for snow on the horizon.
The next best chance comes near the end of next week – just before the record latest date – but prospects aren’t looking too good.
If Denver blows that chance, the city will set a new record for the latest date of first snowfall.
Forecast Discussion: This Weekend
A weak shortwave looks to barely graze the northeast corner of Colorado on Saturday night. This should keep some light snowfall going across the northern Colorado mountains on Saturday, in addition to pushing a dry cold front down the Front Range late Saturday night.
Before then, temperatures will warm to the 60s on the plains, and the cold airmass will quickly erode on Sunday to allow for another day of 60s, thanks to strong downsloping westerly winds.
You could probably find a little bit of fresh, soft snow this weekend in the Park Range and north Front Range.
All you need to worry about are the 90mph gusts on Sunday! But for real, wind gusts are going to be considerable up at altitude, at least in northern Colorado and especially along the Divide. Saturday will be a bit better (30-60mph).
This map shows the strongest wind gusts that are forecasted over the weekend:
Forecast Discussion: Next Week
After the shortwave this weekend, ridging again takes hold over much of the western United States.
Warm air advects into the region on Monday, with winds a bit calmer than the weekend. A mild day is in store for the state.
Another storm has the audacity to barely graze us to the north on Tuesday / Wednesday. The edge of an upper level trough should introduce some large-scale lift into the state, with a good pocket of moisture advecting through the Colorado high country from Tuesday morning to Friday night.
Despite the moisture, we only see dynamic support from Tuesday to Wednesday morning (jet right entrance region) and periodically through Thursday with some very disorganized flow overhead. Large scale subsidence really takes hold into Friday as a jet max moves over Colorado and then puts us in the left entrance region. Near the surface, upslope orographic winds will be light Wednesday through Thursday morning due to divergence aloft.
There’s still some time for models to screw around with this setup, but like the past many systems we’re just not too excited about the prospects for consistent, widespread snowfall. As always, a few areas will likely get lucky with some localized dynamics and deeper moisture, but overall, not a particularly productive pattern.
Even the more optimistic ensemble members aren’t painting too exciting of a picture:
We’ll give you a shout early next week with the latest updates, but we’re not expecting much of a change.
This storm will push another cold front down the Front Range on Wednesday morning-ish, but it doesn’t look to provide enough cold air or moisture to really put a risk of snow in the forecast for Denver.
That said, the very latest run of the Euro has Friday to Saturday looking pretty spicy, but its ensemble members, along with other weather models, wholeheartedly disagree, so we’ll table thoughts about that for now.
Basically, if you’re looking for some fresh turns next weekend….eh:
Let’s see which way the forecast trends over the next couple days!
More frequent updates on Twitter, Instagram, and Facebook:
💙 Supported By 💙
We love doing this and making our work available for free! But it does
take a ton of time to put these forecasts together, develop our custom
graphics, and keep our servers running. Consider throwing some beer or
coffee money our way:
Support us on Patreon!