It's Better to be Lucky than Good

Yesterday was one of those days when if you were skiing at Alta in the afternoon, consider yourself blessed or, alternatively, it's better to be lucky than good. 

From 1100 to 1600 MDT, Alta-Collins picked up 9" of fresh, including 3" in an hour from 1200 to 1300 and then again from 1400 to 1500.  Water equivalent was .46", so this was 5% water content.  Winds on Mt. Baldy during that period never guested over 10 mph.  

I took a look at forecasts from the 12Z models on the prior day (2 April) and the GFS was going for nothing.  The HRRR .15" water and 2.7" of low-density snow.  Even yesterday morning, expectations were low.  The Utah Avalanche Center Forecast that morning called for 0.5 to 1" of snow.

This isn't to throw them under the bus as they do a great job, but just to illustrate that yesterday's snowfall was pretty unexpected.  It wasn't handled well by the models or the forecasters.  So what happened?

Well, morning broke with not much happening other than a few scattered snow showers.  At 1459 UTC (0859 MDT), there were some light returns on radar, but nothing to get excited about.  

However, the flow was light and the airmass unstable, and with a little surface heating, convection began to get going.  By 1857 UTC (1257 MDT), during an hour in which Alta picked up 3" of snow, localized convective snow showers had developed over portions of the central Wasatch, especially around Little Cottonwood, and the high terrain down to Mt. Timpanogos.  

Even then, the radar wasn't all that impressive, but the relationship between radar reflectivity and snowfall rate is not a good one.  Low-density snow of the type that fell yesterday often doesn't light up radar screens.  This is why it's so valuable to have weather cams and automated snow depth sensors to monitor actual conditions at the ground.  

The development of the first snow showers over high terrain was probably favored by the light flow and unstable conditions, with daytime heating yielding upslope flow and convergence over the mountains.  

The convection became more widespread with continued surface heating as evident in the 2033 UTC (1433 MDT) radar image from another period when Alta got 3" in an hour.  Nevertheless, snow showers persisted over upper Little Cottonwood.  

It's very difficult to reliably predict the location and intensity of these snow showers.  We can anticipate their development, but questions of where, when, and how intense are hard to answer reliably.  Neither our current models nor human cognition are very good at distinguishing a situation like the one yesterday from one where the snow showers are less productive and maybe provide light accumulations. The processes are simply too small in scale and too sensitive to small changes in the atmospheric stability and moisture content.  Basically, yesterday really was a tough forecast, at least with lead times of more than a couple of hours. 

I suspect if you were skiing yesterday afternoon, you probably didn't care.

Thoughts on Great Salt Lake Effect Forecasts from Computer Models

Great Salt Lake Effect forecasts from computer models have historically exhibited poor reliability for a few reasons.  

One is that many computer models have had insufficient resolution (or grid spacing) to adequately resolve the Great Salt Lake and surrounding mountain effects.  

Another is that the Great Salt Lake is simply very sensitive to small changes in the ambient flow and surface conditions, so when you look at a handful of model runs, they don't always capture the full range of possibilities.

Let's take a look at how these issues are manifest in current model runs, starting with the GFS.

The effective grid spacing of the GFS is 13 km.  At it's current level, the Great Salt Lake is about 25-km wide, meaning that in northwesterly flow, the GFS at best has two or three grid points that include lake characteristics.  In addition, the GFS is unable to distingish between the Stansbury, Oquirrh, and Wasatch Ranges, so it cannot separate mountain effects from lake effects.

Below is the GFS forecast for 1200 UTC (6 AM MDT) tomorrow.  The background lake and terrain are not from the GFS.  The lake is from an older, "good old days" lake level and the terrain is fairly realistic.  I use them for reference.  Note that the GFS is capable or producing lake-effect precipitation, but as an enormous blob covering a huge area.  While it is possible we will get lake effect, it's not going to look like this.  

The HRRR, on the other hand, has better resolution.  It has 3-km grid spacing and it better resolves the Great Salt Lake and various mountain ranges of northern Utah.  If we look at the HRRR forecast valid at the same time, we see considerably more structure due to this better resolution.  It produces a lake band that is more concentrated and it also produces orographic precipitation over the central Wasatch.  This is somewhat more realistic.  

That's good, but studies have shown that high resolution models still have problems. Several years ago John McMillen, a graduate student in my group, did retrospective forecasts of 19 lake-effect events using the Weather Research and Forecasting (WRF) model with 1.33 km grid spacing, even finer than the HRRR.  He found that those forecasts were "band happy."  While they produced banded forecasts of banded events, they also produced banded forecasts of non-banded events.  Basically, the model could not reliably distinguish events that were banded from non-banded.  So, when the HRRR generates a band, that's great, but you can't have a lot of confidence that Mother Nature will do the same.  

He also found that the bands were often in the wrong place and most commonly to the right (relative to the downstream flow direction) of the observed band position.  This is something that has also been identified in other model forecasts over other lakes, such as Lake Ontario.  

Another challenge for getting the lake-effect location right is sensitivity to the large scale flow.  The centers of the GFS and HRRR bands in the forecasts above are clearly different, but so are the flow fields.  Small changes in wind direction make a big difference, and current ensemble modeling systems can't paint out the full range of possibilities due either to insufficient resolution or too few members.  

So, many elements are in place for lake-effect tonight.  Many elements are also in place for orographic precipitation.  Either could materialize, but location and intensity are difficult to anticipate at this time.  

I look forward to seeing how this all plays out.

Expectations for the Coming Ski Season

You may have seen some guesses about what will happen this coming ski season.  Such predictions are often issued with great fanfare and overconfidence.  I'll pick on AccuWeather here because they are an easy target.  The outlook they issued in late September 2022 included the following.

"Unfortunately, we have bad news as far as the drought goes in parts of California, Nevada and the Southwest," Pastelok said. "The main storm track will be even farther north than it was the first half of the winter season last year including the late fall."

Then there was the accompanying snow outlook. 

What a spectacular crash and burn!

The reality is that these seasonal forecasts, based largely on the loading of the dice due to the anticipated presence of El Nino or La Nina, have somewhat limited practical utility for most applications.  If you didn't have any knowledge of meteorology, you might expect there to be a 33% chance of below average, 33% chance of near average, and 33% of above average precipitation in any given season.  At best, we can shift these odds just a bit.  For temperature, we can shift them a bit more simply because of the growing influence of global warming.  

So let's look at expectations for this season.  El Nino conditions are currently in place across the equatorial Pacific Ocean with the most anomalously warm sea surface temperatures (SSTs) off the South American coast, but anomalously warm sea surface temperatures extending across the entire equatorial Pacific.  

Source: NOAA Climate Prediction Center

El Nino is expected to persist through this winter (the Climate Prediction Center gives a > 95% chance it will survive through at least February).  This affects thunderstorm characteristics in the tropics, which in turn can affect the midlatitude jet stream.  Correlations between El Nino characteristics and temperatures and precipitation over the western US weigh heavily into seasonal forecasts, with recent long-term warming trends also considered. 

For Nov-Dec-Jan this leads to the dice being loaded for above average temperatures across the western Continental US.  In this case for Utah, there is a 40–50% chance of above average, or what the Climate Prediction Center calls "leaning above" (see scale at lower left).  

Source: NOAA Climate Prediction Center

For precipitation during Nov-Dec-Jan, they are loading the dice slighly for below average in the Pacific Northwest, but indicate equal chances across the rest of the western continental US.  Basically, they don't have any confidence in it going one way or the other.  

Source: NOAA Climate Prediction Center

Below are the projections for Feb-Mar-Apr.  

Source: NOAA Climate Prediction Center

Source: NOAA Climate Prediction Center

I'm a little surprised that they aren't leaning above for precipitation in portions of SoCal with El Nino being expected.  This may be due to the fact that this was issued in mid August and perhaps we'll see that appear if confidence in the El Nino grows over the next month or two.  

Bottom line here is that I see no reason to change the seasonal snowfall forecast that I have issued for the past several seasons in the Wasatch Range.

I have no idea what is going to happen. 

I guess I might add on caveat and that is that I would lean toward above average temperatures for the cool season and a vulnerable low-elevation snowpack.  As we saw last year, a favorable jet stream can still give us an enormous low-elevation snow season, and we will need probably need an active storm track with some colder storms to overcome recent warming trends and what are currently remarkably high global temperatures.  

Limitations on Predictability

Today's 0600 UTC initialized GFS forecast, covering the period from 1800 UTC (1200 MDT) Thursday 27 September through 1200 UTC (0600 MDT) Tuesday 2 October is below and calls for a mid-latitude trough to interact with monsoon moisture and the remnants of Hurricane Rosa to bring precipitation into the southwest U.S., including Utah. 

It is a very believable forecast.  You will almost surely see tweets of forecasts from this and other models with statements about what is likely to happen, but we need to remember that we are still 4 days from the initial arrival of forecast precipitation into Utah.  Further, uncertainty remains concerning the interaction of the mid-latitude trough and the hurricane, something that we discussed in the previous post. 

The likelihood that the remnants of Rosa move across the southwest U.S. is now high, but not 100%.  Most tracks produced by the GEFS ensemble, for instance, bring the low center across Arizona, but there are two that shunt it off across New Mexico and two that keep it over the Pacific. 

The European ensemble favors tracks somewhat further to the west (not shown, or I have to kill you), most going through western Arizona or southern California, but a few remaining out over the Pacific.

This, combined with other details of the moisture surge and the movement of the trough through the western U.S., yields large spread in the precipitation forecasts being produced by the GEFS and other ensembles.  In the case of the GEFS, total accumulated precipitation at the Salt Lake City airport through 0600 UTC (0000 MDT) 5 October varies from as little as .03" to as much as 1.87".  An ensemble like this does not resolve thunderstorms and other small scale features that might create further chaos for the distribution of precipitation. 

Source: NCEP

If you think that is bad, check out the plume diagram for Yuma Arizona.  Note in particular the scale change.  There, total accumulated precipitation varies from .02 to 4.16 inches, most falling over about a 2-3 day period.  

Source: NCEP

This is a situation in which we are going to see a pattern change.  It is likely that we will see some precipitation and that the rainfall could be significant and potentially hazardous in portions of the state, especially burn scars.  However, it is still too soon to anticipate local impacts with any reliability.  That will come eventually as the forecast lead time shortens.  Avoid letting specific predictions and single model solutions shared on social media and the like from biasing your perspective of this event.  There remain a wide range of possible outcomes for precipitation.  

Hurricane Rosa Causing Forecaster Heartburn

Rosa is currently a category 1 hurricane located over the eastern Pacific south of the Baja Peninsula.

Source: National Hurricane Center

Over the next several days, she is expected to drift toward the WNW and interact with a complex a complex sequence of mid-latitude troughs.  The latest (0600 UTC) GFS calls for Rosa to curve toward the northeast, with her remnants migrating through the interior U.S. southwest and eventually contributing to a very strong mid-latitude cyclogenesis event along the U.S. Canadian border.  

The concerns associated with such a scenario are numerous, including the possibility of heavy rainfall on new burn scars (with related debris flows) and what would probably be some exciting weather for early October in the high plains of the U.S. and Canada. 

The problem is this.  These sorts of interactions between tropical cyclones and mid-latitude troughs are notoriously tricky to forecast.  Yesterday, the European center model and ensembles favored a path for Rosa that basically took her westward.  In contrast, the GFS favored a solution similar to the one above.  Basically, you had model solutions that clustered around two tracks, one with a harmless route out to sea, the other with a route through the southwest.  This is known as a bifurcation and it is very common when tropical cyclones interact with midlatitude troughs and it yields dramatically different forecast outcomes.

The latest ensemble forecasts produced by the GEFS (top image below) favor a track through the southwest.  The European ensemble (bottom image below) has more spread.  Most go for the southwest solution, but there are others that go out to see or remain offshore.  These are from Brian Tang's Tropical Cyclone Guidance web site at the University at Albany. 

As things stand now, this is not a concern through the weekend except perhaps very near the U.S.–Mexico border.  Monsoon moisture would likely begin to move into the southwest in earnest late Sunday and early next week, with Rosa's remnants arriving Monday night at the border and moving through the southwest through Tuesday. 

This is a forecast that bears watching, but regional and local impacts cannot be confidently predicted at this time and there are a wide range of possibilities. For Utah, the phasing between the midlatitude trough, Rosa, and monsoon moisture in advance of Rosa will ultimately determine what we see and where moisture affects the state.  This is a prime example of a situation in which a single model run (e.g., the GFS) provide very little useful guidance.  Ignore the icon based forecasts as well, although you should do that all the time anyway.  

Predictability Limits Don't Change Because of the Holidays

Christmas is still a full week away, but that isn't preventing people from making specific forecasts for the big day. 

This is a common issue when a major day or holiday approaches.  The desire to provide specifics far in advance.  10-day forecasts for trick-or-treating weather on halloween.  Guarantees of a white Christmas, etc.  Unfortunately, the existence of a holiday does not magically clear up the meteorological crystal ball or alter the fundamental chaotic nature of the atmosphere. 

I saw one forecast this weekend for a brutally cold Christmas day with a high of 18ºF.  Presumably this was based on forecasts like the one below from yesterday morning's GFS which did indeed have some exceptionally cold air over northern Utah with 700-mb temperatures below -20ºC.

However, last night's GFS paints a totally different picture with a ridge over the west and 700-mb temperatures of -4ºC over northern Utah.  For the ski areas, that's a swing in temperature of about 17ºC (30ºF)!

Cold air outbreaks on a continental scale can often be anticipated may days in advance, but the specifics for one particular area are more difficult to nail down.  Northern Utah is difficult because we tend to be on the western edge of the cold air.  The shift in the model forecasts above reflects the uncertainty in the forecast more than a model trend.  It doesn't mean we're going to miss the cold air (or be warm).  It simply means that we have a range of possibilities due to the chaotic nature of the flows that generate these cold-air outbreaks. 

We don't produce a 700-mb temperature thumbnail plot from the GEFS, but below is one for the "1000-500-mb thickness" (dashed lines, some in red), which is proportional to temperature in the lower atmosphere.  These are valid for 11 PM Christmas Eve, when Santa will be delivering presents to your home.  Some of these runs would be quite cold (e.g., lower left panel and 2nd panel from right in center row), but others would be more seasonable or perhaps even warm (e.g., first and second panels from left in center row). 

Source: Penn State E-wall

Thus, endorsing a specific, individual forecast makes little sense at this stage.  On the other hand, it is important for meteorologists to discuss the range of possibilities for big holidays when travel volumes are likely to be high. In that regard, mentioning that there could be a cold surge into the area seems appropriate, if it is also noted that it is still unclear if that surge will push into the Great Basin or remain east of the divide and over the northern Plains. 

Western Showers, Eastern Deluge?

Changes are afoot that will result in a shift in the weather for Utah and a significant rainfall event in the eastern United States.

The pattern currently developing over North America will give weather forecasters headaches from coast to coast.  It features multiple interactions between upper-level closed lows and high-amplitude ridges, with a tropical cyclone thrown in just for good measure.  The interactions between these features are notoriously difficult to nail down and this leads to considerable forecast uncertainty.

The low hanging fruit is in Utah.  Our run of well-above-average temperatures will come to an end on Friday as an upper-level trough from the Pacific swings into the Intermountain West, bringing cooler air and showery weather.

After that, it's tough to say.  The models have been all over the place dealing with the interaction of Friday's trough with another upper-level trough dropping down from British Columbia over the weekend (such an interaction between two troughs is called the Fujiwara effect).  I'm giving up on details for the weekend except to say that temperatures will thankfully be more seasonable.

The real forecast challenge in the east.  A significant rain system is sweeping through the northeast currently, then it looks like a prolonged rain event will develop along the mid Atlantic and southern New England coast through Friday night in advance of tropical cyclone Joaquin.  Such rain events are sometimes called predecessor rain events or PREs.

After that, we have diverging solutions from the various members of the GEFS and Euro ensembles regarding the track and intensity of Joaquin ranging from a direct track into the mid atlantic states (see GFS above) to a track that keeps it well offshore.  The Euro ensemble has more members keeping Joaquin or its remnants offshore, whereas the GFS has more members calling for something moving into the mid atlantic or at least along the coast (see below).

Source: Penn State e-Wall

Put them together and you have a huge range of possibilities.  A good summary is provided by the National Weather Service Sterling, VA area forecast discussion:

"FORECAST CONFIDENCE SIGNIFICANTLY DECREASES OVER THE WEEKEND AS MODELS CONTINUE TO STRUGGLE TO RESOLVE THE BLOCKING PATTERN (CLOSED LOW TO OUR SW...HIGH OVER NORTH ATLANTIC) AND HOW TC JOAQUIN INTERACTS WITH IT. SOLUTIONS RANGE FROM A COASTAL IMPACT TO THE TC REMAINING WELL OFFSHORE. PLEASE REFER TO NHC BULLETINS FOR THE LATEST DETAILS AND BE AWARE THERE IS AT LEAST POTENTIAL FOR HEAVY RAIN AND FLOODING (THE LARGEST THREAT GIVEN THE TWO PREDECESSOR RAIN EVENTS)...WINDY CONDITIONS...AND COASTAL FLOODING OVER THE WEEKEND."

Overall, this is a great example of a low predictability pattern, but one with the possibility of high impact weather.  Rainfall associated with the PREs is likely, but precisely where and how much.  Then, will that be followed by a further deluge?  From a scientific perspective, the challenge is developing ensembles that provide reliable estimates of event probability.  From a communications perspective, the challenge is how to effectively inform the public and decision makers when the range of possibilities is very large.

Astronomical Spring, Meteorological Reality

Astronomical spring arrives today, but no matter what my daffodils say, meteorological reality will beg to differ in the coming days.

This morning we have some snow falling in the mountains and rain in the Salt Lake Valley.  The precipitation began with your classic "cloud storm" scenario.  An incredibly dry airmass was in place over the Salt Lake Valley this morning, so much of the early precipitation evaporated before reaching the ground.  In this morning's sounding, taken about 5 am MDT, there was a 34ºF dewpoint depression at the surface (which equates to a relative humidity of 25%) and even drier air just above the surface layer.

Source: NOAA/NWS/SPC

A cold front moves in tonight, followed by a prolonged period of northwesterly flow that will gradually bring in even colder air on Thursday and Friday.  By Saturday morning, our 700-mb temperatures are forecast to be about -16ºC, which is below average even for January.

The overall change in temperature is perhaps the easy part of the forecast.  Precipitation is far more difficult.  As we have discussed a couple of times this week, there are wide variations in the timing and intensity of precipitation forecast by the models.  Yesterday's NAM, especially the high-resolution nest, clearly blew the precip forecast for today (see Does Precision = Accuracy), providing a good illustration that high resolution is not necessarily a forecast panacea.  For Alta, here are some numbers produced by the 0600 UTC GFS, 1200 UTC NAM-12-km, and 1200 UTC NAM-Nest-4-km:

GFS

6 AM Wed – 6 PM Wed: 0.18" SWE, 1.3" Snow

6 PM Wed – 6 AM Thu: 0.31" SWE, 2.8" Snow

6 AM Thu – 6 PM Thu: .10" SWE, 1.4" Snow

Total: .59" SWE, 5.5" Snow

NAM-12-km

6 AM Wed – 6 PM Wed: 0.05" SWE, 0.5" Snow

6 PM Wed – 6 AM Thu: 0.13" SWE, 1.3" Snow

6 AM Thu – 6 PM Thu: .02" SWE, 0.4" Snow

Total: .20" SWE, 2.2" Snow

NAM-Nest-4-km (no snow algorithm applied)

6 AM Wed – 6 PM Wed: 0.23" SWE

6 PM Wed – 6 AM Thu: 0.87" SWE

6 AM Thu – 6 PM Thu: .09" SWE

Total: 1.19" SWE

So the SWE forecast produced by these three models varies from .20" to 1.20".  Good luck!  We don't run our snow algorithm on the 4-km NAM yet, but using something like a 10-to-1 ratio, which is close to what we are getting from the other models and consistent with the warm temperatures, gives us a range in snowfall of about 2–12 inches.  How's that for uncertainty.  

My read of this event is that something in the 4–8" range is the most likely scenario for the total accumulation in upper Little Cottonwood through tomorrow afternoon.  That accumulation will probably come in a few pieces, with this morning being the first.  Much of the snow will be high density, which we really need to stick to the bone rattling frozen granular that covers the upper elevations of the Wasatch.  

All of this being said, this is a situation where forecast uncertainty is high, as indicated by the model spread above.  

Does Precision = Accuracy?

Public forecasts issued by the National Weather Service and most news broadcasts are usually fairly vague with regards to snow accumulations.  They usually provide a range (e.g., 4-8 inches) for a broad period of time (e.g., tomorrow).

There are, of course, many users who need far greater precision than that.  Determining how much salt to apply to highway for the morning rush hour, or whether or not to close a canyon to traffic to perform avalanche control work requires knowing very precisely how much snow is going to fall and when.  The temperature and the strength of the wind might also be important for these applications too.  

With those examples in mind, let's think about the ramifications of the forecasts being produced for tomorrow (Wednesday) afternoon through Thursday morning by the 12-km NAM and its 4-km high resolution nest, which are remarkably different in terms of the timing and amount of precipitation.

For the three-hour period ending at 3 PM tomorrow afternoon, the 12-km NAM sags precipitation into Salt Lake County and the central Wasatch.  In contrast, the 4-km nest keeps the precipitation to the north.  For Little Cottonwood Canyon, this affects how early tomorrow road crews need to start treating and plowing the highway as skiers are preparing to leave the resorts (Note: it is going to be warm tomorrow, with snow levels near 7500 ft, which will help some, but I'm using this as an example, so go with it).     

Once the storm sags into northern Utah, the 4-km NAM generates a stronger storm, that lasts longer, as illustrated by the 3-h accumulated precipitation for the period ending at 9 PM tomorrow night.  

Here's another way to look at it.  Below is a time series of 3-h accumulated precipitation from the 12-km NAM (green) and the 4-km nest (white) from 1200 UTC (6 AM MDT) this morning through 0000 UTC 22 March (6 PM MDT Friday).  There is a tendency for the 12-km NAM to smooth out the precipitation, generating less abrupt peaks and valleys, whereas the 4-km NAM has greater variability, with a more pronounced peaks and valleys, but also more overall precipitation.  

What you are seeing in the 4-km NAM is the direct influence of resolution, which allows it to produce precipitation features with greater structure and detail.  That's great, but precision does not necessarily equal accuracy.  A high resolution model can produce intense, narrow precipitation bands that may look more physically realistic, but if they are in the wrong place, the forecast is arguably worse.  A time series like the one above can be very misleading in such an instance.

These are the growing pains that meteorology is presently experiencing.  Eventually, we will have ensembles of high resolution models, which will allow us to deal with some of the predictability challenges posed by precision.  Until that happens, we have to carefully interpret high resolution forecasts produced by a small number of models.  

Gut-Check Forecast

The surface cold front is through and we're starting to see precipitation fill in behind it over northwest Utah.

This is a gut-check forecast if ever there was one.  The models are producing a narrow band of heavy precipitation behind the surface front.  The 1800 UTC NAM dumps out more than 0.75" of liquid precipitation in areas along the band by 1200 UTC (0500 am) tomorrow.  Snow levels will be flirting with the benches soon.

1800 UTC NAM accumulated precipitation (inches following scale at top)
from 1800 UTC 6 Mar – 1200 UTC 7 Mar 2012

The precipitation band, however, is quite narrow and any error in position or movement has major ramifications.  One way to measure the forecast uncertainty is to run lots of different forecast models and see how many produce large precipitation accumulations.  The Short Range Ensemble Forecast (SREF) system run by the National Centers for Environmental Prediction includes 21 model forecasts which, when added to the NAM forecast, enables us to look at 22 different model forecasts.   The plot below shows the percentage of those forecasts that produce at least 0.5" of liquid precipitation from 1500 UTC (0800 MST) 6 March to 1500 UTC (0800 MST) 7 March.   

Percent of SREF members producing at least 0.5" of
liquid precipitation from 1500 UTC 6 Mar – 1500 7 Mar 2012
Source: SPC

About 50% of the ensemble members go for at least 0.5" of precipitation immediately south of the Great Salt Lake (such as the NAM forecast above), while only 10% go for at least that much in a band that extends across northern Utah.  Many of these forecasts include a band of precipitation, but there is variability in the intensity, position, and movement, which results in low probabilities of > 0.5".  In other words, there will be precipitation, but where, when, and how much is a very difficult forecast.  And that's before we even begin to talk about snow accumulations.  Snow levels will be flirting with the benches and valley floors through midnight and small changes in temperature could have a big impact on accumulations beneath the band.

The Post-Frontal Crapshoot

Source: http://www.consumertraveler.com

Some forecast situations are fairly predictable with the tools we have today, some aren't.  Today's computer forecast models do a fairly good job simulating cold-frontal passages over northern Utah.  Thus, while we have a beautiful day today in the Salt Lake Valley and Wasatch Mountains, we know that change is coming tonight and that the Cottonwoods will see a pretty good burst of snow tonight.  There are uncertainties with regards to the distribution, intensity, and amount of snow, but the basic idea that there will be a snowstorm tonight is fairly predictable.  That wasn't the case 10 years ago when forecast models were nowhere near as good as they are today.

In contrast, the post-frontal environment tomorrow is more of a crapshoot when it comes to snowfall.  There are a number of reasons for this:

1. Post-frontal precipitation is produced by shallow convective clouds that are small in scale and typically initiated over the Wasatch Mountains or, in some cases, the Great Salt Lake.  
2. Shallow convective clouds, whether initiated by flow over the Wasatch Mountains or the Great Salt Lake, are very sensitive to small changes in relative humidity and temperature.  Therefore,  small errors in the relative humidity and temperature of the incoming flow can greatly reduce forecast accuracy.  
3. Current computer models, even those run at 4-km grid spacing (e.g., the WRF model run locally by the National Weather Service), do not adequately resolve the processes responsible for these clouds.  Lower resolution models, like the GFS (~25-km grid spacing) and the operational NAM (12-km grid spacing) not only fail to resolve these clouds, but also fail to adequately resolve the Wasatch Mountains and Great Salt Lake.  
4. Often lake-surface temperatures (and salinity) are poorly initialized in computer forecast models.   Sometimes this is because the developers of those models don't make the effort to incorporate recent lake-surface temperatures, which can only be obtained via satellite.  Other times, it's simply because there hasn't been a recent cloud-free satellite overpass.   
5. Small particles known as aerosols, may play an important role in post-frontal precipitation efficiency.  At this time, such aerosol effects are poorly understood, pretty much entirely unobserved, and not considered at all by our computer models.  

In my view, this is an area ripe for immediate research.  In particular, would a dramatic increase in forecast model resolution to what we call cloud-permitting scales (horizontal grid spacings of 250 m or less) lead to forecast improvements?  Would computer model ensembles at such grid spacings lead to significant improvements in forecast skill?  Or, are uncertainties in terms of the land surface (e.g., the Great Salt Lake temperature), cloud microphysics (including aerosols), and the incoming large-scale flow too much to overcome?  

These are questions in need of answers, especially for those of us trying to figure out where to ski tour tomorrow.

Visualizing Forecast Uncertainty

Earlier posts this week have discussed the uncertainty in forecasts of front trough and snowband position for this evening.  Here's a nice loop showing a sequence of forecasts from the NAM beginning with the 84 hour forecast from 1800 UTC 22 Feb and ending with the 12 hour forecast from 1800 UTC 25 Feb (today), all valid for 0600 UTC 26 Feb (11 PM tonight).  Meteorologists sometimes refer to this as "DModel/Dt".  Notice how the position of frontal trough and precip band jumps back and forth, from as far south as central Utah to near the Utah-Idaho border.

Now, if you think that now we are dealing with a shorter lead time (24 hours or less), so presumably the forecast is more certain, think again.  The target area has narrowed, but for 1200 UTC 26 Feb (5 AM Saturday morning) the NAM puts the front right on top of Salt Lake City with heavy precipitation to the north, whereas the GFS puts the front roughly over Provo.

NAM Forecast Valid 1200 UTC (0500 MST) 26 Feb 2011

GFS Forecast Valid 1200 UTC (0500 MST) 26 Feb 2011

So, the frontal snowband remains, but the precise positioning remains uncertain.  It's going to be an interesting night.

Two Different Cyclones and Ongoing Forecast Challenges

Today provides an excellent opportunity to examine the differences between a western US "Intermountain" cyclone and an eastern US cyclone.

Radar composite with RUC2 sea level pressure (white contours)
and 850-mb temperature (red contours) at 1400 UTC 25 Feb 2011

In the eastern United States, the cyclone is an event.  Midlatitude cyclones are responsible for much of the cool-season precipitation in the east, including most of the major winter storms.  The precipitation shield accompanying today's cyclone is extensive, although those with a good eye may notice that to the south of the low center, precipitation is falling well ahead of the surface cold front.  This is not unusual in this part of the world and indeed overnight a very impressive pre-frontal squall line with numerous severe weather reports rumbled through the southeast.

In contrast, your Intermountain cyclone is a sub-synoptic-scale event.  The frontal structure is presently quite disorganized, although that will change during the day today.  Further, orography is presently playing a dominant role in the precipitation distribution, with heavy precipitation over northern California.  Over the Intermountain West, precipitation is trying to organize along a frontal trough that is developing over northern Nevada and Utah.  It is this trough and its placement has caused meteorologists, especially yours truly, much consternation the past couple of days.  Forecast models have placed the frontal trough and snowband anywhere from central Utah to the Utah-Idaho border.

The observations coming in this morning are consistent with a frontal band developing over northern Utah today during Intermountain cyclogenesis.  We'll see some rain or snow showers in the Salt Lake Valley, but the strong front and snowband eventually develop near the Utah-Idaho border this evening.

As discussed ad nauseam in previous posts, the predictability of the location of the frontal band has been very low.  Nevertheless, I think we can have some confidence now that this evening the real action will be to the north of the Salt Lake Valley, but the subsequent fate of the front remains uncertain.  I'm not making any wagers for late tonight and tomorrow.

My Clueless Forecast

Steenburgh's Intermountain Cyclone Forecast System
Photo: Wikipedia Commons

I'm just about ready to give up.  As discussed in the previous two posts, there's quite a bit of uncertainty with regards to the details of the weather forecast over the next couple of days.  We know there's an Intermountain cyclogenesis event on tap, but the forecast models can't seem to settle on a track or frontal position, especially for tomorrow afternoon and evening.

I was encouraged to see that the NAM and the GFS were converging on a solution this morning, but that doesn't necessarily mean they were converging on the right solution.  Indeed, the SREF showed that there was a lot of spread when one looked at all the available model solutions, so uncertainty remained.

So, perhaps I shouldn't be surprised when I pull up the 1800 UTC 24 Feb NAM and GFS and find that they have shoved the frontal trough and precipitation band for tomorrow afternoon back to the north compared to the 1200 UTC 24 Feb initialize runs.  Oh the humanity!

NAM forecast valid 5 PM MST 25 Feb

GFS forecast valid 5 PM MST 25 Feb

Like the roulette wheel in Vegas, you simply can't beat a chaotic system.  It is a fact of life that some patterns are more predictable than others and in a pattern like this it is simply inappropriate to issue a simple, deterministic forecast.  Snow is coming, but the specifics of where, when, and how much remain uncertain.  The bottom line is to be prepared, but stay tuned.

Intermountain Cyclogenesis and Forecast Uncertainty

As discussed yesterday, we're on track for an Intermountain cyclogenesis event tomorrow and Saturday.  Yesterday the NAM and the GFS forecast cyclone tracks and frontal positions for the event that were quite different.  Today, they have converged on a solution that moves the low center across southern and central Utah.

Over the next 24-36 hours, a surface front intensifies and slides eastward and southward into Nevada and northern Utah.  As shown by the 1200 UTC 24 Feb initialized NAM, the passage of this front through northern Utah (see the cyclonic wind shift in the image below) produces a good round of snow for the Salt Lake Valley and Wasatch mountains tomorrow afternoon.

Concurrently, cyclogenesis occurs over central Nevada, downstream of the High Sierra along the western portion of the frontal boundary.  

Although both the NAM and the GFS eventually move the cyclone across southern Utah, the real forecast problem for Friday night concerns the placement of the frontal trough tomorrow evening and night.  During this period, a band of snow will likely develop along and to the north of the trough.  This can be seen above and in the forecast analyses valid for 0300 and 0900 UTC 26 Feb (8 PM MST Friday and 2 AM MST Saturday).

Vigorous, slow-moving fronts of this type can produce substantial snowfalls in the valleys of northern Utah, but positioning is everything.  The precip band can be 50-100 km wide and a shift in north or south can influence whether or not Ogden, Salt Lake, Provo, or Nephi get the goods.

The National Centers for Environmental Prediction Short Range Ensemble Forecast system (SREF) provides forecasts from several models with different initial conditions.  The SREF can be used to help assess forecast uncertainty.  Meteorologists often will view the mean and standard deviation of a forecast variable to do this.  The 0900 UTC 24 Feb initialized SREF forecast valid 0300 UTC 26 Feb drapes the sea-level pressure trough over central Utah, perhaps a bit south of the NAM position.

Note that the standard deviation of sea level pressure is high across the Intermountain West, indicating that there is quite a bit of spread amongst the model forecasts.  In other words, the forecasts that make up the SREF vary in the intensity and position of the frontal trough.  For these forecast details, the atmosphere has low predictability for a forecast of 36-48 hour lead time.  Be prepared, but stay tuned.

Upcoming Intermountain Cyclogenesis Event

One of the more remarkable transitions that occurs in Utah climate during the spring is the increase in the frequency of Intermountain cyclogenesis.  As shown by Jeglum et al. (2010) using the ERA-Intermim, NARR, and NCEP/NCAR atmospheric reanalyses, the frequency of occurrence and genesis of Intermountain cyclones increases monotonically from a minimum in December or January to an absolute maximum in May.

Mean monthly Intermountain cyclone
frequency and genesis (Jeglum et al. 2010).

The intensity (or amplitude) of Intermountain cyclones also tends to be greater in the spring.

Two-dimensional histogram of peak 850-mb Intermountain
cyclone amplitude vs. month (Jeglum et al. 2010).

Not coincidentally, Shafer and Steenburgh (2008) found that the frequency of strong Intermountain cold frontal passages is also highest in the spring, although the peak is sharper and in June. 

Monthly frequency of strong cold frontal passages over
the Intermountain West (solid) and western United States
(dashed, Shafer and Steenburgh 2008)

If the models are on track, we have a great Intermountain cyclone event on tap for Friday and Saturday.  The 1800 UTC 23 Feb initialized NAM produces Intermountain cyclogenesis in the direct lee of the southern "High Sierra" for 0000 UTC 26 Feb (5 PM MST 25 Feb, Friday afternoon).  

The low center then tracks into northeast Nevada, with Salt Lake City in the so-called "warm sector" and strong southerly flow ahead of the low center at 0600 UTC 26 Feb (11 PM 25 Feb).  

As the low center moves into southwest Wyoming, the cold front rotates across Salt Lake City early Saturday morning.  

The NAM cyclone is as close to a "classical" frontal cyclone as you can get in the Intermountain West.  Although cyclogenesis occurs in the lee of the Sierra Nevada and orographic forcing is important, the low center forms along a pre-existing frontal boundary.  The amplitude of the frontal wave then increases, resulting in a "open wave" cyclone as the low center moves across northern Utah.  The Bergen School meteorologists responsible for the Norwegian Cyclone Model would be quite proud!

This is all fine and dandy, but a look at the GFS illustrates that there is great uncertainty in this case with regards to cyclone track.  In particular, the GFS forecast for 0600 UTC 26 Feb puts the low center in southern Nevada, with a surface trough and frontal zone draped across central Utah, well to the south of the NAM trough and frontal zone.  

In contrast to the southerly warm-sector flow predicted by the NAM, the GFS forecast puts Salt Lake in cold, post-frontal northerly flow for Friday night bar hopping!    Further, the GFS puts a band of what would be heavy snow across Utah Country, whereas the NAM frontal band is near the Utah-Idaho border.

This case provides a great example of how you can have confidence in a synoptic event, but differences in positioning make weather forecasts for specific locations very difficult.