Little Cottonwood Forecast Guidance

Last week we described a new HRRR-derived forecast product for Snowbasin.  So my friends in the central Wasatch don't feel neglected, we've also dropped a similar product for Little Cottonwood Canyon on weather.utah.edu.  It's available from a link along the top bar or directly at https://home.chpc.utah.edu/~steenburgh/ml/hrrrlccforecast.html.  It should update regularly, but we're not watching things all the time, so always check the dates before using.  

Like the new Snowbasin forecast guidance, the Little Cottonwood forecast guidance is based on the HRRR and uses machine learning to improve the prediction of several variables.  There's a graphical product (the examples below are the forecast from 0000 UTC 4 October and not todays):

and a tabular summary:

We provide wind forecasts for three locations: Mt. Baldy (AMB; 11,066 ft), Top of the Collins chair (ALT; 10,443 ft), and Cardiff Peak (IFF; 10,059 ft).  AMB is a wind-exposed location above Alta Ski Area and IFF the only upper-elevation wind-observing site on the north side of the canyon.  Each of these is important for avalanche forecasting and mitigation in the upper canyon.  We provide temperature forecasts for AMB, ALT, the Collins Snow Study Plot (CLN; 9662 ft), Base of Alta (ATH20; 8752 ft), and Elberts in the mid canyon (ELBUT; 7600 ft). All of the wind and temperture forecasts are based on machine learning algorithms applied to the HRRR.  

There is also the height of the 0°C wet-bulb temperature level, which is a rough estimate of the top of the melting layer (snow level is typically several hundred feet below this).  The precipitation fields, including snow-to-liquid ratio/water content, hourly and totally precipitation, and snowfall amounts, are for CLN.  Water equivalent precipitation is directly from the HRRR. Snow-to-liquid ratio/water content are based on a machine learning algorithm developed by Michael Pletcher, a graduate student in my group.  We apply these to the water equivalent precipitation to obtain the snowfall forecasts.  

The graphics are organized so that the precipitation related plots are plotted on the left and the temperature, RH, and wind plots are on the right.  From top to bottom, the precipitation-related plots include:

1: Hourly water equivalent (bars) and total water equivalent (black line) at CLN.
2: 700-mb temperature (purple line) and relative humidity (green line) over CLN.  These are useful for estimating cloud conditions at crest level.
3: Wet-bulb zero level (black line) and 1000 ft below the wet-bulb zero level (green line) with color-fill between.  This is based on a high-resolution profile over Salt Lake City and provides an approximation for the melting layer.  We use the upstream profile so we can have data down to the Salt Lake Valley and because the National Weather Service provides a very high resolution profile from the HRRR over the Salt Lake City airport.  Color fill is used to indicate the elevation of key locations in Little Cottonwood Canyon.
4. Hourly snow-to-liquid ratio (bars) at CLN with light blue used for non-precipitating period and dark blue for precipitating periods. Inclusion of non-precipitating periods allows one to have an estimated snow-to-liquid ratio during periods when the HRRR doesn't produce precip (but Mother Nature might). 
5. Hourly snow (bars) and total snowfall (black line) at CLN.

From top-to-bottom, the temperature, RH, and wind plots are:

1. Hourly temperature at AMB, ALT, CLN, ATH20, and ELBUT following the embedded legend.
2. Hourly RH at AMB, ALT, ATH20, and ELBUT  following the embedded legend (CLN has not had a RH sensor to enable training for this variable).
3. Hourly wind speed (red line), wind gusts (blue line), and wind direction (black dots) at AMB. Wind speed and gusts based on the left-hand y-axis.  Wind direction the right-hand y-axis.
4. Same as 3 except for ALT.
5. Same as 3 except for IFF.  Winds at this site will be weaker and more erratic, and that is evident in this plot.

The primary strengths of this product are its intentional design for Little Cottonwood Canyon and the use of machine learning to improve the HRRR forecasts, which are too low-resolution to adequately capture local effects.  The weaknesses are that its based on one model (the HRRR), rather than an ensemble, and we are not currently using any machine learning to improve the water equivalent forecasts (that could be done but I only have so much time in the day).  One, however, can also consult our experimental plumes from the RRFS Ensemble (https://weather.utah.edu/index.php?runcode=2025100606&t=rrfsqsf&d=PL&r=CLN) and Utah Snow Ensemble (https://weather.utah.edu/index.php?runcode=2025100600&t=ensgefsds&d=PL&r=CLN).  In doing this, one should recognize that the plumes are for the nearest grid point to CLN and that the RRFS and Utah Snow Ensemble have different terrain representations than the HRRR.  

Comments appreciated.  Criticism ignored.  Actually, that's not true, but like all of our online products, we do what we can with limited time.  Yes, we know the web page is not mobile-device friendly.