Some Updates on weather.utah.edu

I'd like to share a few updates that we've made to products on weather.utah.edu since last winter. 

The Little Cottonwood forecast guidance derived from the HRRR (https://www.inscc.utah.edu/~steenburgh/ml/hrrrlccforecast.html) and GFS (https://www.inscc.utah.edu/~steenburgh/ml/gfslccforecast.html) has been updated with retrained machine learning algorithms that include last season in the training dataset.  An example from the GFS is below.  

We also added a temperature forecast for the Collins (9662 ft) observing site, which is a bit more representative of mid-mountain conditions than Mt. Baldy, which is still included as it is useful for ridge-top temperatures.

The new snow-to-liquid ratio (SLR) algorithm is based on a technique known as a random forest trained with observations from Alta-Collins.  It is more accurate and should produce a somewhat wider range of SLRs (up to about 25:1 if conditions are right) than the technique we used last year.  We thank Alta Ski Patrol and Alta Ski Area for continuing to provide access to weather observations and snow measurements so that we can produce products like this.  

We also updated the SLR algorithm used in our experimental HRRR snowfall products, and example of which is below.  

This also uses a random forest, but trained with data from 12 snow-safety sites in the western continental United States.  We thank everyone staffing those sites for providing data so we can develop products like this.  

I get a lot of requests for "point forecasts" from this product.  I'm trying to move away from autogenerating those.  Instead, you can do it dynamically for any point in the continental US at https://www.inscc.utah.edu/~steenburgh/hrrrqsfts/hrrrqsfts.html. This link is also available if you click on "Point Time Series" under HRRR-Snow (Experimental) in the left hand nav bar on weather.utah.edu.  Below is an example for Paradise Ranger Station on Mt. Rainier.  They will take a minute or two to pull up.  I recommend that you compare the elevation of the HRRR grid point (printed in the second graph) to your site as sometimes there are large differences that can be important for interpretation.  

We are also working on products from the Rapid Refresh Forecast System (RRFS) which the National Weather Service is currently testing and developing to replace the HRRR.  The RRFS is run at comparable grid spacing to the HRRR (3 km) and is currently run as a small ensemble with six members.  

Remember that these are experimental products and not official forecasts.  Feedback appreciated.