When winter comes to Salient Predictions, speculating about the upcoming ski season is more than just a pastime: it's a data-driven exercise in predictive analytics. Here's how to plan your 2023-24 Northern Hemisphere ski season with the power of Salient's season-ahead forecast.

In this analysis, we'll focus on the Europe, North America, and Japan IKON/EPIC ski resorts.

We will combine 3 different forecasts from the 2023-11-15 model run: subseasonal (weeks 1-4), seasonal (months 2-3), and long-range (quarter 2). Snow accumulation depends on daily weather over time, so we'll temporally downscale these coarse probabilistic forecasts to an ensemble daily timeseries.

Note: Though Salient does not yet natively forecast snow and ice in the main product, this analysis shows a downstream process that converts our precipitation and temperature predictions into snowpack and runoff over time.

Results

Three regions stand out for snow quantity in 2023-2024: the Pacific Northwest, the Alps, and the Rockies. Salient is forecasting a winter that is warmer and (in some regions) slightly wetter than average. The key to success this season will be taking advantage of any extra precipitation without getting caught in the rain. The best strategy is to head for higher elevations, where the snow will be colder and drier.

Snow Water Equivalent (swe) is a measure of the amount of water contained in the snowpack. It correlates with snow base depth. A more full accounting of snow quality would take into account snow density, which advantages cooler locations with lighter, fluffier snow.

If you don't see your favorite mountain on this list, we combined some neighboring resorts like the Cottonwood Canyons, Aspen, and Killington/Pico since they performed similarly. We also masked resorts with below-average swe since artificial snowmaking is a key contributor at many lower-elevation resorts.

‍Conclusion - glühwein & fondue

The best hill is always the one that you're on. If you have a choice of mountains, the best bet this year is the Alps. Mountains like Chamonix, Andermatt-Sedrun, and Zermatt are best positioned to capitalize on this year's extra precipitation with cooler temperatures that maximize snowfall and minimize rain/melt. If you're in North America, Mammoth's balance may enable it to repeat its amazing 2022-23 season.

Wherever you make your turns, have a great season.

All code and data for this analysis is available to Salient customers in our examples repository.

Methodology - How we calculated this

We're going to look at 3 different ways to forecast the ski season: simple anomalies from historical averages, seasonal sums of snow/melt conditions, and a daily estimate of snow water equivalent accumulation/ablation.

1. Anomaly Analysis - Examining Differences from Historical Averages

Salient generates weekly forecasts natively in terms of difference from historical averages. The scatterplot below shows mean anomalies across all 51 ensembles for Dec 1 - April 30 at each ski resort. For the Northern Hemisphere 2023-24 ski season, Salient is predicting a slightly warmer than typical winter. New England is the most extreme example of the warming trend, with a +2°C temperature anomaly forecast. The Alps, Rockies, and Pacific Northwest predict a smaller +1°C positive temperature anomaly. Precipitation is forecast to be about average in most cases; the anomalies shown in the plot are small in magnitude compared to overall variation.

Of course, the weather doesn't care about seasonal-scale anomalies. To estimate snowfall, we need to look at precipitation and temperature on any given day so we know whether it's falling as rain or snow.

2. Seasonal Sums of Snow / Melt Conditions

A good ski year is one with lots of fluffy light snow and little rain or melting. At each location, we can estimate snowfall by summing precip when temperatures are below freezing. We can quantify melting by summing "melting degree days", the number of degrees above freezing each day.

Here we see 3 distinct profiles that are on the "efficient frontier" of snow volume and quality:

• Rockies (Aspen, Vail, Lake Louise) - cooler with very little melting or rain, but also drier
• Pacific Northwest (Whistler, Stevens Pass) - exceptional snowfall, but also melt/rain
• Alps (Chamonix, Zermatt, Andermatt-Sedrun) - good mix of high snow volume and quality
• Special mention - Mammoth combines great volume and quality, with a high precipitation baseline overcoming a dry forecast anomaly

A few caveats to this analysis:

• Temperature and precipitation type can vary throughout the day and even on different parts of the mountain
• This analysis doesn't consider the extra melting that rain causes
• This analysis doesn't consider artificial snowmaking
• Because the accumulation numbers begin in December, early-season snow isn't included

3. Daily Accumulation / Ablation

The analysis of "melting degree days" is expedient, but there are more sophisticated ways to model snow accumulation and melt. The University of Washington's Snow17 model (github, MIT license) takes into account excess melting from rain, mixed rain/snow in transitional temperatures, and the amount of liquid water held by the snowpack. It estimates how much precipitation falls as snow, how much snow melts during warm days, and how much extra snow is washed away by rain.

We can apply the snow17 model at all 51 ensemble members for each resort to get a more nuanced picture of accumulation and melt. This chart shows the mean of all 51 ensembles for each resort.

Note that snow water equivalent isn't the same as base depth, since snow density also plays a role. Snow density can vary from about 0.05 to 0.20 depending on the temperature at which the snow falls, plus increases from settling, packing, and melting.

3.1 Ensemble Variation at 4 Resorts

Focusing on our example locations, we see that the probabilistic nature of forecasting creates significant variation across ensembles. At Whistler, peak accumulation ranges from 400 to 1400 mm snow water equivalent.

The length of the season depends on when temperatures consistently rise above freezing. Whistler's ensembles peak in late March, with warmer temperatures magnified by extra rain. Copper and Andermatt-Sedrun continue to accumulate snow well into the Spring season.

3.2 Impact of Temp & Precipitation on Accumulation / Ablation

If we visualize precipitation and temperature from each day in each ensemble, we can see the co-occurrence of rain and melting. In the plot below, each dot is colored according to the degree of accumulation or ablation. Note that there are some blue dots above zero because the snow17 model allows for some mixed precipitation in a range around freezing. There are some warm days with little loss of snowpack because the base depth is already zero.

Copper's lower temperatures mean that it rarely experiences any above-freezing days. Andermatt-Sedrun is fortunate enough to receive little precipitation when temperatures are above freezing, so it does not experience extra rain-driven melt. Whistler's excellent precipitation is offset by some late-season rain, which explains its springtime drop in snow cover.

This analysis demonstrates how to feed Salient's Subseasonal-to-Seasonal weather forecasts into a deterministic timeseries model. You can do the same thing with wind generation, energy demand, or agricultural yield. Contact Salient to explore how probabilistic forecasts can improve your business.