The average global temperature is rising. In 2015, the contiguous United States (CONUS) average temperature was 2.4°F above the 20th century average. In 2016 it was 2.8°F above the the avg. In 2017 it was 2.6°F above the average. What does this mean for snowfall in the Pacific Northwest? We can look at historical weather data and do math.
The northwest avalanche center has 48 weather stations operating in Washington and the northern part of Oregon. Each of these stations has a temperature sensor and 21 have a precipitation can. Most of the stations without a precipitation can are close to one with it. It's a simple matching exercise to add precipitation to the 27 stations that need it. Here's a table with each temperature sensor and how close the nearest precipitation can is:
|Weather Station||Elevation||Distance to Precip measurment|
|Berne Snow Camp||2700||0.00km|
|Chinook Pass Base||5500||0.00km|
|Chinook Pass Summit||6240||0.82km|
|Crystal Green Valley||6230||1.83km|
|Dirty Face Summit||5980||7.37km|
|Mission Ridge Mid-Mountain||5160||0.00km|
|Mission Ridge Summit||6730||1.75km|
|Mt Baker - Heather Meadows||4210||0.00km|
|Mt Baker - Pan Dome||5020||1.11km|
|Mt Hood Meadows Base||5380||0.00km|
|Mt Hood Meadows Blue||6540||1.31km|
|Mt Hood Meadows Cascade Express||7300||2.21km|
|Mt St Helens - Coldwater||3260||0.00km|
|Ski Bowl Base||3660||0.00km|
|Ski Bowl Summit||5010||1.67km|
|Snoqualmie Pass - Dodge Ridge||3760||1.14km|
|Snoqualmie Pass - East Shed||3770||8.52km|
|Stevens Pass - Brooks||4800||0.00km|
|Stevens Pass - Brooks Wind||4850||0.08km|
|Stevens Pass - Grace Lakes||4800||0.81km|
|Stevens Pass - Schmidt Haus||3950||0.00km|
|Stevens Pass - Skyline||5250||0.40km|
|Stevens Pass - Tye Mill||5180||1.70km|
|Timberline Magic Mile||6990||1.71km|
|Washington Pass Base||5450||0.00km|
|Washington Pass Upper||6680||0.93km|
|White Pass Base||4470||1.87km|
|White Pass Pigtail Peak||5970||0.36km|
|White Pass Upper||5800||0.00km|
These stations produce data every hour, and there are three seasons worth of data readily available. That creates 941,370 weather station-hours to work from. We can estimate the snowfall at each station by doing some simple math on the temperature and precipitation. If it's 25°F out, and there is .1 inch of precipitation (water) measured, there will be about 1 inch of snow. If it's 35°F out, it's raining. Using this logic I created a simple linear scheme to calculate the snow density off of temperature. In this manner we can build hourly snowfall for each weather station. I then subtracted the annual climate temperature anomaly from each temperature sensor and redid the math. Difference between the two snow calculations is the change the temperature anomaly would make. Here's an example from Crystal during February 2017. Blue is the actual snowfall. Orange is what is estimated to fall had it been 2.6°F cooler. Small difference in snow density during colder temperatures, big difference when we are near the rain/snow threshold.
What's it all add up to? The below is the snow loss in inches due to global warming by weather station.
Here is the same chart but expressed as a percent loss compared to the 20th century average temperature:
Let's step it forward. Here are projections from 8 different climate models.
By 2050, we are looking at just over an additional 1°C of warming, or 2.2°F. Here's the decrease in snowfall as a % of what would have been expected given 20th century climatological norms:
This puts many of the ski areas in the Northwest out of business. For reference the Paris Climate Accord is attempting to limit our warming to an additional 1C. This is the best case scenario. In a high emissions scenario, the generally accepted warming in the 21st century would be 2.6 to 4.8 °C (4.7 to 8.6 °F). Here's the percent of snow loss at the high end of the scale (8.6 °F):
There's a lot of assumptions with the above, accurately modeling the future state of the climate,especially in a regional manner is an insanely complex challenge. I took a very simple approach, but I think most of the sources of error cancel each other out, as most are on either end of the transition from rain to snow. One area I don't have good data is at Stevens and Snoqualmie Pass. East flow creates difficulties in predicting snow vs rain due to freezing rain being a possibility. I could flag freezing rain by mathing out the snowfall being measured vs the temperature and precipitation, however you would not be able to effectively model a freezing rain prediction for the warming scenario using the same methods. For the time being I just left it as is, this will inflate snowfall totals for those areas. There are also a number of sensor errors and of the sort in this data, I'm doing some basic cleaning but not getting into the weeds. I would be interested in working with anyone that has ideas on calculating snowfall density using only temperature and precipitation at the surface, plus possibly RH.