It was a stormy Monday morning in Klamath Falls for a drive up to Crater Lake. As I scaled old Mount Mazama’s slope, the forest along Highway 62 grew thicker and the trees grew larger. Soon, at around 5,000 feet, the rain turned to sleet — and quickly snow. The road’s asphalt grew white, and snowbanks on either side of it rose higher.
The ranger at the park’s entrance station seemed a little confused to encounter a visitor like me during a winter storm — but her skeptical glance faded once I told her who I was meeting: The snow guys.
Soon, two government pickups pulled into the lot, and I met Chris Gebauer, a soil scientist for the Natural Resources Conservation Service, and Matthew Kritzer, a hydrologist for the Bureau of Reclamation. Just days after the start of the new year, the two were embarking on their first snow survey for the 2021 water year.
We strapped into snowshoes and climbed the snowbank behind the park headquarters. After trudging for 15 minutes through the powder, we stopped at a yellow diamond sign that read “Snow Course Marker,” and the two began to assemble a long metal pole. Kritzer balanced the pole on his shoulder and Gebauer followed behind, carrying a tape measure and a fanny pack of tools. As the unrelenting snow continued to fall, the two men got to work evaluating the snowpack.
Staying on (snow) course
Scientists have been conducting regular snowpack measurements across the west since the early 20th Century, when Dr. James Church postulated that monitoring snow accumulation in the winter could help predict how much water would be available during the dry summer months in the Tahoe Basin.
Various research groups in different states started their own snow monitoring programs during the 1920s, and the catastrophic dust bowls in the 1930s spurred the formation of the Soil Conservation Service, which eventually took over the practice and became the Natural Resources Conservation Service. Crews would venture out to established snow courses and take measurements once a month between January and May.
Many snow courses have been replaced by snow telemetry (SNOTEL) sites, which automate the collection and include other environmental data like temperature and soil moisture. They also take measurements more frequently, giving NRCS modelers a higher-resolution dataset with which to make water forecasts. But the park headquarters snow course is still in operation, providing some ground truth for the nearby SNOTEL at Annie Springs.
Each snow course involves a specific number of measurements taken at roughly the same location each month for consistency. Gebauer and Kritzer moved in an “L” shape through the trees, eventually making it into an open field covered in feet of untouched snow. Using the tape measure, Kritzer spaced each measurement 100 feet apart, moving to the left of our path, away from its packed snowshoe tracks.
“One of the hardest parts of the job” Kritzer began as he trudged along, trying to triangulate his location by remembering the trees that had been there last year, “…is remembering this course,” Gebauer finished.
Wielding the pole like Excalibur, Kritzer plunged it into the ground. After reading the depth etched into the metal, he twisted it slightly and pulled it back out. He then peered into slits on the side of the poll to measure the length of the snow core he had captured, relaying the measurements to Gebauer. Gebauer then scraped the traces of dirt collected at the bottom of the pole (it measures all the way to the ground, after all).
Finally, attaching a scale to the end of his snowshoe pole and balancing it on his shoulder like a brindle, Gebauer helped Kritzer weigh the core sample.
Ultimately, we were standing on more than 6 feet of snow. Not bad for January.
While snow depth might indicate how wet a winter is, it doesn’t properly correlate with how much water will be available the following spring and summer. That’s why NRCS researchers use snow water equivalent (SWE), which factors in the density of the snowpack as well as its depth.
“You could have 100 inches of snow, but the water content is going to differ depending on how it fell,” Kritzer said.
Because the scale is specially calibrated to the diameter of the snow pole, its value corresponds directly to the SWE of a given sample. Eventually, modelers average the values taken at each of the five spots to arrive at the SWE for the park headquarters snow course at large.
The snow guys invited me to take a measurement myself. After struggling to balance the more-than-10-foot rod, I awkwardly plunged it into the ground and measured a snow depth of 84 inches, subtracting an inch of dirt I had unwittingly captured at the bottom (Gebauer reassured me that minimizing dirt in a sample takes practice). The scale told me my sample contained 25.5 inches of water, meaning that a snow depth of 83 inches produced only about 30% of that in liquid form.
Gebauer told me that as winter progresses, SWE values tend to increase: The weight of additional precipitation compresses the accumulated snow below it, making it denser.
“The water content as a percent of the snowpack goes up,” he said.
I looked around the glistening meadow. Trees, clad in globs of white, backed up against a colorless sky. I could just barely make out the slope up to the caldera’s rim beyond, blurred by the downpour of snow. It was difficult to imagine that even a place like this, in the throes of winter, could be impacted by rising global temperatures. But annual snowfall recorded in the park has been decreasing for decades, as an uneven distribution of atmospheric energy and moisture has resulted in drier winters. According to data from the National Park Service, the average annual snowfall at the park’s headquarters in the 1930s was 614 inches—a whopping 51 feet. By the 2010s, average snowfall had reduced to 474 inches, a decrease of nearly 23%.
“This is as winter wonderland-y as it gets,” Gebauer said. But, especially over the past few decades, not all of that snow has made it to the Klamath Basin.
Tougher to forecast
The only trend researchers have observed in the snowpack that feeds the Klamath watershed is that water years are getting more difficult to predict. NRCS releases monthly basin forecast reports based on SWE data compiled from SNOTEL and snow course sites, which normally increase with confidence as the year progresses. Lately, though, that’s not always a given.
Scott Oviatt, snow survey supervisory hydrologist for NRCS Oregon, said snow accumulation on any given month during the winter is becoming less of an indicator of how much water will be available during the summer, particularly because of the timeframe within which it melts. He estimated that mountain snowpack is responsible for between 50 and 80% of water availability in the basin.
“The Klamath Basin is pretty dynamic,” Oviatt told me. “Especially over the last 20 years, we’ve found high variability in terms of snow accumulation and snowmelt runoff.”
Unseasonably warm winter days—more common due to climate change—can bring rain instead of snow to the mountains, melting off some of the accumulated precipitation. And early springs—also a symptom of warming global temperatures—can cause early snowmelt runoff into streams and storage areas (like Upper Klamath Lake), which lose more of that water to evaporation than if it had remained snow.
Gebauer recalled years where above-average snowfall during January, February and March suggested a high water year, but an unusual April runoff event set things back more than expected. Things can also be variable in a positive direction, with late spring storms bringing surprise snow to the mountains.
“If the snow melts off really fast and it all flushes out of the system, then we don’t have the long-term storage and slow release,” he said.
NRCS uses statistical models in their monthly forecast updates, which compare current conditions to a 30-year median curve of snowpack measurements from 1981 to 2010. Optimal water years will consistently see snowpack at or above that median, while low water years fall below the curve. The reports then model a set of streamflows based on that month’s snowpack and the precipitation in that water year so far, assigning a probability that each of those values will be exceeded.
The service’s January 1 summary for Oregon, which was just released this week, said snowpack in the Klamath Basin was 87% of normal levels, based on 53% at this time last year. Upper Klamath Lake is storing about 81% of the water it normally would in January. Based on current conditions, NRCS predicts that streamflow between March and September will range between 65% and 74% of normal levels.
The Crater Lake park headquarters snow course, reporting the data Gebauer and Kritzer collected earlier this week, fared among the best of all snowpack measurement sites in the basin. Its SWE of 24.5 clocked in at 119% of its historical median, compared to an SWE of less than half that this time last year.
Inflows to Upper Klamath Lake between March and September could be anywhere in the range of 119,000 acre-feet to 810,000 acre-feet, with a 95% chance flows will exceed that lower value and a 5% chance they’ll exceed that higher value. Water users can view that range of probabilities as a spectrum from dry to wet: The 95% flows would occur under drier-than-normal conditions, and the 5% flows would occur under wetter ones.
But that’s all based on conditions in January, meaning it’s too early to say whether water year 2021 will be better, worse or about the same as water year 2020. And with all the climate-related variations that could occur between now and April, Oviatt said it’s even getting harder to nail down streamflow probability later in the winter, as temperatures and precipitation are more erratic than they used to be.
“With the temperatures being widely variable and causing things we haven’t seen in the past, these statistical models don’t handle that necessarily well,” he said.
And while the probability paradigm theoretically allows water operators to readjust their allocations if conditions seem drier or wetter than normal, Oviatt said federal laws and management practices often bind those folks to the 50% probability value, which may not necessarily reflect reality if an early melt event occurs between two monthly reports.
Oviatt said NRCS is developing new modeling technology that’s more physically based rather than statistically based, and that adapts in real-time to changes in temperature and precipitation.
But the timeline on that is unknown. In the meantime, NRCS can only try their best to forecast with the available data and trust water managers in the basin to plan for the upcoming water year as conservatively as possible.
“All we can do is say, ‘Here’s the number, and good luck,’” Oviatt said.
Alex Schwartz is an environmental reporter for the Herald and News and a member of Report for America, a national service journalism corps. He can be contacted at firstname.lastname@example.org or at 541-885-4477.