Snow at lower elevations always melts first… or does it?
Synchronous Snowmelt and Streamflow in the Sierra
by Jessica Lundquist, Doctoral Candidate
Hydroclimatology Group, Scripps Institution of Oceanography

Author Jessica Lundquist contemplates how snow melts at tarns near Tioga Pass.

My first summer living in the Sierra was 1995, the year the Tioga Road (Highway 120) didn’t open until the fourth of July.  I had a job serving hamburgers at the Tuolumne Meadows Grill, but my main goal was to hike as many miles and see as many vistas as possible.  Where to begin?

“The trail up to Gaylor Lakes is still covered in snow,” explained the ranger in the visitor center, “and the Rafferty Creek trail up to Vogelsang is closed for the same reason.  You’d better head downstream, toward Glen Aulin and lower elevations.”  The meadows, covered with a lake of water, glistened in the July morning sun, while the surrounding peaks gleamed white with thick blankets of snow.  I shouldered my pack and headed downstream, to the lower altitudes where the snow melted first. 

After a summer of hiking the high country and watching the snow gradually disappear and be replaced by wildflowers, my heart echoed John Muir’s statement in My First Summer in the Sierra, “For my part, I should like to stay here all my life or even all eternity.”  Six years later, a graduate student in oceanography, the call to the High Sierra had not diminished.  Remembering my hikes and explorations, I proposed to study snowmelt and streamflow, but as a woman on the plane asked me a couple months ago, “Why would anyone care about that?”  I had more than a few words to say to her!

In 1888, explorer John Wesley Powell reported that the lack of water in Western North America was a serious obstacle to unbridled settlement. However, in the century that followed, dams, reservoirs, and aqueducts allowed cities and agriculture to flourish.  In the West, water is power, and over half of the water supply is derived from mountain snowmelt where the snow provides a natural reservoir, delaying runoff and providing water in the spring and summer when it is needed most.  However, while the population continues to grow, an alarming change has been noted in recent decades in the western U.S. late season runoff. The initiation of spring melt has come progressively earlier in the season, and runoff from spring and summer snowmelt has declined markedly (Cayan et al. 2001). In the long run, it is estimated that, in response to projected global warming of 3 °C, the spring-summer snowmelt will be diminished by about one-third (Roos 1987).  These studies indicate that winter floods will increase, and less water will be available in the summer, when demands from both humans and ecosystems are high.  Thus, understanding snowmelt processes from determining the timing, magnitude, and spatial variability of snowmelt runoff, to better understanding climatic change, has become crucially important.

At a park research workshop in spring 2001, Yosemite National Park was identified as having a special role in the earth sciences as a locus for studies of the responses of natural systems to global and regional climate change, and scientists from the Park, the California Department of Water Resources (CDWR), United States Geological Survey (USGS), Desert Research Institute (DRI), and Scripps Institution of Oceanography (SIO) were quick to collaborate with each other and rise to the challenge.  Fortunately, my thesis topic fell into the right place at the right time, and I helped design an instrument network to monitor how snow melts at different locations and then moves through the river system.

By August 2001, I was back in Tuolumne Meadows.  This time, instead of serving hamburgers to hungry campers in a sharp-looking red-flannel shirt, I was wearing patched, frayed shorts and wheeling a metal cart full of cake-pan-shaped concrete anchors up the trail to Rafferty Creek.  Backpackers, equipped with high-tech, low-weight gear, stared at the cart with appalled faces.  “Are you sure you have enough stuff?” one man asked, warily.  “Oh yes,” quipped my advisor, Dan Cayan (SIO and USGS climatologist), who carried a shovel and a coil of metal cable. “We never go out in the wilderness without all the essentials.”

Figure 1 The Levelogger makes hourly measurements of pressure (stream depth) and temperature.

At the Rafferty Creek Bridge, we stopped. The creek was completely dry, but we wanted to measure the return of water when the snowmelt cycle began again next spring.  The research team clamored under the bridge, and Mike Dettinger (USGS hydrologist) began attaching cable to a strut under the bridge, while Larry Riddle (SIO meteorologist) pulled out a Solinst Levelogger (figure 1). The Levelogger, smaller than the palm of my hand, is placed in the bottom of the stream and measures temperature and pressure every hour.  Most of the pressure comes from the weight of the overlying water, and the pressure measurement is used to determine the depth of the stream.  Atmospheric pressure variations are measured with a barometer and subtracted from the Levelogger measurement.  To keep the lightweight Levelogger from moving in the streambed, we attached it to a PVC pipe inside a concrete weight, with holes drilled in the top to allow pressure to equilibrate with the stream (figure 2).  For extra security, Mike clamped a cable from the concrete weight to the bridge to assure that even if high flows moved the weight, it would not go far (figure 3).

Figure 2 Jessica Lundquist drills holes through the concrete and pvc pipe that will hold a Levelogger below the Rafferty Creek bridge.  In the background, the metal cart full of equipment waits to be wheeled down the trail.	Figure 3 Mike Dettinger attaches cable to the concrete weight to make sure it won't be swept away by high flows.

Figure 4 In June 2002, Mike Dettinger is impressed that the Levelogger's latex freezing-protection shield has survived the winter.

“Looks pretty good,” surmised Dan as Julia Dettinger (Mike’s daughter and an able volunteer) and I moved rocks to conceal the concrete weight and cable, “but there’s no water here now, and this spot could freeze solid before the winter snow comes.”  If water around the Levelogger freezes solid, the expanding ice could seriously damage the instrument.  “I think this calls for freezing protection,” Larry announced as he began to pour 100-proof alcohol into a latex condom.  He inserted the Levelogger into the alcohol-filled condom and sealed the entire contraption (figure 4), carefully placing it into the PVC pipe. Because alcohol freezes at a much lower temperature than water, it should remain liquid and protect the instrument from ice.  At the same time, the liquid and latex would be flexible enough to allow the instrument to sense small pressure (stream depth) variations.  Still, Larry worried about what the business department at Scripps would say when they saw his receipts and reimbursement request for this important field equipment.

Figure 5 Mike and Julia Dettinger make measurements of water depth and flow velocity across the Dana Fork of the Tuolumne River.

With the instrument secured and protected from freezing water, we had one step left: to record the actual river flow at the measurement site.  Flow is measured as the volume (cubic feet or cubic meters) of water moving past the instrument each second.  The water depth, or stream height, measured by our Leveloggers can be used to infer the flow after measuring both the depth and the flow during a number of different flow rates and establishing a rating curve (which relates one measurement to the other).  Today, at Rafferty Creek, the task was easy.  I wrote a big “zero” in my notebook. Yesterday, Mike and Julia had waded across the Dana Fork of the Tuolumne River with a tape measure, carefully recording the distance across and the depth at increments across the river’s width.  At each increment, Mike inserted a flowmeter into the stream (figure 5). As the ring of cups spun with the current, Mike recorded the number of turns per minute and calculated the water velocity.  Once the cross-sectional area and the velocity are known, the flow is easily calculated by multiplying the two.

Figure 6 Elevation vs. distance downstream from Parker Peak, showing the range of elevations monitored by Tuolumne River gages.

By the end of the summer, we had ten Leveloggers installed in the Tuolumne River Basin, with elevations varying from 4,000 ft (about 1200 m) at the head of Hetch Hetchy reservoir to over 9,000 ft (about 2740 m) at Gaylor Creek (figure 6).  Now, praying for our instruments to work, we just had to wait to see where the streams would rise first and what role elevation would play.

On average, temperature decreases 6.5C per km elevation gain (3.6F/1,000 ft) in the troposphere, the lowest layer of the atmosphere. At the surface, the sun warms the mountaintops, and the average temperature difference is less, about 4-5C/km (2-3F/1,000 ft) in the Sierra. A dry parcel of rising air (such as wind forced up and over the Sierra during a winter storm) cools even more with elevation (9.5C/km or 5.2F/1,000 ft). This occurs because as the parcel rises, less of the overlying atmosphere weighs down on it, and it experiences less pressure. Under less pressure, the air expands, and air molecules come into contact with each other less frequently, resulting in less rapid motion and cooler temperatures. Cooler air can hold less moisture than warm air, so this process also increases precipitation with altitude. This is why, in eneral, higher elevations receive more snow than lower elevations each winter. These temperature differences also explain why we expected snow at lower elevations to start melting first. [For more information on temperature, moisture, and rising and sinking air, see Adiabatic Processes and Lapse Rates.]

Figure 7 In 2002, every subbasin of the Tuolumne River, regardless of elevation, started to rise on March 29th, the 88th day of the year.  Flow is normalized by basin area, to show how many mm of snow water depth would be melting on average everywhere in the basin.

Synchronous Snowmelt
In June of 2002, we were ecstatic.  Every single instrument had survived the winter and recorded a story of snowmelt and streamflow.  I hurried to process the data, using our very limited rating curves (more measurements will make these more accurate over time) to estimate streamflow throughout the Tuolumne Basin.  As I plotted one record after another (figure 7), I furrowed my brow.  They all looked the same!  At the beginning of the 2002 snowmelt season, all of the streams, regardless of elevation, began to rise on March 29^th.  It seemed that the snow had suddenly started melting everywhere at once.  Was 2002 an odd year, or is this usually the case in the Sierra, only never measured before?

Fortunately, the CDWR has been measuring the snowpack for years, and snow pillows (which weigh the snow on top of them, see https://sierranaturenotes.yosemite.ca.us/SnowSurvey.htm ) have been reporting the amount of water contained in the snow each day, at sites ranging between 5,000 and 12,000 feet elevation.   A large number of stations have been operational for over a decade, and some sites date back to the early 1970s.  Armed with data from 44 snow pillows in the central Sierra from 1992 to 2002, I asked the question, when does the snow stop accumulating and start melting at each snow pillow?  In other words, when does spring begin?

Figure 8  Blue circles (corresponding to elevation in feet on the left axis) show the day that spring snowmelt began at each snow pillow.  The red line (corresponding to the right axis) shows streamflow along the Merced River at Happy Isles (in cubic meters per second) for the same time period.  The Merced River, like the Tuolumne, rose suddenly on March 29th, when the snow started melting at all elevations.

Figure 9  On average, the lower elevation snowpillows (red dots, elevation in feet on left axis) start melting earlier in the season, as the temperature (blue line, right axis) rises.

For 2002, the results were striking.  With very few exceptions, every snow pillow recorded the start of spring melt on March 29^th, the same day that all our streams rose (figure 8).  In most years (7 out of the 11 years initially examined) and on average (figure 9), snow started melting at lower elevations first, but the onsets of spring in 1998, 1999, and 2000 were sudden and synchronous, just like 2002.  What caused this phenomenon?  Could we forecast it in future years?  A sudden onset of spring, where all elevations start melting synchronously, causes sharp rises in rivers across the state, and understanding when this might occur would be useful for reservoir, hydropower, and fisheries management operations.  The sudden appearance of these simultaneous onsets in the last several years of the records raised ominous questions about the role of climate regimes and changes as well.

Figure 10 Mean 850 mb (about 1500 m or 5000 ft) temperature for the three weeks prior to snowmelt initiation for synchronous years.

Figure 11 Mean 850 mb (about 1500 m or 5000 ft) temperature for the three weeks prior to snowmelt initiation for nonsynchronous years.

What was the weather like leading up to these synchronous springs?  The National Oceanic Atmospheric Administration (NOAA) Climate Diagnostics Center (http://www.cdc.noaa.gov) operates online tools that provide plots of daily averages of meteorological data at several levels in the atmosphere from 1948 to the present.  To determine the weather patterns affecting the Sierras, I looked at the average temperatures at the 850-mb level during the three weeks prior to the start of spring snowmelt for groups of synchronous and nonsynchronous years. [Millibars (mb) are a measure of atmospheric pressure, which decreases with height; 850-mb corresponds to about 1500 m or 5000 ft in elevation.]   During the three-week periods, synchronous years have more cold winter storms than usual (figure 10).  When these late spring storms clear, the atmospheric circulations rebound like a spring, and temperatures rise dramatically all across the Sierra. In contrast, during nonsynchronous years, the three weeks prior to the onset of spring did not produce many storms (figure 11).  These periods of  clear skies allowed the snowpacks to ripen (that is, warm to 0°C throughout and become denser so that the addition of further heat causes melting) gradually.  Thus, in the nonsynchronous years, the temperature rise is less dramatic, and lower elevations warm above freezing sooner and melt first (figure 12).

Figure 12 In synchronous years (b), the average temperature rise on the day of snowmelt initiation (marked by a 0 on the lower axis) is about 10°C (compared to a smaller rise of about 6°C on nonsynchronous years) because late spring storms keep the temperature depressed until warm, dry air moves into California and forces snow to start melting everywhere at once.

This is an exciting area of research, and much work remains to be done. Will 2003 be a synchronous spring?  Studies by my advisors (Peterson et al. 2000, Cayan et al. 2001) have shown that large rivers all over the west typically rise and fall together each spring, within a few days of each other.  Is this how widespread the synchroneity of my “synchronous springs” is?  What happens during the second half of the melt season, as snow runs out?  Higher elevations accumulate larger snowpacks in the winter, so even during synchronous melt years, snow should disappear from lower elevations first.  Right now, sitting in sunny San Diego, I can hardly wait for summer to return to the mountains so that I can retrieve my instruments and learn what other stories the streams have to tell.

Acknowledgements:
This research is funded by a Canon National Parks Scholarship, and by the California Institute for Telecommunications and Information Technology (Cal-(IT)²), NOAA, NSF ROADNet, and the California Energy Commission.  Special thanks to Patricia Lundquist and Michael Dettinger for edits and comments. 

References:
Cayan, D. R., S. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D. H. Peterson, 2001. Changes in the onset of spring in the Western United States. Bull. Am. Met. Soc., 82, 399-415.

Peterson, D. H., R. E. Smith, M. D. Dettinger, D. R. Cayan, and L. Riddle, 2000: An organized signal in snowmelt runoff in the western United States. J. Am. Water Resour. Ass., 36, 421-432.

Roos, M., 1987. Possible changes in California snowmelt runoff patterns. Proceedings of the 4^th Annual PACLIM Workshop, Pacific Grove, CA.