Snow Net: Joint Berkeley-Merced Snowpack Tracking Project’s Test by Fire

by Gordy Slack

California is experiencing its third dry winter in a row. Rainfall stands at levels far below normal and some reservoirs are nearly empty, but the real problem registers at higher altitudes in the Sierra Nevada where most of the state’s water wealth is stored as snow. The Sierra snowpack is at only about 17 percent of average level for this time of year. That is hard news to swallow for big water users across the state: farmers, town and city governments, hydro-power producers, and wildlife managers.

Worse, says the California Department of Water Resources, new climate models suggest that these dry conditions may be California’s new normal. Someday, desalination and other freshwater-generating technologies may provide new alternatives to the precipitation California relies on today, but for now the best that resource managers can do is maximize the efficient use of current water supplies.

Imprecise estimates of how much water the snowpack holds, however, make such efficiency planning difficult from the top of the system down. Currently, in the winter, the most remote Sierra snowpack sites are visited by humans only once a month and measurements taken on these trips are very coarse, providing only approximate data about how much frozen water is accumulating and will flow to reservoirs in the spring and summer.


These NeoMotes, data loggers fitted with transceivers, were attached to sensors in each snow-monitoring station. Designed for this project by Professor Steve Glaser and his former student Branko Kerkez, the motes are now commercially manufactured by Metronome Systems in San Jose.

A joint project between UC Merced and UC Berkeley attempts to radically improve the quality and availability of data about the amount of water stored in mountainous parts of the vast American River Basin, which runs from the Sierra Nevada to Sacramento. The snow at the top of this watershed is one of California’s primary freshwater storage sources. The CITRIS-supported project, the Development of a Basin-Scale Water-Balance Instrument Cluster for Hydrologic, Atmospheric and Ecosystem Science, aims to provide real-time, high-resolution data about how much water is stored in and being released from the Sierra snow pack. Armed with that knowledge, water managers can begin to plan allocations for the spring and summer even as the winter snows accumulate.

The project, initially profiled in a 2011 CITRIS Signal feature, took a major step forward this summer when a team of faculty members and students traveled into the Sierra high country and installed ten sensor networks, each composed of ten or more stations that measure snow depth, temperature, and humidity.

David Rheinheimer, a post-doc at UC Merced and an expert on the economics of water management, has just begun an economic analysis of how the sensor network could best be employed to help water managers. He plans to study the implications of high-temporal-resolution stream-flow data for maximizing the efficiency of coordinating hydropower plant operations with other renewable electricity sources. Hydropower plants, some of which have very little storage capacity, can kick in when power sources like wind and solar wane due to weather conditions.

“With better and more information about what’s coming into your reservoirs—especially the smaller reservoirs—you can much more efficiently operate your hydropower system,” says Rheinheimer. “I suspect there is a high value to the electricity grid of having better forecast information, which is what these sensor networks are all about.”

Several more networks will be installed next summer and other sensors measuring solar radiation, soil moisture, and sap flow will be added. Because the networks are based on super-low-energy, wireless communications technology and can send data back to Berkeley in near real time, they could make high-value data from remote areas both relatively inexpensive and readily available. But for the system to gain widespread acceptance among the famously conservative (aka risk-averse) community of water managers, “it must also prove to be both reliable and robust,” says Ziran Zhang, the UC Berkeley graduate student coordinating installation of the instruments.

“Reliable and robust. Those two words are repeated again and again,” says Zhang. “Data must be accurate when they come and we have to be able to count on them under all different conditions.” The system’s stations had been in place for only a couple of months when they underwent trial by fire. The American Fire in August burned nearly 30,000 acres in and adjacent to the area containing two of the networks. Fire crews pulled out all but one of the threatened sensor stations at the Duncan Peak site. The one left behind was destroyed by the fire and another replacement was crushed by a falling tree later in the season.

Curious animals have also been a challenge. Bears are attracted to the wires connecting the sensors to the transmitters and one chewed on a battery after carefully opening the enclosure, says Zhang. In addition to tracking the snowpack, each node on the network also monitors its own health and, as long as only a limited number of nodes malfunction simultaneously, the system can work around a broken network, too. Each site communicates wirelessly with multiple others, so if one connection is lost, data will still find their way home to Berkeley. When a node goes down, that fact is automatically transmitted to the researchers, so they can send someone out to repair it promptly, says Zhang. The rough wilderness terrain raises several challenges for the researchers. Radical altitude changes, rock outcroppings, and irregular vegetative cover all complicate the task of gauging the strength of the wireless signals, says UC Berkeley engineering professor Steven Glaser, co-PI on the project. “The network can be working perfectly in the lab, but when you install something like this in the wild, there will be surprises.”

Roger Bales stresses the importance of properly locating each station so that it can both communicate with the others and collect representative data. “We will be taking lessons learned this year back into the field with us next summer,” says Bales, co-PI on the project and engineering professor at UC Merced. In addition to addressing basin-scale questions about how much water is stored in the snow and how much it is releasing at any given time, other more locale-specific questions may be answered by the sensor networks as well, says Bales. For example, the project is examining how much water is drawn up and held by vegetation in different areas. “The trees are like giant pumps that pull water up into their trunks,” says Glaser.

At one experimental site near Shaver Lake, plant transpiration “pulled out” more than half of the water, he says. In the much larger American River Basin study, area research on the influence of vegetation on water availability is still ongoing, says Bales, but preliminary results suggest that trees are trapping a significant amount of water. “It is our hypothesis that removal of one-third to one-half of the biomass in a forested area, such as is proposed to restore forests to historical conditions and reduce the risk of catastrophic wildfire, will result in a 5 – 15 percent gain in runoff, averaged over many years,” says Bales.

Due to decades of fire suppression practices in the watershed, much of its forests are more densely wooded than old-growth would have been hundreds of years ago. The new data may add extra incentive for land managers to reduce vegetation in some parts of the American River Basin. In some areas, fire suppression has resulted in a near doubling of the biomass over the past century, says Bales. The extra trees, compared to 10-150 years ago, don’t just take water out of circulation; they also constitute dangerous amounts of fuel for infrequent but potentially catastrophic wildfires.

This summer, both the Rim Fire and American Fire were so destructive because burning had been suppressed in those regions over the past century. Warmer temperatures and dry fuel conditions are also factors. There is just too much fuel. Before the forests were managed, small fires would periodically burn through and reduce the amount of younger, smaller vegetation, leaving the big older trees intact. Re-introducing controlled burns, or thinning the trees, may well both restore the forest ecosystems and boost badly needed water supplies, says Bales.

“We will know a lot more when our instruments have had a full season or two in the field,” says Zhang. Right now he and the research team wait impatiently for the snow to fall and the first winter’s data to flow.