A Watershed Moment

More Than Half the World’s Water Comes From Mountainous Watersheds Now Impacted by Environmental Change

Extreme weather, wildfire, land-use change, early snowmelt, and other disturbances are significantly reshaping hydrology and biogeochemical interactions within mountainous watersheds all over the world. Interactions between a watershed’s soil, plants, microbes, fluids, and minerals govern the cycling of water, nitrogen, carbon, and other watershed constituents within a watershed, which support all terrestrial life on Earth.

Sixty to 90 percent of the world’s water resources originate from mountainous watersheds. Headwater catchments are also the genesis of biogeochemical cycling of nutrients, carbon, and metals. As such, water resource management, including accurate forecasting of water availability and water quality, depends upon scientists having a good understanding of how watersheds respond to factors like temperature and snowpack–now and in the future.

Determining this can be difficult even under a normal climate scenario, given the complex interactions taking place across scales and compartments of a watershed. Now that prolonged drought, wildfires, floods, and inconsistent snowpack mark what many have called a “new normal” state, we can no longer depend on past historical trends to project future watershed behavior, but instead need to develop approaches to predict how watersheds will respond to increasingly frequent disturbances.

The Biological and Environmental Research (BER) program at the U.S. Department of Energy (DOE) has approved more than $20 million to renew the Berkeley-Lab led Watershed Function Scientific Focus Area (SFA) project. As one of the most comprehensive watershed research projects in the world, the Watershed SFA team is poised to develop new conceptualizations and insights, as well as novel approaches for characterizing and predicting aggregated watershed hydro-biogeochemical responses to abrupt perturbations.

Did you know?

While watersheds are recognized as Earth's key functional unit for assessing and managing water resources, predicting their behavior remains a significant challenge.

“Despite the impact that watershed function can have on energy production, agriculture, water quality, and other important societal benefits, uncertainty associated with predicting watershed behavior remains high. We are striving to understand how complex, multi-scale interactions can lead to a cascade of downstream effects on water availability, nutrient and metal loading, and carbon cycling.”

Susan Hubbard

Watershed Function Project Lead

Taking the Long View

Susan Hubbard, Associate Laboratory Director for the Earth and Environmental Sciences Area at Berkeley Lab, leads the DOE Watershed Function SFA project. “Despite the impact that watershed function can have on energy production, agriculture, water quality, and other important societal benefits, uncertainty associated with predicting watershed behavior–and particularly watershed response to disturbances–remains high.”

“We are striving to understand and tractably predict how complex, multi-scale interactions can lead to a cascade of effects on downstream water availability, nutrient and metal loading, and carbon cycling. Our goal is to uncover new discoveries about mineral-soil-microbe-plant-fluid interactions, and combine those insights with tools that we are developing, such as 4D digital watershed characterization and scale-aware high performance modeling and machine learning tools,” says Hubbard. “With these advances, we can predict how exports from watershed subsystems aggregate to yield a cumulative downgradient signature of water discharge and water quality in the river.”

To advance the ambitious project goals, the team has turned the pristine mountainous East River watershed of the Colorado Upper Gunnison Basin into a technology-rich living field laboratory. Their research findings at this field lab of the future are relevant not just to the 40 million Americans who depend on water from the Colorado River, but to advancing understanding of how vital mountainous watersheds throughout the world are responding to environmental change, and to developing new approaches and technologies that can be potentially transferred to watersheds worldwide.

The researchers are taking a “system-of-systems” approach, which considers archetypal subsystems that are sufficiently representative to be studied in detail, and how subsystem exports aggregate to lead to cumulative watershed discharge and water quality. They have installed stations that provide distributed, synoptic, long-term measurements throughout the watershed and have developed several heavily instrumented ‘intensive’ watershed subsystem study sites. They have partnered with other scientists to establish a network of specialized ‘satellite’ study sites, which make available a series of remotely sensed layers of watershed bedrock, topography, vegetation, and snow. The East River Watershed is arguably the most characterized watershed in the world.

Concern Over Western Water Supply Is At All-Time High

For decades, hydrologists and resource managers could make relatively sound predictions about how watersheds would likely behave based on observations paired with information from fairly constant historic seasonal patterns of timing of snowmelt, depth of snowpack, precipitation, and temperature.

Today we are witnessing extreme fluctuations in these factors, many occurring over subseasonal timescales. In Colorado for example, 2019 was the first year without drought since the U.S. Drought Monitor began keeping records 19 years ago. In the last three years of instrumenting the East River Site, the team’s study site has experienced one of the largest snowpacks on record as well as one of the lowest snowpacks on record. Variability in snowpack and snowmelt timing is now the “new normal.”

The DOE Watershed Function SFA was developed to address the increasing uncertainty that exists around predicting watershed behavior, at a time when concerns over water availability and water quality run high, particularly in Colorado and other parts of the American West. While the researchers are poised to investigate a range of disturbance–including floods, droughts, and wildfires–much of their research to date has focused on the cumulative response of watersheds to changing patterns of snow accumulation and snowmelt timing necessary for optimizing management of water resources in this region.

The Biological and Environmental Interactions: From Hillslopes to the River Corridor

Already the DOE Watershed SFA research team has begun to cast light on the influence that these deviations from past trends can have on downgradient water quantity and quality. Snowmelt, for example, is occurring about a month earlier than it did a quarter century ago. This change in timing could impact the synchrony that currently exists between when snowmelt water is delivered to soil microbes and plants, and how those communities use and cycle water and nutrients. It could also impact the stage of a river, and associated biogeochemical-hydrological interactions that occur across terrestrial-aquatic interfaces.

Watershed SFA project researchers are exploring how natural variability as well as artificially induced snowmelt dynamics influence these complex interactions. For example, they have deployed black tarps to change the radiation and thus snowmelt timing relative to adjacent plots along an elevation gradient. They are monitoring vegetation, soil, and bedrock hydro-biogeochemical processes in accelerated and controlled plots to document the impact of snowmelt timing on exports.

Their measured observations and computer simulations show that the soil microbial community beneath snowpack contributes to an earlier and larger peak of nitrate export with early snowmelt compared to a more typical snowmelt scenario. Moreover, the study revealed that the water-table fluctuations associated with snowmelt infiltration led to biogeochemical reactions in the deeper Mancos Shale bedrock, which contributed a significantly larger pulse of bedrock nitrogen export than has been documented before.

The work depicts how different compartments of the watershed contribute differently to early snowmelt, and underscores the importance of understanding such response at both the local and aggregated scales The work has recently led to a ‘two nitrogen world’ hypothesis that is being tested in the new phase, focused on soil-microbe-plant interactions that contribute to a shallow nitrogen export and a deeper microbe-mineral bedrock interaction that contributes to a deeper export.

According to SFA team member Nick Bouskill, “Within mountainous ecosystems, the lack of available nitrogen can limit plant growth and photosynthesis, and microbial activity. Coarse soils, sparse vegetation, and strong hydrological events, such as snowmelt and monsoonal precipitation, can flush nitrogen prior to it being retained by plants and microbes. Predicting how nitrogen cycles throughout this ecosystem necessitates a greater understanding of these two nitrogen worlds, and the contributions that they make to productivity and downstream exports.

SFA team member Heidi Steltzer also describes how in both scenarios of early snowmelt and decreased snowpack, shrubs have replaced grasses and wildflowers as the dominant vegetation. Those grasses and wildflowers rapidly take up nitrogen and other elements from water within snowmelt. It’s not yet clear if these new plants can quickly assume the roles of their predecessors and prevent nitrates or other elements from entering the river and traveling downstream.

The researchers are also investigating how hydro-biogeochemical exchanges between river corridor elements (hyporheic zones, off-channel wetlands, and floodplains) alter riverine chemical signatures, as a function of river stage and river order. SFA team member Dipankar Dwivedi leads a team in using high-performance computing to predict flow and reactions through and out of river meanders.

“Development, application, and validation of advanced, high performance computational models that can predict hyporheic zone processes in high resolution have enabled us to explore how exports of carbon, iron, and other geochemical species vary–from a single meander to larger reaches, and across changing river stage conditions,” Dwivedi says. “An accurate model for exploring these hydro-biogeochemical interactions is important because the hyporheic zones play an outsized role in linking watershed hillslopes with rivers.”

The 4D Watershed

The Colorado River Basin extends across 250,000 square miles of varying ecosystems, including many different ecozones that host distinct plant communities sitting above diverse soils and bedrock. Innumerable interactions occur across various watershed compartments, governing watershed discharge and quality over vastly different time and space scales. Examples include rapid interactions between microbes and plant roots that influence nitrogen transport during snowmelt to decadal-scale groundwater flux tied to bedrock weathering.

Berkeley Lab research scientist Haruko Wainwright is investigating new approaches to rapidly characterize the organization of watersheds using extreme watershed data layers and machine learning.

“We now have an unparallelled watershed dataset complex at the East River, consisting of many 2D data layers that describe the lateral variability in bedrock, soil, elevation, snow, and plant properties,” Wainwright says. “We are using data-driven approaches to identify regions in the landscape that encompass a unique set of properties relative to neighboring regions that potentially influence the contribution that zone makes to the behavior of the larger watershed.”

These “functional” zones, or watershed “zipcodes,” are expected to have a characteristic response to a perturbation. One functional zone might be grass-dominated, north-facing hillside on volcanic bedrock, while another could be south-facing conifer forest on granite bedrock. These different zones are likely to respond uniquely to a perturbation such as early snowmelt or drought, and contribute a unique signature to the downgradient integrated water discharge and water-quality signature.

“This allows us to use increasingly available remote-sensing datasets to map out the location of zones, and then to install new autonomous above-and-below-ground intensive monitoring stations in different zones to provide information about how fluids and materials move through watershed subsystem compartments over the 4th dimension of time. The team is using the watershed zonation and associated property information to inform numerical models, and has also started to explore how to use remote sensing imagery of plants to autonomously detect subsurface biogeochemical activity.

Microbial ecologists Eoin Brodie and Romy Chakraborty lead SFA team members in testing the hypothesis that the presence and distribution of plants on the surface serves as an ‘easy-to-monitor’ biological sensor network that provides information about subsurface properties that are difficult and expensive to measure. “There are more than 22,000 watersheds in the U.S. alone,” Chakraborty says. “We simply cannot instrument every watershed across the planet, but this approach of relying on plants as sensors could open the door to using readily available and less expensive satellite spectral data to monitor changes in watershed behavior at much larger scales.”

The First Scale-Aware Watershed Model

Emerging high performance computing capabilities offer the possibility for scientists to precisely simulate a range of biogeochemical interactions taking place across a watershed–accurately, rapidly, and in unprecedented detail. For example, the DOE is already preparing to maximize the benefits of Exascale—future supercomputers that will be 50 to 100 times faster than our nation’s most powerful system today—for U.S. economic competitiveness, national security, and scientific discovery.

However, it may not be necessary to simulate everything everywhere. A key technical goal of the Watershed Function project is to develop models that can use the scientific insights and 4D digital watershed information developed through the project to predict how watersheds respond to perturbation over vast spatial scales and abrupt time scales, but in a tractable and computationally efficient manner.

Toward that end, the team is developing the first scale-adaptive watershed simulation capability which allows them to adequately quantify integrated watershed response to perturbation through process investigations and characterization performed in distinct watershed zones. This method uses variable resolution modeling approaches to “telescope” into “hot spot” regions that play an outsized role in watershed behavior.

When combined with machine learning and functional zone approaches, the new scale-adaptive capability is expected to provide a path forward for balancing accuracy and tractability in simulating watershed hydro-biogeochemistry. Paired with the Watershed SFA computational advances, the “ExaSheds” project represents the first systematic effort to advance powerful machine learning and Exascale computing to transform our ability to predict watershed behavior and increase the use of ever-larger and more-complex data obtained from watershed field observations.

Scale Adaptivity and Machine Learning to Transform Watershed Predictions

Scale-Adaptive Watershed modeling capabilities paired with machine learning approaches are expected to transform our ability to tractably yet accurately predict multi-scale watershed behavior.

“Combining machine learning with Exascale computing on the new GPU-based heterogeneous computer architectures will make it possible to replicate an entire river basin or catchment area in a computer simulation without losing finer details, such as the meander, or even a localized ‘biogeochemical hot spot’ of microbially mediated activity along a meander bend.”

Carl Steefel Senior Scientist and Modeling Lead, Watershed Function SFA

East River Community Watershed

In addition to advancing scientific objectives, the Watershed Function team is helping to build community, capacity, and a distributed watershed philosophy. The team and sponsoring DOE Subsurface Biogeochemistry Program describe many watershed prediction challenges as indivisible, meaning that they are of a nature that no one person, discipline, or organization can solve alone. Tackling such challenges requires a cultural shift toward distributed science, where scientists, scientific teams, and networks of teams are motivated to work together to advance and widely share codes, methods, data, tools, insights, and field sites. The Watershed SFA is committed to this cultural shift and to further developing the East River as a ‘Community Watershed’ that is optimized for catalyzing research opportunities and collaborations needed to address indivisible challenges.

Since the inception of the project, the Watershed Function project has hosted over 450 scientists to work at its Colorado Field Sites. Berkeley Lab Staff Scientist Ken Williams is the project Deputy Director, who directs on-site investigations taking place at the East River Watershed SFA research site. As the project enters a new phase, Williams said the team hopes to build upon the “community watershed” approach to watershed science they’ve developed at the East River site over the past years. In addition to the 66 scientists involved in the specific Watershed Function project, individuals from two additional national laboratories, four federal and state agencies, 27 universities, and four small businesses are collaborating with the project at the East River Watershed site, bringing their own questions and expertise to the grand challenge of watershed function predictability.

“We’re committed to answering the fundamental question: How do mountainous watersheds retain and release water, nutrients, carbon, and metals? But having now established the East River as a ‘community watershed’ providing a collaborative platform from which the scientific community can address this question, we have the unique opportunity to also develop novel approaches that leverage advanced computing and similar cutting-edge tools to the study of watersheds that can be applied to watersheds across the U.S. in other regions, or even other parts of the world.”

Ken Williams

Project Deputy, Watershed Function SFA

In Case you Missed These videos...

The Watershed Function Scientific Focus Area (SFA) program is useful for predicting how disturbances to mountainous watersheds impact the downstream delivery of water, nutrients, carbon, and metals.

A team at the Department of Energy’s Lawrence Berkeley National Laboratory (Berkeley Lab) studies changes to plant communities in a research area along the East River catchment in Colorado.

At the East River, Colorado field site, teams of scientists led by Berkeley Lab and supported by the Department of Energy are working to identify major inputs and exports of nitrogen in a pristine watershed system.

4D observations allow us to watch watershed behavior across scales and track the spatial variability of nutrients, metals, and contaminants from bedrock-to-canopy over time and under changing conditions.