Subsurface Hydrology Data Integration and Sensing Methods

Knowledge Graph (193 nodes, 1309 connections)

Research Primer

Background

Mountain watersheds like those surrounding the Rocky Mountain Biological Laboratory store, route, and release water through a complex plumbing system that extends from the snowpack on the surface down through soils, fractured bedrock, and deeper groundwater. Understanding this plumbing is essential for predicting streamflow, sustaining ecosystems, and managing water resources in the Gunnison Basin, where downstream users depend on snowmelt generated at high elevations. Because much of this system is hidden underground or spread across rugged terrain, scientists increasingly rely on integrated sensing methods — combining instruments mounted on aircraft, installed in the ground, and carried on foot — to build a coherent picture of what is happening across scales.

Several core concepts run through this area of research. Ground-penetrating radar (GPR) is a geophysical technique that sends electromagnetic pulses into the snow or ground and measures how long they take to bounce back, allowing researchers to infer properties like snow density or subsurface layering. Remote sensing integration refers to the practice of fusing measurements from different platforms — for example, combining airborne lidar (which maps surface elevation and snow depth with laser pulses) with GPR travel-time data to produce maps that neither tool could generate alone. Groundwater storage dynamics describe how water volumes held beneath the surface rise and fall with snowmelt, rain, and drought, while fracture density in bedrock controls how quickly that water can move. Porewater chemistry — the chemical signature of water in soil and rock pore spaces — reveals where water has been and what it has dissolved along the way.

Tying these measurements together requires careful data harmonization and synthesis: different sensors report in different units, on different grids, and at different times. Quality assurance procedures and time-series data management are therefore as critical to discovery as the sensors themselves. In the Gunnison Basin, where snowpack is the dominant water input and where vegetation, geology, and topography interact in complex ways, this integrated approach is the foundation for understanding how a changing climate will alter water delivery to streams, soils, and ecosystems.

Foundational work

Early site-based studies in the East River watershed helped establish that surface patterns observable above ground carry important information about what is happening below. Berman's (Berman, 2011) survey of vegetation along a southwest-facing hillslope of the upper East River showed that plant community distributions are strongly tied to underlying lithology: aspen stands correlated with shale units while conifers were more commonly associated with sandstone, and dry meadows dominated across all three lithologic classes examined. This finding reinforced the idea — central to subsurface hydrology in the Gunnison Basin — that bedrock type exerts a quiet but powerful control on water availability, rooting depth, and ecosystem structure, and that surface observations can serve as a first-order guide to subsurface conditions.

Key findings

A central thread running through recent work is that no single instrument can capture the full picture of mountain water storage; the most reliable results come from combining sensors. Meehan et al. (Meehan et al., 2024) demonstrated this by merging airborne lidar snow depth retrievals with ground-penetrating radar travel-time observations at Grand Mesa, Colorado, during the NASA SnowEx 2020 campaign. Using an automated layer-picking routine and machine learning to distribute density estimates across the landscape, they produced spatially continuous maps of snow depth, bulk density, and snow water equivalent with root-mean-square errors of 11 cm for depth, 27 kg per cubic meter for density, and 46 mm for snow water equivalent — a major step toward solving the long-standing problem that airborne lidar can measure how deep the snow is but not how much water it contains.

Forest structure emerged as a critical modulator of these patterns. Hojatimalekshah et al. (Hojatimalekshah et al., 2023) applied a convolutional neural network to airborne lidar at Grand Mesa and found that fine-scale forest structural and topographic metrics significantly influence snow depth during the accumulation season (R-squared of 0.64). Canopies with complex vertical foliage arrangements intercepted more snow, leaving shallower snowpack beneath them. Together with the Meehan et al. (Meehan et al., 2024) finding that wind, terrain, and vegetation jointly control bulk density, these results paint a consistent picture: the trees standing on a hillslope are not passive bystanders but active participants in shaping how much water the snowpack will eventually deliver.

At the hillslope scale, the Berman (Berman, 2011) vegetation survey added a complementary insight by showing that lithology — specifically the contrast between shale and sandstone — structures plant communities along East River hillslopes, with aspen preferentially occupying shale-derived soils. Because vegetation patterns influence snow interception and subsurface water movement, this geological template likely propagates into the hydrologic signals that remote sensing and geophysical methods are designed to detect.

Current frontier

The temporal trajectory is striking: foundational site characterization in the early 2010s has given way, since 2020, to data- and sensor-intensive approaches that merge airborne, ground-based, and computational methods. The Hojatimalekshah et al. (Hojatimalekshah et al., 2023) deep-learning analysis and the Meehan et al. (Meehan et al., 2024) ground-based and airborne sensor integration reflect a broader shift toward machine learning as a tool for bridging scales and filling in what sensors cannot directly measure. Community discussion of GPR research and project outcomes We hope you enjoy this report of GPR research and findings. signals that these methods are becoming part of the everyday toolkit for water managers and researchers alike.

Looking ahead, the frontier is moving toward brokered data integration frameworks that can pull together heterogeneous observations — geophysical surveys, porewater chemistry, time-series groundwater measurements, and remotely sensed snow products — into unified analyses. Emerging methods such as electrical resistivity tomography, airborne electromagnetic surveying with data inversion, and repeated snowmelt sampling promise to extend the integrated sensing paradigm from the snowpack down into the weathered bedrock zone where much of the Gunnison Basin's groundwater storage resides.

Open questions

Several major questions remain. How can snow density be reliably estimated from space, not just from airborne campaigns, so that mountain snow water equivalent can be monitored globally? How do the controls on snow accumulation documented at sites like Grand Mesa translate to the steeper, more heterogeneous terrain of the upper Gunnison Basin? How deeply do vegetation-lithology relationships documented at the surface propagate into groundwater storage and streamflow chemistry, and can integrated sensing detect those linkages in real time? And how should data harmonization and quality assurance standards evolve so that community, agency, and research datasets can be combined without loss of meaning? Answering these questions over the next decade will likely require continued fusion of airborne remote sensing, ground-based geophysics, machine learning, and long-term in-situ monitoring of the kind that RMBL is uniquely positioned to support.

References

Berman, C. (2011). Vegetation Pattern Dependency Along the Southwest-Facing Hillslope of the Upper East River Watershed. →

Hojatimalekshah, A., et al. (2023). Lidar and deep learning reveal forest structural controls on snowpack. Frontiers in Ecology and the Environment. →

Meehan, T. G., et al. (2024). Spatially distributed snow depth, bulk density, and snow water equivalent from ground-based and airborne sensor integration at Grand Mesa, Colorado, USA. The Cryosphere. →

Concept (18) →

Protocol (5) →

Place (3) →

Author (5) →

Publication (3) →

Document (1) →

Dataset (45) →