The Yellowstone caldera is home to thousands of geothermal springs and 75% of the world’s geysers, with kilometers-deep groundwater flow systems that tap magmatic heat sources. As that hot groundwater rises toward the surface, it interacts with shallower, cooler groundwater to produce multi-phase mixing, boiling, and a huge array of different hydrothermal features. While the deep, geothermal water is sexy and merits both the tourist and scientific attention given to it, there’s a largely untold story in the shallow groundwater, where huge volumes of cold water may advect more heat than the hydrothermal features.

Grand Prismatic Spring. (Photo by Alaskan Dude on Flickr.)

Yellowstone is a rhyolitic caldera that has produced 6000 cubic kilometers of ash flow tuffs, rhyolites, and basalts that form a poorly-characterized, heterogeneous fractured rock aquifer, hosting both hot/deep and cold/shallow flow systems. The Yellowstone volcanics lie on top of the Rocky Mountain Cordillera, which itself is a complex hydrogeologic system, ranging from low permeability metamorphic rocks to high permeability limestones.
In a paper in the Journal of Hydrology, Gardner and colleagues (2010) use stream hydrographs and groundwater residence times to characterize the cold, shallow groundwater of the greater Yellowstone area. Stream hydrographs, or the time series of stream discharge, are useful indicators of groundwater dynamics, because in between rain or snowmelt events, streamwater is outflowing groundwater. The recession behavior of a hydrograph during periods between storms can be used to estimate aquifer volumes. In the Yellowstone region, the annual hydrograph is strongly dominated by the snowmelt peak, and Gardner et al. used the mean daily discharge record from 39 streams to characterize the recession behavior of streams on different lithologies. What they found was that streams flowing in watersheds dominated by volcanic rocks have much less variable hydrographs than those on other rock types. The figure below uses data from the USGS to illustrate these differences, which are in line with studies in the Oregon Cascades* and elsewhere which suggest that young volcanic rocks produce groundwater-fed streams with muted hydrographs.

Daily discharge for the Firehole River (USGS gage #06036905) and Soda Butte Creek (USGS gage #06187950) for the 2006-2007 water years, expressed on a unit area basis.

Using a nifty technique to separate the recessions into components attributable to snowmelt versus groundwater, Gardner et al. were able to calculate a ratio of the groundwater discharge to the total discharge of each stream and to calculate the hydraulic diffusivity, which is a ratio of permeability (how easily a fluid moves through a rock) compared to the amount of water stored in the system. If hydraulic diffusivity is low, the flow in the stream decreases slowly over time, like the Firestone River in the figure above. But hydraulic diffusivity can be low either because of low permeability or large aquifer storage volumes, so being able to tease apart those two components is key to understanding the hydrograph behavior. Gardner et al. did this by looking at the ratio of groundwater discharge to maximum discharge and using that as an index of aquifer storage. Based on these ratios, Gardner et al. separated the streams in the Yellowstone area into three groups (runoff-dominated, intermediate, and groundwater-dominated) with contrasting hydrogeologic properties.

Geologic map of a portion of the Yellowstone Plateau, with approximate locations of stream gages of interest noted. Modified from Christiansen (2001, USGS Prof. Pap. 729-G).

Soda Butte and Teton Creeks are runoff dominated, with low groundwater storage and middling recession behavior. Since there is little groundwater storage, in order for hydraulic diffusivity to be low, then permeability must also be low. Sure enough, Teton Creek lies on top of Precambrian gneiss and granite, and unfractured metamorphic and intrusive igneous rocks like these have the lowest possible permeabilities. The Soda Butte Creek watershed comprises Eocene Absaroka volcanics, and older volcanic rocks like these can be quite weathered to clays and relatively impermeable.
The intermediate watersheds of Tower Creek and Cache Creek have significant ratios of groundwater discharge to maximum discharge, but their hydrographs recede rapidly over the summer. This means that they have high permeabilities relative to their aquifer storage volume. The Tower Creek watershed has Eocene tuffs and glacial valleys with alluvial fill, and Cache Creek watershed has Paleozoic carbonates. These materials are known for their high permeabilities, and the low storage volumes can be explained if those layers thinly overly less conductive materials.
The Firehole River, Gibbon River, and Snake River above Jackson Lake are groundwater-dominated, with very high permeabilities but even larger aquifer storage volumes. All of those streams drain primarily Quaternary Yellowstone volcanics, and this hydrologic behavior is in keeping with other young volcanic terrains.
Not content to stop with this hydrogeologic classification of the Yellowstone area, Gardner et al. collected water samples from small, cold springs to analyze CFC and tritium concentrations, which are useful tracers of groundwater travel times. For the springs they sampled, they found an average travel time (from recharge to discharge) of ~30 years. Using those CFC-derived groundwater transit times and some back-of-the-envelope estimates of aquifer geometry, Gardner et al. estimate that the Quaternary Yellowstone volcanics have a permeability of 10^-11 to 10^-13 m², which is in line with estimates of young volcanics elsewhere. They also estimated that the aquifer depth represented by these small springs was ~70 m, but speculated that deeper flowpaths might have been discharging directly into the streams, out of reach of their CFC and tritium sampling abilities.
Finally, Gardner et al. note that the cold springs they studied are actually not as cold as they should be. In fact, they appear to be what are coming to be called “slightly thermal” springs. Groundwater recharge temperature is commonly assumed to be approximately mean annual temperature, and in the Norris Geyser Basin area, that’s around 4-5 &deg C. But the cold springs in the area are around 10 &deg C. Using this temperature difference and a handy equation from Manga and Kirchner (2004), Gardner et al. are able to calculate the heat flux advected by these cool springs. Their value of ~3800 W/m² for the springs around Norris is about 10% of the heat flux from the Norris and Gibbon Geyser Basins themselves. That number becomes even more astonishing when you consider the relative scales of the cool versus the thermal groundwater systems. Geyser basins cover ~10 km² of the Yellowstone Plateau, whereas cool groundwater drains under the entire ~1000 km² plateau, and could be discharging far more heat than those showy thermal springs and geysers themselves.
So if you happen to go to Yellowstone this summer, in between gawking at Old Faithful, Artist Paint Pots, and Mammoth Hot Springs, take a few moments to appreciate the waters of the less dramatic cool rivers and streams. Their waters too are profoundly shaped by the geologic history of Yellowstone, and they are taking an awful lot of heat.
*Disclaimer: My PhD research focused on the hydrology and hydrogeology of volcanic aquifers and streams of the Oregon Cascades.
Payton Gardner, W., Susong, D., Kip Solomon, D., & Heasler, H. (2010). Snowmelt hydrograph interpretation: Revealing watershed scale hydrologic characteristics of the Yellowstone volcanic plateau Journal of Hydrology, 383 (3-4), 209-222 DOI: 10.1016/j.jhydrol.2009.12.037