US Geological Survey


Date of this Version



Published in U.S. Geological Survey Professional Paper 1717, 1-35, (2007)


The extraordinary number, size, and unspoiled beauty of the geysers and hot springs of Yellowstone National Park (the Park) make them a national treasure. The hydrology of these special features and their relation to cold waters of the Yellowstone area are poorly known. In the absence of deep drill holes, such information is available only indirectly from isotope studies. The δD-δ18O values of precipitation and cold surface-water and ground-water samples are close to the global meteoric water line (Craig, 1961). δD values of monthly samples of rain and snow collected from 1978 to 1981 at two stations in the Park show strong seasonal variations, with average values for winter months close to those for cold waters near the collection sites. δD values of more than 300 samples from cold springs, cold streams, and rivers collected during the fall from 1967 to 1992 show consistent north-south and east-west patterns throughout and outside of the Park, although values at a given site vary by as much as 8 ‰ from year to year. These data, along with hot-spring data (Truesdell and others, 1977; Pearson and Truesdell, 1978), show that ascending Yellowstone thermal waters are modified isotopically and chemically by a variety of boiling and mixing processes in shallow reservoirs. Near geyser basins, shallow recharge waters from nearby rhyolite plateaus dilute the ascending deep thermal waters, particularly at basin margins, and mix and boil in reservoirs that commonly are interconnected. Deep recharge appears to derive from a major deep thermal-reservoir fluid that supplies steam and hot water to all geyser basins on the west side of the Park and perhaps in the entire Yellowstone caldera. This water (T ≥350°C; δD = –149±1 ‰) is isotopically lighter than all but the farthest north, highest altitude cold springs and streams and a sinter-producing warm spring (δD = –153 ‰) north of the Park.

Derivation of this deep fluid solely from present-day recharge is problematical. The designation of source areas depends on assumptions about the age of the deep water, which in turn depend on assumptions about the nature of the deep thermal system. Modeling, based on published chloride-flux studies of thermal waters, suggests that for a 0.5- to 4-km-deep reservoir the residence time of most of the thermal water could be less than 1,900 years, for a piston-flow model, to more than 10,000 years, for a well-mixed model. For the piston-flow model, the deep system quickly reaches the isotopic composition of the recharge in response to climate change. For this model, stable-isotope data and geologic considerations suggest that the most likely area of recharge for the deep thermal water is in the northwestern part of the Park, in the Gallatin Range, where major north-south faults connect with the caldera. This possible recharge area for the deep thermal water is at least 20 km, and possibly as much as 70 km, from outflow in the thermal areas, indicating the presence of a hydrothermal system as large as those postulated to have operated around large, ancient igneous intrusions. For this model, the volume of isotopically light water infiltrating in the Gallatin Range during our sampling period is too small to balance the present outflow of deep water. This shortfall suggests that some recharge possibly occurred during a cooler time characterized by greater winter precipitation, such as during the Little Ice Age in the 15th century. However, this scenario requires exceptionally fast flow rates of recharge into the deep system.

For the well-mixed model, the composition of the deep reservoir changes slowly in response to climate change, and a significant component of the deep thermal water could have recharged during Pleistocene glaciation. The latter interpretation is consistent with the recent discovery of warm waters in wells and springs in southern Idaho that have δD values 10–20 ‰ lower than the winter snow for their present-day high-level recharge. These waters have been interpreted to be Pleistocene in age (Smith and others, 2002). The well-mixed model permits a significant component of recharge water for the deep system to have δD values less negative than –150 ‰ and consequently for the deep system recharge to be closer to the caldera at a number of possible localities
in the Park.