Editor's note: Yellowstone Caldera Chronicles is a weekly column written by scientists and collaborators of the Yellowstone Volcano Observatory. This week's contribution is from Mark Stelten, research geologist with the U.S. Geological Survey and deputy Scientist-in-Charge of the Yellowstone Volcano Observatory.

Geologists from the Yellowstone Volcano Observatory are often asked to estimate how likely future eruptions are at Yellowstone, but it’s no walk in the national park.

People visit Yellowstone National Park every year to observe its wildlife and vast array of hydrothermal features. One question that lurks in the back of many visitors’ minds as they traverse through one of the world’s largest active volcanoes is: When is Yellowstone’s next eruption going to be? When a volcano is restless, this question can be addressed by examining trends in monitoring data, like seismicity, ground deformation, and gas emissions. But what about dormant volcanoes, like Yellowstone, that are showing no signs of stirring anytime soon?

For currently dormant volcanoes, we don’t usually predict the dates of future eruptions, but rather the probability that an eruption will occur during some time frame (for example, over the next year or next 10 years). This is sort of like long-term weather forecasts—for example, estimating the probability that the upcoming hurricane season will have more hurricanes than an average year.

To an extent, forecasts of volcanic eruptions rely upon knowledge of the frequency at which eruptions occur at a given volcano. As an analogy, let’s say that you live next to a baseball field, and you want to get an idea of the next time a baseball will be hit into your yard. One way to forecast this would be to calculate an average recurrence rate by dividing the number of baseballs in your yard by the duration of your observation period (let’s say, 1 year), to derive the number of baseballs in your yard per year. This average recurrence rate can then be turned into a probability of a baseball being hit into your yard over the next day, week, month, etc. Similarly, forecasting volcanic eruptions requires knowing the number of eruptions that have occurred over time. Geologists achieve this by combining geologic mapping with geochronology to determine a volcano’s eruptive history.

Map of Yellowstone caldera showing the locations and ages of the most recent rhyolite eruptions at Yellowstone, the Central Plateau Member rhyolites. Unit boundaries are from Christiansen (2001). The West Thumb region of Yellowstone Lake is indicated because it is thought to be the location of an explosive eruption and the source vent for the Tuff of Bluff Point. The Central Plateau Member rhyolites are broken into five informal groups based on new 40Ar/39Ar eruption ages. Each informal eruption group is shown in the same color. Numbers on the map and legend are included to indicate the location of different lava flows. Group mean ages and their 95% confidence intervals are included next to the list of units.

Knowing the average rate of volcanic eruptions is only the start. Geologists also need to understand if volcanic eruptions are one-off events that happen independent of other eruptions, or if they occur in groups as part of a bigger volcanic event. Going back to the baseball analogy, because baseball is played during only parts of the year, it is much more likely that baseballs will be hit into your yard during the baseball season rather than in the off-season. Recent research has shown that many volcanic systems, including Yellowstone, work in a similar way, with multiple eruptions occurring in rapid succession, separated by long periods with few to no eruptions. To accurately forecast volcanic eruptions, this “grouping” of eruptions needs to be well-characterized.

Schematic summary of rhyolite eruptions in the Yellowstone Plateau volcanic field over the past 1.3 million years. Smaller rhyolite eruptions are known intracaldera eruptions, meaning they occurred within existing caldera structures. Additional rhyolite eruptions that occurred outside the caldera are not included in the figure.

Determining the rate and pattern of volcanic eruptions is only part of the job. Once the history of volcanic eruptions through time is known, the next task is to try to understand where the volcano currently stands in terms of its life cycle. Returning to the baseball analogy one last time, this is like trying to figure out if it is currently the baseball season or the off-season. The difficulty with places like Yellowstone is that they produce large but infrequent eruptions, with thousands to hundreds of thousands of years between eruptive episodes (where an episode could include one or more eruptions). This means there are few observations upon which to base our forecast, and there are (fortunately) not many opportunities to test these forecasts. For example, no eruptions have occurred in Yellowstone National Park during the past 70,000 years. From 160,000 years ago to 70,000 years ago, however rhyolite lava flows (or groups of lava flows) were erupting approximately every 20,000 years on average. Does this mean we are currently in the volcanic off-season? Or does it mean we are “due” for an eruption (which, by the way, is never really true)? The reality is that we cannot say for sure based on statistical forecasting methods alone. Instead, we must combine these types of forecasts with real-time monitoring of the volcano to assess the state of the volcanic system.

Based on our current knowledge of Yellowstone’s eruptive history, the annual probability of a volcanic eruption is on the order of 0.001%, but even this low number is probably an overestimate for the short term. There are no signs of an impending volcanic eruption based on monitoring data, and we know that the magmatic system beneath Yellowstone is mostly solid. But one day, perhaps thousands or tens of thousands of years from now, the volcanic off-season in Yellowstone may end, and volcanologists will be watching for signs of incoming baseballs.

Panoramic of the West Yellowstone rhyolite lava flow taken along Highway 20 (between the West entrance of Yellowstone National Park and Madison Junction). The flow is approximately 111,000 years old and has a volume of about 41 km3 (10 mi3).