Volcano Watch — Mauna Loa eruption would give an hour's notice

Several weeks ago, on August 12, the Center for the Study of Active Volcanoes at the University of Hawai`i at Hilo and the U.S. Geological Survey's Hawaiian Volcano Observatory held a public seminar on the prediction and mitigation of volcanic eruptions and earthquakes. The seminar consisted of four panels of participants, each of whom made a brief presentation followed by a question-and-answer period. The seminar was well-attended, with roughly 100 people in the audience.

The first panel consisted of five scientists, who presented information on how well we are able to forecast and predict geologic events before they happen. A forecast is long-range and concerns the likelihood that certain events will occur within rather long time frames, whereas a prediction is short-term and more precise about the size and location of the predicted future event.

Examples of forecasts include such statements as "Kīlauea, Mauna Loa, Hualālai, and even Mauna Kea will probably all erupt in the future because Kīlauea and Mauna Loa have erupted within the past few years or decades, Hualālai has erupted within the last several hundreds of years, and Mauna Kea has erupted within the last several thousands of years."

A shorter-term forecast might be "Mauna Loa is likely to erupt within the next five years, because magmacontinues to accumulate within the shallow magma storage reservoir; however, the roughly year-long increase in seismicity that preceded the 1975 and 1984 eruptions has yet to begin." Based on our current understanding of Mauna Loa, a prediction of the next Mauna Loa eruption will probably only precede the actual eruption by a little less than one hour, which is the duration of volcanic ground vibration (tremor) that occurred before the last two eruptions. The Hawaiian Volcano Observatory is currently working on identifying eruption precursors that would allow for a prediction with a lead time longer than one hour, but shorter than the roughly one year-long increase in seismic activity.

There are many kinds of data that we collect to help predict future volcanic events. In January 1983, the migration of earthquake epicenters along the East Rift Zone of Kīlauea Volcano traced the advance of a blade-like body of magma precisely enough that geologists were waiting at the site where the current eruption broke out. At other volcanoes, such as Mount St. Helens, rapid changes in the shape of the ground surface (upward bulges), coupled with migration of earthquake epicenters, indicated that a large eruption was going to take place within a few days or weeks. The area was evacuated before the large-scale eruption took place, and many lives were saved. Similarly, at Mt. Pinatubo in the Philippine Islands, these same precursors and the emission of large plumes of volcanic gas provided the eruptive precursors that led scientists to recommend to civil authorities that it was time to evacuate a large region, thereby saving thousands of lives. The scientists were correct in predicting a large eruption of Mount Pinatubo and recommending evacuation. Many of these explosive volcanoes, like Mount St. Helens and Mount Pinatubo, lie dormant for hundreds, or even thousands, of years before they show new signs of activity. In addition, each volcano has a different threshold before it erupts, and each eruption changes the internal structure of the volcano so that the threshold at a single volcano changes from one eruption to the next. These ever-changing variables make it unlikely that the scientists will ever develop a fail-safe set of parameters to allow for precise prediction of eruptions.

Earthquakes are in many ways easier to forecast but more difficult to predict. The reason why they are easy to forecast is that earthquakes tend to occur along well-defined fault structures, the long-term rates of movement along may of these faults can be determined, and they occur frequently enough to have established patterns during historical time. In other words, large earthquakes do not occur randomly in either space or time. The patterns of past earthquakes help us forecast where, and how frequently, future earthquakes are most likely to occur. Forecasting earthquakes has progressed to the point where scientists can develop statistical models of the likelihood of an earthquake of a certain magnitude occurring within a certain time period along a specific section of faultline for high-risk areas. An example of an earthquake forecast was presented several weeks ago when we outlined reasons why we think the south flank of Kīlauea Volcano is likely to experience yet another moderate-sized earthquake within the next few years. This forecast is based on the similarity of events now to those that preceded the 1989 and 1975 earthquakes in the same region. Prediction, on the other hand, requires that some identifiable precursors occur. Earthquakes invariably precede large eruptions, whereas for large earthquakes, no reliable precursors have yet been identified. Considerable research effort is focused on the search for short-term precursors to earthquakes.

In future columns, I will outline some of the discussions that took place at the symposium during the three panels of citizens and business—Civil Defense, the media, the National Park Service, and public utilities—and State and County officials responsible for land-use planning and building codes.