Inter-Island Distance Learning

INTER-ISLAND DISTANCE LEARNING EXCHANGE PROJECT

FORECASTING VOLCANIC ERUPTIONS

Written By

R. B. Trombley, Ph.D. Southwest Volcano Research Centre Apache Junction, Arizona USA

1 Introduction

Forecasting the time, place, and character of a volcanic eruption is one of the major goals of volcanology. It is also one of the most difficult goals to achieve. At the Southwest Volcano Research Centre (SWVRC, an experimental computer programme, specifically designed for the MS-DOS & Windows based PCs has been developed and tested over the past seventeen years in an attempt to forecast long-range volcanic eruptions. The ERUPTION Pro 10.5 software package’s intent is to forecast the next eruption event of volcanoes about the world. For the past 7 years, ERUPTION Pro 10.5 ‘s accuracy has been over 90%. For those readers interested in the paper written and mathematics, etc. of the programme, go to the following website URL: http://www.swvrc.org/other.htm and click on Paper, "ERUPTION Pro 10.3 - The New & Improved Long-Range Eruption Forecasting Software".

The Principal Inputs to Forecasting

Forecasting is primarily done with the monitoring of a volcano’s activity. It is very much like going to a physician when one is not feeling well. The physician makes certain “measurements”, e.g., takes your temperature, blood pressure, general looks, etc. which will allow the physician to examine the symptoms and thereby determine the cause of the illness. Much is the same with the monitoring of volcanoes. We look at volcanoes in very much the same way. That is, we look at the volcano’s “vital” signs and determine the symptoms which will now allow a volcanologist to determine the probability of an eruption event.

In addition to the normal probability contribution in forecasting from the historical data, there are several other symptoms that contribute to the overall analysis. Those other symptoms are: Input from Correlation Spectrometer (COSPEC), Thermal Imaging, Volcanic-Seismicity, Crater Lake Temperature (where applicable), Deformation and the volcano’s Frequency of Eruption analysis. Let’s take a look at these different inputs, learn about them, and see how they help scientists (volcanologists) determine if a volcano is about to erupt.

2 Correlation Spectrometer (COSPEC)

A Correlation Spectrometer (COSPEC) is a ground-based, remote sensing, ultraviolet correlation spectrometer which measure the flux of the sulfur dioxide gas emissions from a volcano. There are several places where readings can be taken; from the ground or the air.

On a tripod near a vent of a volcano. Flying beneath volcanic gas plume to measure SO2 Photograph by S.R. Brantley on June 16, 1982.

The COSPEC functions in the ultraviolet part of the electromagnetic spectrum, with the sky used as the source of ultraviolet energy. Light enters the instrument and travels through a series of mirrors, lenses, and slits to reach a detector and photomultiplier, where the light is converted into electric pulses and amplified. If gas is converted into electric pulses and amplified then one can examine it with a Correlation Spectrometer. If a gas is present in the path of the instrument, the

COSPEC detects the amount of ultraviolet radiation that is absorbed by the SO 2. Since the COSPEC is measuring the concentration of gas through a thin cylindric slice of the atmosphere, the concentration of SO2 in this slice is expressed as parts per million-meters (ppm-m), rather than simply as ppm for an instrument which directly measures gas within a plume. Typically, the COSPEC is mounted in a vehicle, with the telescope, mirror, and lens assembly oriented vertically upward. The vehicle then makes traverses beneath the gas plume, with the traverses oriented perpendicularly to the plume. If there is no road access, the COSPEC can be placed in an aircraft which is flown under the plume. The COSPEC also can measure from a stationary position. Since the volcanologist's goal is to measure the amount of SO2 released from the

3 volcano over a period of time, it is essential to measure the width and speed of the plume as it is dispersed downwind from the actively degassing crater or craters. Unfortunately, the speed of the plume frequently is difficult to estimate, and scientists often rely upon the wind speed as a guide.

Thermal Imaging

In modern volcanology and with the advent and availability of satellite technology, it is relatively easy to determine and measure sources of heat on earth from space. Another addition to the probability contribution input is the input due to an increase on thermal output from the volcano under analysis. This accomplished through satellite based thermal imaging. Input for this contribution is obtained from the Geostationary Operational Environmental Satellite (GOES) namely, GOES-8, GOES-9, GOES-10, the Operational Significant Events Imagery (OSEI), and the Advanced Very High Resolution Radiometer (AVHRR) satellite imaging.

Some examples of thermal imaging are shown below.

Left view show the profile of the Big Island of Hawaii while An AVHRR view of Kamchatka the right image show the resulting thermal imaging, in this (Russia). The two bright spots in case of the Mauna Loa area. (Note the red showing heat.) the middle are volcanoes Kliuchevskoi (top) & Bezymianny (bottom).

Direct satellite data reception of high temporal frequencies and automated processing enable near-real-time, near-continuous thermal monitoring of volcanoes. Current capabilities focus on 2 instruments: the Advanced Very High Resolution Radiometer (AVHRR) and the Geostationary

4 Operational Environmental Satellite (GOES) imager. Collection of 10 AVHRR images per day covering Alaska, the Aleutians, and Kamchatka (Russia) allows routine, on-reception analysis of volcanic hot spots across these regions.

The high inclination (98o) orbit of the NOAA polar orbiting satellites, on which the AVHRR is flown, coupled with its wide field of view width (~3,000 km) means that high regional area coverage ( > 2 images/day/satellite) can be obtained at all latitudes. NOAA’s policy of ensuring that at least 2 satellites are in orbit at any one time means that regional coverage is increased to >4 images per day at equatorial latitudes. Sidelap and converging orbits towards the poles allows almost hourly coverage for volcanoes towards the pole of 60o. Because the low data rate (~665 kbits/sec) permits direct reception at relatively low cost there has been much effort to develop techniques for near-real-time monitoring using AVHRR data.

Because of their extremely high regional area coverage and wide field of view (full earth as viewed from an altitude of 35,770 km) , the GOES stationary satellites provide a powerful tool for volcano hot spot monitoring. The geostationary orbit has the advantage of regularly covering all volcanoes over a large area and provides a constant viewing geometry, making it easier to carry out automated image co-registration and production of time-lapse movie loops. A major benefit of GOES data is its high regional coverage resolution. In routine operating mode the GOES imager scans the latitudes of the United States once every 15 minutes, and South America once every 30 minutes. Once every 3 hours the imager takes 30 minutes to scan the full earth disc.

To be of value to the non-remote sensing specialist (e.g., volcanologists, etc.), satellite derived hot spot information must be presented in a fully processed form and in a way that is clear and easy to interpret (see views above). Suitable formats include colour images with flagged hot spots and radiance time series that show peaks and troughs due to variations in volcano thermal activity. To be useful in an operational and hazard monitoring role, the information must also be available in a timely fashion. Automated and routine processing of AVHRR and GOES data and rapid information dissemination via the Internet have shown the way forward.

5 Volcanic-Seismicity

Seismicity plays an important and major role in the probability determination relative to a volcano’s potential (probability) of having an eruptive event. It is, arguably the largest contributor to the probability determination of an impending eruption.

In order for a volcano to erupt, magma must move to the surface. These underground movements or a change in volume of molten rock generally produce signals that can be detected by geological, geophysical, and geochemical observations before an eruption begins. In some cases the changes may be obvious to people living near a volcano; for example, they may feel a swarm of earthquakes or note consistently increased fuming from the crater. More often, the changes in earthquake pattern, ground surface deformation, and the composition of gas emissions are small and subtle and can be detected only by continuous monitoring with sensitive instruments.

Here is an example of seismic activity, a seismogram of Mt. Erebus, in Antarctica, from 31 May 2004.

6 7 An increase in earthquakes at shallow depths beneath a volcano is generally an indication of some significant change, perhaps an increase in volume of magma in the storage chambers. Some earthquake swarms may be caused by changes in the regional stress system and are concentrated beneath a potentially active volcano because the hot rocks there are weaker and more prone to break than are colder rocks in the surrounding region. In fact, some magma chambers beneath volcanoes are so plastic that they deform without the fracturing the causes earthquakes. In this case, locating the envelope of tiny earthquake centers in the weak but still brittle rocks that surround a magma chamber provides a subsurface image of the location and size of that reservoir. Besides recording individual earthquakes, seismometers near an active volcano often pick up continuous ground vibrations called volcanic tremor. This tremor is nearly always seen during volcanic eruptions, and it is often recorded before an eruption starts. The causes of volcanic tremor probably include magma moving through conduits at depth, gases boiling out of magma or groundwater, or a close sequence of many small earthquakes. Because some volcanic is caused by shallow intrusion of magma, it is of great importance in the seismic monitoring of active volcanoes.

Crater Lake Temperature

Not all volcanoes have a lake at their crater but those that do provide an additional observation and measurement that may be used by the volcanologist to determine what’s going on inside the volcano. It is very much like putting a teakettle on the stove. When one turns the heat on, the water begins to heat as well. In a volcano, when the heat is “turned on” (magma entering the magma chamber from the earth’s interior), the water in the crater lake begins to heat up as well.

Volcanic lakes occur all over the world with a higher percentage in volcanic arcs. These lakes may look like other natural reservoirs that accumulate water from precipitation, runoff, and inflowing streams. Yet many volcanic lakes are hot and contain large amounts of dissolved solids in their waters. In fact, a few of these lakes consist of concentrated mixtures of sulfuric and hydrochloric acids that represent the most acidic and mineralized natural waters on earth. Some also host long-lived molten sulfur bodies, a spectacular phenomenon shared only with a

8 moon of the planet Jupiter, Io. Lastly, a few volcanic lakes can store large amounts of potentially lethal gas in their bottom waters. As an example, changes in the temperature of the crater lake on Taal Volcano in the Philippines sometimes warn of forthcoming eruptions. In 1965 the temperature began to rise well above its background level of 33oC in July; the volcano erupted in September. The 1966 eruption was preceded by a much less obvious temperature rise, and the 1967 eruption could not have been forecast on the basis of temperature alone.

The crater lake temperature profile of volcano Taal in the Philippines.

Deformation (Tilt)

As with other contributions, ground deformation plays a role in the probability determination of an eruption. Ground deformation on volcanoes may be due to several causes. These include: (a) inflation/deflation of a buried magma storage zone, (b) injection of a dike or sil which may or may not be an eruption conduit, (c) subsidence due to lava loading, gravitational settling, or spreading of the entire volcano, and (d) slope movement caused by slope creep prior to failure or by magma pressure variations on steep slopes. Combinations of these causes frequently occur to produce a complex pattern of deformation.

9 Slight changes in the slopes or distances between survey points on the summits and flanks of active volcanoes provide another major method of diagnosing the internal changes taking place. Several techniques are involved, including conventional leveling and the determination of distances with reflected laser beams. Tiltmeters, which can detect changes in slope smaller than 1 part per million, are also used. An angular change in slope of 1 part per million (1 microradian) is equivalent to lifting the end of a rigid board 1 kilometer long by only 1 millimeter. A promising new technique called GPS (Global Positioning System) uses satellites and portable receivers to determine the relative position of the ground stations to accuracies of a few millimeters. Radio code and time signals broadcast by the satellites arrive at ground-surface

An example of a recent set of Tiltmeter readings from Kilauea volcano.

survey stations in a unique pattern. Comparison by computer processing of the different patterns recorded at two survey stations determines their relative location.

10 Both elevation and distanceA recent changes set of GPS are readingsdetectable and by data GPS from measurements; volcano the receivers can be Kilauea. many kilometers apart, and need to "see" the satellites but not each other. Although the present instruments and data-reduction computers are quite expensive, important observations can be made in a few hours with only one person at each receiver.

Repeated at frequent intervals, these surveying techniques reveal the tiny, unseeable deformations of the volcano's surface caused by changes in magmatic pressure or volume inside the volcano. Deformation monitoring, the collective name for all of these surveying techniques, has become the second most important means of forecasting volcanic eruptions, preceded only by the monitoring of earthquakes.

Volcano’s Frequency of Eruption

Another of the important inputs to forecasting the probability of an eruption event is a look at the volcano’s eruption frequency, i.e., what is the eruption rate or cycle if any ? In the case of ERUPTION Pro 10.5’s software package inputs, only the last ten years of a particular volcano is examined. For example, if a particular volcano (e.g., Kilauea in Hawaii) has erupted every year for the past 10 consecutive years, there is a high probability that it will erupt in the eleventh year. Volcanoes that have erupted at some time in the last 10 years, but not every year, will have their probability contribution lessened depending on how many times it has erupted within the 10 year period.

11 Eruption History

The value of the recent volcanological record is obvious to volcanologists. The events of the present provide some of our best clues in interpreting the volcanic products of the past few billion years of the earth’s history. Eye-witness accounts, photographs, and instrumental documentation all build a picture of the processes, rates, and interrelationships of events that cannot be found from products alone and yet are essential for understanding volcanism.

In studying a particular eruption, the accounts of similar historical eruptions can provide valuable guides in investigating a specific type of volcanic process or product. The historical record gives a useful dynamic context and in assessing the likelihood of future activity at a particular volcano. And lastly, of course, the historical behaviour of similar, but better instrumentally documented volcanoes, is often helpful.