Literature Review on Predicting Volcanic Eruptions
Francis Marion Alcorn IVVolcanic eruptions pose a significant threat to humans, presenting a risk of damage to property and to industries. For example, an eruption of an Icelandic volcano in 2010 shut down air travel across the Atlantic for a significant length of time (Adam 2010), preventing transportation. As a result, a significant amount of research has been conducted for the purpose of predicting the occurrence and duration of volcanic eruptions. Multiple methods have been proposed for predicting eruptions, such as chemical analysis of volcanic gases and magma and physical analyses of volcanoes themselves. However, as of yet, the only successful methods for predicting eruptions utilize data from seismic activity preceding eruptions. Additionally, other research has been done to predict the duration of earthquakes by using historical data on prior eruptions.
Since small earthquakes preface many volcanic eruptions, the most successful methods for predicting impending eruptions utilize models created through the analysis of data on seismic activity near a volcano. A model of this type was successfully used to predict the eruption of the Oshima volcano in Japan (Miyazaki 2008). The researchers were able to predict four dates on which eruption was likely; the volcano finally erupted on the final date predicted, July 31, 2006, (Miyazaki 2008). Researchers had even more success in predicting the eruptions of the Kamchatka volcanoes in Russia (Senyukov 2013). For one of these volcanoes, Bezymyannyi, the researchers were able to create a model from seismic data collected from prior eruptions. The model was used to determine the probability of an eruption based on an increased frequency of earthquakes; a moderate increases in frequency corresponding to a 50% chance of eruption, and a further increase in frequency corresponding to 90% chance of eruptions. Using this model, the researchers predicted eight out of ten eruptions that did occur, with only one additional false prediction. By similar methods another two Kamchatka volcanoes, Klyuchevskoi and Kizimen, were successfully predicted to erupt. Surprisingly, the model for predicting the eruption of Kizimen was based on data from other, similar volcanoes, as there was no data from its prior eruptions (Senyukov 2013). These methods of producing models for predicting eruptions from seismic data can further be applied to any other well monitored volcanoes.
In addition to monitoring seismic activity, it has been proposed that monitoring the chemistry of volcanic gases and magma can be used to predict eruptions. A study was performed in 1987 on the gases released from the Hawaiian volcano, Kilauea (Crowe et al. 1987). Gases were collected using microfilters at a vent during eruptive and dormant phases, and from a nearby fumarole, or a vent that releases gases but not magma. Analysis of these gases revealed that the concentration of certain trace elements, such as cadmium, rhenium, and gold, correlate with eruptions. It was identified that the ratio of the concentrations of chlorine to fluorine in the gas decreases and the ratio of the concentrations of cadmium to rhenium increases during eruptive phases compared to during dormant phases (Crowe et al. 1987). It was presented that the ratio of cadmium to rhenium can be used as an indicator of eruptions, as high values for this ratio may mean the volcano is in an eruptive phase. The mechanism for this trend was investigated in a later study, which looked at the diffusion of volatile, or gaseous, metals from a melted mixture of metals and metal oxides, to try to determine whether monitoring for trace metals can be used to predict eruptions (Mackenzie and Canil 2008). The metals tested include trace elements such as cadmium, rhenium, lead, thallium, and multiple others. The researchers discovered that the rate of diffusion of cadmium was highest, and that of rhenium was the lowest, with the other metals having intermediate rates of diffusion. This indicates that a high cadmium to rhenium ratio would produce gas bubbles at a high rate, compared to a low ratio, thus increasing pressure and presenting a raised risk of eruption. Additionally, the researchers found that the less viscous a melt is, or the more easily the melt flows (the melt being representative of a volcano’s magma), the faster gases diffuse from it. Thus, the researchers propose that less viscous magma promote eruptions (Mackenzie and Canil 2008). This method for predicting eruptions suffers from the major flaw that it remains untested as of yet. It is likely to remain only theoretical, however, because this method poses risk to research personnel and equipment, especially when compared to remote seismic sensing, a more successful method. It also suffers from the necessity of having a laboratory nearby a volcano for timely analysis of samples. Otherwise, it limits this method’s ability to predict eruptions before they occur.
Another method proposed for predicting eruptions involves utilizing physical methods to predict volcanic processes and assess the risk these pose. It has been proposed that changes in mass of a magma chamber can be calculated using micro-gravity measurements, and that changes in the volume of the volcano’s magma chamber can be determined from ground deformation. (Rhymer and William-Jones 2000). These data can be used to calculate changes in the density of magma in the magma chamber, these changes being indicative of certain processes that may be occurring within the volcano. The authors state that increased density poses low risk of eruption since this means the magma is trapped and cannot rise. However, a decrease in density of the magma poses a high risk of eruption, since this indicates the magma is not crystalizing and is rising and is likely to erupt (Rhymer and William-Jones 2000). This method, like the chemical methods, also suffers from a lack of testing on actual eruptions, and likely will also only remain theoretical
While much research has gone into predicting when eruptions will occur, very little has been done into predicting how long an eruption will last. However, researchers have created a model for predicting the duration of eruptions of Mount Etna in Italy by using historical data on its eruptions (Gunn et al. 2014). They analyzed eruptions since the 1600s, an eruption being defined as eruptive phases, when material is actually released from the volcano, separated by no more than ten days. The data was limited to only after 1600, because data prior to this time were deemed too unreliable. However, despite the data restriction, the data still suffers from conflicting reports, due to human error in recording the duration of eruptions. To address this problem, the researchers incorporated uncertainty in the data into their statistical model. The model is based on a survival analysis curve, originally devised for medical patients in order to record the duration of disease before death, where the relative frequency, or percentage, of eruptions with a certain duration or longer are plotted against said duration. Thus, 100% is at the eruption with the shortest duration, in the upper left corner, and 0% is at the eruption with the longest duration at the bottom right corner. The researchers used this model to propose that there is a 66% chance of an eruption lasting at least seven days and a 33% chance of an eruption lasting at least 86 days (Gunn et al. 2014). The researchers also state that this model can easily be applied to other well-documented volcanoes with a large set of data. However, this is also this method’s major flaw; in order for such a model to be created background data is needed, so this model cannot be applied to many volcanoes lacking recorded data.
While there have been improvements in prediction of eruptions, this science is still imperfect as there is no method that can predict eruptions without flaw. Additionally, while some methods can predict a time interval in which eruption is likely, down to even a few days, no method can predict exactly when a volcano will erupt. These flaws are especially true of other methods than analysis of seismic data, which have not yet seen success and are only theoretical. It is not surprising, however, that these other methods have not been tested with actual eruptions, since they require personnel to get near a potentially high risk volcano and thus presents a danger when compared to monitoring seismic activity. An even larger flaw with this science is the lack of methods to predict the duration of eruptions, and therefore the extent of damage they may cause. However, the science is advancing and scientists are able to limit the loss of life by giving people warning about imminent eruptions, allowing them to reach a safer area.
Works Cited
Adam, Karla, and Ashley Halsey III. “Iceland Volcano Wreaks Flight Chaos; Ash Clouds Shut down Airports in Europe, Paralyzing Travel.” The Washington Post. N.p., 16 Apr. 2010. Web. 21 Sept. 2014.
Crowe, Bruce M., David L. Finnegan, William H. Zoller, and William V. Boynton. “Trace Element Geochemistry of Volcanic Gases and Particles From 1983–1984 Eruptive Episodes of Kilauea Volcano.”Journal of Geophysical Research 92.B13 (1987): 13708- 3714. No Records. Web. 13 Sept. 2014.
Gunn, L. S., S. Blake, M. C. Jones, and H. Rymer. “Forecasting the Duration of Volcanic Eruptions: An Empirical Probabilistic Model.” Bulletin of Volcanology 76.1 (2014): n. pag. Springer. Web. 7 Sept. 2014.
Mackenzie, Jason M., and Dante Canil. “Volatile Heavy Metal Mobility in Silicate Liquids: Implications for Volcanic Degassing and Eruption Prediction.”Earth and Planetary Science Letters 269.3-4 (2008): 488-96.ScienceDirect. Web. 7 Sept. 2014.
Miyazaki, T. H. “Prediction of Volcanic Eruptions by Pseudoanalytic Functions.”Physics of Particles and Nuclei Letters 5.3 (2008): 290-93. Springer. Web. 7 Sept. 2014.
Rymer, Hazel, and Glyn Williams‐Jones. “Volcanic Eruption Prediction: Magma Chamber Physics from Gravity and Deformation Measurements.”Geophysical Research Letters 27.16 (2000): 2389. Wiley Online Library. Web. 7 Sept. 2014.
Senyukov, S. L. “Monitoring and Prediction of Volcanic Activity in Kamchatka from Seismological Data: 2000–2010.” Journal of Volcanology and Seismology 7.1 (2013): 86-97. EBSCO. Web. 7 Sept. 2014.