Keeping Current Mini-Review Earthquake Forecasting

REVIEW ARTICLE

Gerald T. McNeil III1*

In 2016, Ecuador and Italy both experienced deadly earthquakes, with death tolls of over 800 people even with a commonly used earthquake prediction system in place. The seismometer system, which is the current system used in earthquake prediction, provided no help or warning of the devastating earthquakes that occurred. This method only looks at patterns of previous earthquakes to give the prob- ability of an aftershock once the first earthquake occurs. Newer methods currently being studied look at physical changes in the earth and atmosphere caused by tectonic plate shifting which begin before the main event of an earthquake. In this paper, we will review the use of ionosphere electron measurements, changes in water chemistry, pre-seismic tremors, magnetic field changes, and changes in air chemistry, in the field of earthquake prediction and discuss why combining multiple methods can create a more accurate way of making pre disaster predictions.

INTRODUCTION (Scoville, Sornette, & Freund, 2015). Combining these new methods of earthquake prediction with current methods, early earth- During the course of an earthquake many physical changes oc- quake forecasting could significantly reduce the risks of false posi- cur in the earth’s crust causing waves to form in different patterns tive results. Here we review several methods that are relatively and direction: Primary or P-waves which shake in the earth in the new and are still being researched in the field of earthquake fore- direction in which they propagate, Shear or S-waves, which shake casting and the importance of a multi-parameter earthquake pre- the earth perpendicular to the direction from which they propa- diction platform. gate (U.S. Geological Survey, 2016), Love waves which shake the earth both perpendicular to and in the direction in which they Future Prediction Methods propagate, and Rayleigh waves which shake the earth in elliptical Earthquakes occur when two tectonic plates either slip against one patterns (Harkrider, 1964). For many years, people have attempted another, pull apart from one another, or crash into one another. to predict earthquakes by picking up P-waves on seismometers to Currently, seismometers are used to measure P-waves from an al- see that an earthquake is about to occur, P-wave tremors are the ready occurring earthquake to determine the direction in which the first signals to reach seismographs and happen only shortly before main event will go, this method has yet to warn of an earthquake’s an earthquake’s main event (Allen et al., 2009). Progress in creat- approach more than a few seconds in advance. Though seismic ing more sensitive seismographs would possibly allow detection of waves of smaller magnitude in multiple directions possibly longer before the main earthquake strikes than current seismometers allow. Recent studies have shown that prior to earthquakes, the earth has measurable electrical and chemical changes.PROOF These changes are likely to occur because the shifting of tectonic plates applies an enormous amount of pressure on igneous rocks deep in the earth’s crust, which results in the changes to the rocks’ molecular struc- ture.

1 Eastern Nazarene College, 23 East Elm Ave, Quincy, MA 02170, USA

*To whom correspondence should be addressed: [email protected] Figure 1. Schematic representation of the break-up of a peroxy bond Except where otherwise noted, this work is licensed under in a silicate matrix. Step I marks how the break-up of the dormant peroxy https://creativecommons.org/licenses/by/4.0/ defect. Step II marks the electron transfer from an O2− into the decoupled doi:10.22186/jyi.34.1.6-10 peroxy bond and the generation of an electron–hole pair (Freund, 2007).

JYI | January 2018 | Vol. 34 Issue 1 © McNeil 2018 6 REVIEW ARTICLE

PROOF

Figure 2a. Images of simulations of Total Electron Content (TEC) Figure 2b. Images of heat maps of Total Electron Content (TEC) over over the 100 closest receivers to the epicenter of the earthquake a the 100 closest receivers to the epicenter of the earthquake on the month before the earthquake (February 11, 2011) for four different day of the earthquake (March 11, 2011) for four different time points. time points. These points are from the same times of day as (A) the en- These times show (A) enhanced TEC, (B) a sudden depletion, and (C-D) hancement, (B) depletion, and (C-D) fluctuations. The X- and Y-axes rep- the large fluctuations following. The X- and Y-axes represent the longi- resent the longitude and latitude, respectively (Hammerstrom & Cornely, tude and latitude, respectively. (Hammerstrom & Cornely, 2016). 2016).

Figure 3. pH and conductivity data from simulated earthquake conditions in comparison to their control tests. pH results show major drops every time voltage is applied which stop immediately as voltage is removed. There are also significant increases in pH and conductivity over time in the experimental tests (25mV) when compared to the controls (0mV) (McNeil & Cornely, 2017). measurements are still being investigated as possible earthquake in TEC, were also studied on the days leading up to the earthquake. predictors, there are several other changes that occur in the earth, The solar data showed no significant events occurring on the day as a result of an earthquake, which are being investigated for their of the earthquake, thus effects of the solar activities could be ne- use as earthquake predictors. Two major fields of study are The To- glected (Hammerstrom & Cornely, 2016). tal Electron Content (TEC) of the ionosphere, and water chemistry Researchers are still investigating magnetic anomalies related (Balk et al., 2009). These changes likely occur because of shift- to earthquakes. Some research suggests that magnetometers, which ing tectonic plates break unstable peroxy defects shown in Figure record changes in magnetic fields, have picked up disturbances 1. Breaking peroxy bonds allows electrons to flow freely through in magnetic fields prior to earthquake main events (Fraser-Smith igneous rocks. According to Freund (2007), the electrical potential et al., 1990). However, other studies suggest that these anoma- generated can travel far and wide. This would mean that the earth’s lies were likely coincidental sensor errors, as other earthquakes surface could be affected by these charges. PROOFhave not had the same effect on magnetometers since that event Total Electron Content (TEC) of the ionosphere is commonly (Thomaset et al., 2009). Data on magnetic anomalies is difficult used to eliminate possible errors due to signal delay in GPS track- to collect, it requires expensive equipment to be placed near the ing (Hagen, 2009). TEC data has recently been investigated as a epicenter of an earthquake, and it must be collecting data before, possible earthquake forecasting tool. For example, Hammerstrom during, and after an earthquake. The fact that we are unable to and Cornely (2016) examined TEC data from the day of the 9.1 predict where and when an earthquake will occur makes such data magnitude earthquake which occurred in Japan on March 11th relatively rare, the only few cases of recorded data occurred by 2011 at 2:46PM JST, shown in Figure 2a and 2b. This data showed coincidence, a manometer happened to be located near the epicen- a spike in TEC an hour before the earthquake occurred, then a drop ter of an earthquake, collecting the rare data. Though difficult to in TEC shortly before the earthquake and abnormal fluctuations in record, researchers are developing more sensitive magnetometers TEC values for several hours after the earthquake. Normally, TEC to better record data from earthquakes (Katori et al., 2014) over an area remains stable and any changes occur slowly and Pre earthquake stress can crack igneous rocks deep in the earth smoothly. Solar activities, which are the main cause of anomalies which contain uranium in significant quantities. Uranium breaks

Figure 4. A Venn diagram comparing possible sources of error in three methods being researched for possible use in earthquake forecasting. If the normal alpha value of 0.05 is used for each method, then using all three together would give an alpha value of 0.000125, which is only a 0.0125% chance of a type I error. down into daughter product thorium, radium, and eventually ra- experiments that inducing high stress on igneous rocks caused an don via alpha decay. Alpha decay is the release of an alpha par- electrical potential to form. It is thought that this potential would ticle, or helium nucleus, from an unstable radioactive atom which cause Redox reactions in ground and well water when an earth- forms and atom of an element two atomic numbers lower (Larsen quake occurs likely changing the pH of the water. Another study & Gottfried, 1960). Once radon gas is formed, it becomes trapped found anomalies in ion concentrations of water collected from lo- inside of the igneous rocks until some force releases the trapped cal water shortly after an earthquake (Chen et al., 2014). Mod- gas. Pre-seismic pressure on these igneous PROOFrocks can crack them, ern technology allows for constant monitoring of water acidity releasing the radon gas. Radon decays to form alpha particles and and ionic changes with devices available to just about everyone. daughter products; polonium, bismuth, and lead sequentially, each These chemistry sensors could be implemented for use in earth- of which also decay forming alpha particles. Specialized sensors quake forecasting in the near future. One study simulated known which detect alpha particles, then collect data on all of the radio- electrical changes caused by pre-earthquake conditions in samples active decay which could then alert to the possibility of an earth- of collected pond water then measured chemical changes that oc- quake. Radon’s short half-life of 3.8 days and the fact that it is curred as a result of this simulated earthquake. This experiment easily dispersed by wind and diffusion allows many readings in an found significant changes in both pH and conductivity values in area over several days. If the sensors frequently detect radon or the water when compared to control tests shown in Figure 3. The its daughter molecules, there is a high likelihood of an earthquake data collected also showed evidence that voltages that occur in (Jashank et al., 2014). igneous rocks due to pressure caused by tectonic activity, could be Water chemistry has been studied mostly in simulated earth- immediately detected by pH probes placed in water touching said quake conditions. Takeuchi, Lau, and Freund (2006) found in their rocks. Patterns in pH change and observed anomalies in collected

data are being investigated for possible indication of earthquake Jashank, M., Gupta, S. J., & Nair, J. (2014). Development of Radon Gas Sensor to magnitudes and time until earthquake’s main event (McNeil & Monitor the Precursors of Earthquake. IOSR Journal of Engineering, 4(1), 10-15. Katori, Y., Okubo, K., Hato, T., Tsukamoto, A., Tanabe, K., Onishi, N., . . . Takechi, Cornely, 2017). N. (2014). Evaluation of High-resolution Geomagnetic Field Observation Dr. Pierre-Richard Cornely hopes to combine several methods System Using HTS-SQUID Magnetometer and its Application. AGU Fall of possible earthquake forecasting into one platform in order to Meeting Abstracts, 1, p.3864. eliminate possible errors due from other sources. Utilizing only Larsen, E. S., & Gottfried, D. (1960). Uranium and Thorium in Selected Suites of Igneous Rocks. American Journal of Science (U.S.), 258-A. one precursor to predict earthquakes allows for many failures due Lindh, A. G., Lockner, D. A., & Lee, W. H. (1978). Velocity Anomalies: An to many possible errors in data collection. However, combining Alternative Explanation. Bulletin of the Seismological Society of America, two or more different forecasting methods eliminates many, if not 68(3), 721-734. all, possible data errors allowing for a more accurate prediction McNeil, G. T., & Cornely, P. J. (Under Review @ JYI 2017). Water Chemistry Changes and Correlation with Seismic and Pre-earthquakes Activities. (Cornely, 2016). Scoville, J., Sornette, J., & Freund, F. T. (2015). Paradox of peroxy defects and Figure 4 compares possible sources of error in TEC, water positive holes in rocks Part II: Outflow of electric currents from stressed chemistry, and pre-seismic vibrations. While some methods have rocks. Journal of Asian Earth Sciences, 114, 338-351. doi:10.1016/j. overlapping sources of error, all three together do not share a pos- jseaes.2015.04.016 Takeuchi, A., Lau, B. W., & Freund, F. T. (2006). Current and surface poten- sible error and therefore viewing the data from all three would tial induced by stress-activated positive holes in igneous rocks. Physics significantly reduce the risk of a Type I error in earthquake predic- and Chemistry of the Earth, Parts A/B/C, 31(4-9), 240-247. doi:10.1016/j. tions. Statistically utilizing multiple different parameters for pre- pce.2006.02.022 dicting earthquakes multiplies the alpha levels of each parameter Thomas, J. N., Love, J. J., & Johnson, M. J. (2009). On the reported magnetic precursor of the 1989 Loma Prieta earthquake. Physics of the Earth and Planetary thus drastically reducing the system as a whole’s chance of falsely Interiors, 173(3-4), 207-215. predicting an earthquake. It is also shown that TEC and Water U.S. Geological Survey (2016). Retrieved April 15, 2017, from https://pubs. chemistry do not share a common source of error, reducing the usgs.gov/gip/earthq1/measure.html number of data sources to just two could significantly reduce the cost of making such an earthquake predicting device. Research in the field continues to find new and useful tools for reading the earth to detect warnings of an impending earthquake is becoming more possible than ever before. With these advances in the field of earthquake forecasting, we may be very close to creating a functional warning system that gives ample time to save hundreds of thousands of lives. REFERENCES Allen, R. M., Gasparini, P., Kamigaichi, O., & Bose, M. (2009). The Status of Earthquake Early Warning around the World: An Introductory Overview. Seismo- logical Research Letters, 80(5), 682-693. doi:10.1785/gssrl.80.5.682 Balk, M., Bose, M., Ertem, G., Rogoff, D. A., Rothschild, L. J., & Freund, F. T. (2009). Oxidation of water to hydrogen peroxide at the rock–water interface due to stress-activated electric currents in rocks. Earth and Planetary Science Let- ters, 283(1-4), 87-92. doi:10.1016/j.epsl.2009.03.044 Chen, Z., Du, J., Zhou, X., Yi, L., Liu, L., Xie, C., . . . Li, Y. (2014). Hydrochemistry of the Hot Springs in Western Sichuan Province Related to the Wenchuan M S 8.0 Earthquake. The Scientific World Journal, 2014, 1-13. doi:10.1155/2014/901432 Cornely, P. J. (2016). Summer Research 2016 Data AssimilationPROOF & Fusion Platform for Earthquake Forecasting. Retrieved from https://www.researchgate.net/profile/ Pierre_Richard_Cornely/publications on April 10, 2017. Fraser-Smith, A. C., Bernardi, A., Mcgill, P. R., Ladd, M. E., Helliwell, R. A., & Villard, O. G. (1990). Low-frequency magnetic field measurements near the epicenter of the Ms7.1 Loma Prieta Earthquake. Geophysical Research Letters, 17(9), 1465-1468. doi:10.1029/gl017i009p01465 Freund, F. T. (2007). Pre-earthquake signals – Part II: Flow of battery currents in the crust. Natural Hazards and Earth System Science, 7(5), 543-548. doi:10.5194/ nhess-7-543-2007 Hagen, J. B. (2009). Radio-frequency electronics: circuits and applications. Cam- bridge: Cambridge University Press. Hammerstrom, J. A., & Cornely, P. J. (2016). Total Electron Content (TEC) Varia- tions and Correlation with Seismic Activity over Japan. Journal of Young Investigators, 31(4), 17–20. doi:10.22186/jyi.31.4.13-16 Harkrider, D. G. (1964). Surface waves in multilayered elastic media I. Rayleihj and love waves from buried sources in a multilayered elastic half-space. Bul- letin of the Seismological Society of America, 54(2), 627-679.