Authors: Michael O'Brien and Pierre-Richard Cornely
Institution: Eastern Nazarene College
ABSTRACT
Earthquakes pose a significant threat to human life, especially when they affect major centers of population. However, there is currently no method of predicting seismic activity that gives enough time to prepare for a major catastrophe. This paper presents a method for predicting earthquakes months before an actual seismic event occurs. A major earthquake, which occurred near Haiti’s capitol in early 2010, is investigated using the Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions (DEMETER) satellite. Electron density in the ionosphere is measured through DEMETER. The electron density measurement is then compared to predictions by the Parameterized Ionospheric Model (PIM), a computer model capable of predicting the electron density at a given point in the atmosphere. This comparison reveals a fluctuation in ionospheric electron density which was not predicted by the model and is not accounted for by other solar or magnetic sources. This data demonstrates a correlation between electron density in the ionosphere and major seismic activity, which can be observed months before the seismic event occurs. With further research, it may be possible to predict when and where earthquakes will strike long before they actually occur.
INTRODUCTION
The connection between the ionosphere and earthquakes has come to light recently with research into how the ionosphere can be used to predict seismic activity. Being part of the upper atmosphere, the ionosphere ranges from approximately 85 to 900km above the Earth’s surface and is filled with electrons and other charged particles. One way to characterize the ionosphere is to measure its electron density, which is most often affected by solar and magnetic activity. Seemingly unrelated, earthquakes usually occur at the fault line between two tectonic plates within the earth’s crust. Although tectonic plates typically move very slowly, it is possible for the plates to experience a sudden jolt, resulting in an earthquake. Earthquakes can be devastating, yet there is currently no reliable method for predicting earthquakes.
The first significant advances in connecting the ionosphere to seismic activity were made by Freund (2002; 2010). There are several contending theories concerning the exact mechanics responsible for the seismic indicators in the ionosphere, but the basic premise is fairly consistent. Prior to an earthquake, due to the seismic forces in the Earth’s crust, a thin layer of particles forms as ions originating in the crust travel to the Earth’s surface and begin radiating outwards due to strong electric fields (Namgaladze, Klimenko, Klimenko, & Zakharenkova, 2009). These ions continue propagating away from the Earth’s surface and eventually enter the ionosphere, where they cause a measurable disturbance in electron density (Park & Dejnakarintra, 1973).
The present study focuses on the Haiti earthquake, which occurred on January 12, 2010. We hypothesize that there will be measurable variations in the electron density of the ionosphere during the months prior to this earthquake. The electron density data reported in this paper is taken from the Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions (DEMETER) satellite orbiting above Haiti. DEMETER carries many instruments, including an Instrument Sonde de Langmuir (ISL) probe. This ISL probe allows us to measure the electron density of the plasma and the electron temperature in the ionosphere, from which we can obtain the relative ion and electron density along the orbit of the DEMETER satellite.
Once electron density measurements are obtained, they are compared to an atmospheric model called the Parameterized Ionospheric Model (PIM) (Daniell et al., 1995). PIM draws upon several regional theoretical ionospheric models, allowing it to be computationally fast while maintaining accurate physics. It can produce electron densities for altitudes between 90 and 25,000km and is able to output the predicted electron density over time based on solar and geomagnetic parameters as well as latitude, longitude, and altitude data. However, sudden electron density variations caused by seismic activity are not represented in PIM models (Cornely, 2003), so according to our hypothesis, any significant difference between outputs provided by PIM and measurements from DEMETER which cannot be accounted for by solar or magnetic sources should be the result of seismic activity.
Having the ability to use the ionosphere as an indicator for seismic activity may allow scientists to predict earthquakes months ahead of time. This type advanced warning would make it possible to minimize damages and loss of life resulting from significant earthquakes.
MATERIALS & METHODS
Figure 1 presents a synthetic plot that illustrates our method for observing electron density variations. If our hypothesis is correct, there will be an observable difference in the two predictions. That difference will be the seismic indicator we are looking for. While this phenomenon could be observed by looking at electron density as a function of both latitude and longitude, only latitude data is presented here for the sake of simplicity.
Figure 1. Electron density in the ionosphere predicted by PIM and DEMETER data of electron density in the ionosphere plotted as a function of latitude. This is a simulated curve illustrates what we expect to see in the months leading up to the Haiti earthquake. The red curve simulates the electron density predicted by PIM, the blue curve simulates the data obtained from the DEMETER satellite, the green bar at the bottom corresponds to the location of Haiti, and the dotted line pinpoints the epicenter of the earthquake. The oval highlights the variation we expect to see when comparing the two curves. The existence of this kind of variation is the seismic indicator we are looking for.
The Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions (DEMETER) satellite orbits the earth roughly twice in every 24hr period. We obtained the electron density data from the satellite by downloading data files from the DEMETER website. Data from each orbit is contained in two files: a “burst” file and a “survey” file. The burst file contains data taken at a high sampling rate over a small distance, and the survey file contains the rest of the data taken at a lower sampling rate. This raw data is stored as binary information, and in order to make the data more user friendly we ran each file through IDL Virtual Machine (Exelis Visual Information Solutions, Boulder, CO), a computer program which converts binary files into ASCII files for easier processing.
Once in ASCII format, we processed the data files using MATLAB (The MathWorks Inc., Natick, MA, 2013). Electron density, latitude, longitude, and altitude data were extracted from each burst and survey file. The data set from each file was then combined to provide the electron density as a function of latitude and longitude as shown in Figure 2.
We then used another MATLAB script to process multiple DEMETER data files collected over a given period of time and displayed data from multiple orbits on a single graph. Using this ensemble method, we can observe general trends in the data for each of the months leading up to the earthquake and one month immediately after. Parameterized Ionospheric Model (PIM), using its inputs of solar and geomagnetic parameters, as well as latitude, longitude, and altitude, was used to predict the electron densities along the same path as the DEMETER orbit. An ensemble method was also used with the PIM data sets, so each graph contains outputs from multiple PIM predictions over the course of each month. In order to simplify the graphs and better observe potential electron density disturbances, electron density is presented only as a function of latitude.
Any variations between PIM prediction and values measured through the DEMETER satellite are not necessarily the result of seismic activity. Typically, other sources of ionospheric disturbances can emanate from solar and magnetic field effects. In order to isolate seismic activity as the primordial cause of the observed ionospheric disturbances, certain parameters need to be checked such as the sun spot number (SSN), the storm time index (Dst), the geomagnetic index (Kp), and the solar irradiance (or solar flux, F10.7). Using the National Oceanic and Atmospheric Administration’s SPIDR (SPIDR, 2012) website resources, for each month of interest, each parameter was checked to ensure it did not create any significant effects in the electron density in the months leading up to the Haiti earthquake.
Figure 2. Example MATLAB output of a single orbit of the DEMETER satellite. Data from each “burst” file (top left) is combined with data from each “survey” file (bottom) to make the final data set (top right.) The location of Haiti is indicated in green.
RESULTS
When the Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions (DEMETER) satellite measurements are compared to Parameterized Ionospheric Model (PIM) predictions during the months leading up to the earthquake, they appeared to have the same general spatial features everywhere except at around 20° latitude. For example, during two and three months before the earthquake (Figure 3a and 3b), both PIM and DEMETER data generally agree on the distribution of the electron density over Haiti at all latitudes. However, one month before the earthquake (Figure 3c), a small, yet definite, variation emerges. During the month prior to the earthquake (Figure 3d), the difference between PIM and DEMETER data is even more obvious. One month after the earthquake (Figure 3e), electron density returns to the behavior observed two and three months prior to the earthquake. These results demonstrate the electron density around 20° latitude was higher than expected over Haiti during the months leading up to the earthquake and higher than at any other time period, which suggests a correlation between seismic activity and electron density in the ionosphere.
In order to isolate seismic activity as the source of this variation in electron density, other common solar and magnetic parameters were also considered. These parameters include sun spot number (SSN), storm time index (Dst), geomagnetic index (Kp), and solar irradiance (or solar flux, F10.7). The solar flux and sun spot number are proxies for the solar extreme ultraviolet (EUV) flux (Chakrabbarty et al., 2012).
Solar EUV flux is the radiation that ionizes the neutral atmosphere to produce the ionosphere. These two indices correlate with the overall magnitude of the electron density in the ionosphere, but not with quiet or active conditions. SSN >80 and F10.7 >150 correspond to a higher than normal electron density. Generally, Kp 6 corresponds to active conditions, which are usually associated with geomagnetic sub-storms. When Kp is between four and six, auroral activity is moderate and the effects do not usually extend into mid-latitudes. High Kp usually means that the aurora extends in an equatorial direction and the ionosphere is usually not significantly affected. For example, a Kp of eight or nine is typically associated with full geomagnetic storms. The presence of a geomagnetic storm is indicated by Dst, which is typically a negative number and indicates how much the quiet time geomagnetic field is reduced by the ring current in the magnetosphere. If Dst is more negative than -100nT, it will typically create observable effects on the mid-latitude ionosphere. If it is more positive than -50nT, ionospheric effects are negligible. When Dst is between -50nT and -100nT, ionospheric effects will be highly variable depending on the time of day. Looking at these parameters using the SPIDR website, Dst, Kp, F10.7 and SSN were all within quiet or normal ranges during the time period of interest leading up to the earthquake, indicating that they had no significant effect on the ionospheric electron density.
DISCUSSION
The work presented in this paper addresses the problem of earthquake prediction by analyzing the electron density in the ionosphere during the months leading up to the 2010 Haiti earthquake. Comparing predicted electron densities from the Parameterized Ionospheric Model (PIM) and the measured electron densities collected by the Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions (DEMETER) satellite, there is an observable variation one and two months before the earthquake. Ruling out other common sources of electron density disturbance leaves us with seismic activity as the most likely source of this observed variation.
While the results of this work are promising there are also some limitations to this study. Using the DEMETER satellite for characterizing the ionospheric electron density is accurate, but it only provides information for the exact path of the satellite, leaving electron densities in other parts of the ionosphere unknown. Additionally, the DEMETER scientific mission ended in late 2010, so current DEMETER measurements are no longer available. While past DEMETER data can still be used for data analysis of seismic behaviors and their potential correlations with ionospheric disturbances, more ionospheric electron density data is required to study the real behavior of the background ionosphere near fault lines over earthquake prone regions such as Haiti.
Although additional work is required in order to establish a definitive system for earthquake prediction, the methods described open the potential for a predictor of seismic activity which could be effective as early as one or two months before the actual event occurs. Such a predictor would be of significant utility to earthquake prone countries.
ACKNOWLEDGEMENTS
This investigation is based on observations with the IAP experiment conducted on the DEMETER satellite, launched by Centre National D’etudes Spatiales (CNES). The authors thank J.J. Berthelier, the Principal Investigator of IAP for the use of the data.
REFERENCES
Chakrabbarty, D., Bagiya, M. S., Thampi S. V., & Lyer, K. N. Solar EUV flux (0.1-50nm), F10.7 cm flux, sunspot number and the total electron content in the crest region of equatorial ionization anomaly during the deep minimum between solar cycle 23 and 24. Indian Journal of Radio & Space Physics, 41, 110-120.
Cornely, P. (2003). Flexible Prior Models: Three-dimensional Ionospheric Tomography.Radio Science, 38(5), 1087.
Daniell, R. E. Jr., Brown, L. D., Anderson, D. N., Fox, M. W., Doherty, P. H., Decker, D. T., Sojka, J. J., & Schunk, R. W. (1995). Parameterized ionospheric model: a global ionospheric parameterization based on first principle models. Radio Science, 30(5), 1499-1510.
Demeter Data Server. (n.d.)., 2012, from http://demeter.cnrs-orleans.fr/
Freund, F. (2002). Charge generation and propagation in igneous rocks.
Journal of Geodynamics, 33 (4-5), 545-570.
Freund, F. (2010). Toward a unified solid state theory for pre-earthquake signals. Acta Geophysica, 58(5), 719-766.
IDL Virtual Machine. Exelis Visual Information Solutions, Boulder, CO
MATLAB version 8.2. The MathWorks Inc., Natick, MA, 2013.
Namgaladze, A. A., Klimenko, M. V., Klimenko, V. V., & Zakharenkova, I. E. (2009).
Physical mechanism and mathematical modeling of earthquake ionospheric precursors registered in Total Electron Content. Geomagnetism and Aeronomy, 49(2), 252–262.
Park, C. G., & Dejnakarintra, M. (1973). Penetration of thundercloud electric fields into the ionosphere and magnetosphere: 1. Middle and auroral latitudes. Journal of Geophysical Research, 78(28), 6623–6633.
Sharma, K., Das, R. M., Dabas, R. S, Pillai, K. G. M., Garg, S. C., & Mishra, A. K. (2008). Ionospheric precursors observed at low latitudes around the time of koyna earthquake. Advances in Space Research, 42(7), 1238–1245.
SPIDR. (n.d.). 2012, from http://spidr.ngdc.noaa.gov/spidr/