Overview | Earthquake | Drought | Fire | Flood & Cyclones | Landslide & Soil Erosion | Volcano
Detecting Earthquake Precursor: A Thermal Remote Sensing ApproachArun K. Saraf*, Swapnamita Choudhury1, Vineeta Rawat*, Priyanka Banerjee2, Sudipta Dasgupta3 and J.D. Das** *Department of Earth Sciences **Department of Earthquake Engineering Indian Institute of Technology Roorkee Roorkee – 247667, India saraffes@iitr.ernet.in
Abstract
Seismologists have not yet found a reliable earthquake precursor. In order to look for a reliable earthquake precursor a remote sensing based thermal technique has been employed recently based on the concept that stress accumulated in rocks in tectonically active regions may be manifested as temperature variation through a process of energy transformation. Rise in land surface temperature (LST) before an impending earthquake has been detected for 23 earthquakes using pre- and post-earthquake datasets of thermal sensors such as NOAA-AVHRR, Terra/Aqua-MODIS and SSM/I for the different parts of India and the world. Analysis revealed a transient short-term thermal rise in LST ranging from 2-12° C around epicentral areas of different earthquakes. In the studied earthquakes, pre-earthquake thermal anomalies generally started developing about 1-10 days prior to the main event depending upon the magnitude and focal depth, and then disappeared just before the main shock. This temperature increment prior to an impending earthquake can also be attributed to degassing from rocks under stress and/or to p-hole activation in stressed rock volume and their further recombination at the rock-air interface. A precise correlation of LST maps of Bam (26 Dec 2003, Iran) and Zarand (22 Feb 2005, Iran) with InSAR generated deformation maps; published by other workers (Stramondo et al., 2005a, 2005b, Parsons, 2005); also provides evidence that the thermal anomaly is ground related phenomena, not an atmospheric one. 1.0 Introduction Satellite thermal remote sensing has emerged as a potential tool in detection of pre-earthquake thermal infrared (TIR) anomaly in land surface temperature in and around epicentral regions. Thermal data in seismic studies was first put to application in Russia in 1985 (Tronin, 2000) and results were published (Gorny et al., 1988).This paper presents observations made by post-earthquake analysis of 23 past earthquakes occurred in different parts of the world and using thermal data from different satellites (Choudhury, 2005, Saraf and Choudhury, 2005a, 2005b, 2005c and Panda et al. 2007). Bam earthquake, Iran (26 Dec 2003) and Bhuj earthquake, India (26 Jan 2001) are two classic examples and are discussed here in detail. The Bam earthquake (table 1) brought devastation over the SE Iran, killing about 31,000 people, injuring 75,600 and rendering 75,600 homeless (Choudhury et al., 2006). In case of Bhuj earthquake (table 1) the death toll was estimated about 20,083 according to Gujarat Government figures and was accompanied by the wide scale damage to property and economy (Saraf and Choudhury, 2005a). 2.0 Concept of Thermal Infrared Anomaly Before an imminent earthquake earth’s crust passes through an earthquake preparatory phase. Accumulation of stresses and resultant pressure built up leads to the rise in LST. The enhanced TIR emission from earth’s surface retrieved by satellites prior to earthquakes is known as “thermal anomaly” (Freund et al., 2005).
*Earthquakes here discussed in detail. There may be various physical explanations for thermal anomalies appearing before an impending earthquake. However, authors prefer to explain the physical mechanism of thermal anomaly mainly by two leading theories: Earth-degassing Theory (Zu-ji et al., 1999) and p-hole activation Theory (Freund 2000, 2002 and 2003). During the preparatory phase of an earthquake, in high tectonic stress, pore spaces in the rocks are reduced due to subsurface pressure releasing the trapped gases. These gases on reaching earth’s surface are already at an elevated temperature with respect to the air temperature due to subsurface geothermal heat and secondly, it also creates a local greenhouse effect on the land surface thus serving as the source of outgoing anomalous radiation. Positive thermal anomalies at a regional scale were observed with the examination of around 9000 thermal images for the Middle Asian earthquake. These anomalies were attributed to the green house effect that was caused before the earthquake by an increase in gases like CO2, CH4 etc (Tronin, 1996). A new theory of charge generation in rocks prior to earthquakes is given by Freund (2000, 2002 and 2003). This theory keeps parity with laboratory experiments (Qiang et al. 1997, Ozounov and Freund 2004) and also provides explanation for other observed geophysical precursors. Electronic charge carriers can be free electrons or sites of electron-deficiency in the crystal structure, the latter known as p-holes (Freund, 2000). These p-holes in their dormant state are in form of Positive Hole Pairs (PHPs). A PHP is devoid of any charge, as outer shells of both of the two O- atoms are filled. Introduction of PHPs in minerals during rock-genesis and alteration has been explained explicitly by Freund (2002). Interestingly, O- -O- bond-distance (1.5A°) is almost half of O2–-O2- bond distance (2.8–3.0A°). It implies that the peroxy-bound O- has a small partial molar volume, thus having a tendency to be favored by high pressure. Being dissociated under high-accumulated stress a PHP introduces two holes (charge deficiencies) into the valence bond to attain electrostatic stability, causing the insulator to become a p-type semiconductor. Positive holes propagate though an oxide or silicate structure by electron hopping and immediately rush towards the Earth’s surface with high speed. Immediately after reaching a non-solid medium like the atmosphere or water, a positive hole acquires an electron to become O2- from O- . The recombination leads to vibrationally highly excited O-O bonds. These O-O bonds can de-excite either by emitting specific mid-IR photons in the 11-12 µm region or by channeling the excess energy into neighbouring Si-O and Al-O bonds, which in turn will emit in the 8-10 µm region (Freund et al., 2005, 2007). The observation and the spectral signature of the emitted radiation provides strong evidence that the underlying effect is a kind of mid IR luminescence arising from the recombination of p-holes at the rock surface. The energy therefore emitted by this electron acquirement increases the surface temperature. Several Remote Sensing Rock Mechanics (RSRM) (Wu et al, 2000) experiments and other studies (Pellet et al., 2007) conducted to predict material behavior on stress application and time-space forecast of rock failure have validated the infrared emission from the rocks under stress. However, it should be noticed that this heat is starkly different from the frictional heat that develops at the fault surfaces during rupture. This frictional heat, that takes a lot more time to come up to the surface, actually develops at the time of the earthquake itself, and hence has no contribution towards the pre-earthquake thermal anomaly (Banerjee, 2007). 3.0 Data and Methodology Day and nighttime High Resolution Picture Transmission (HRPT), Global Area Coverage (GAC), Local Area coverage (LAC); AVHRR datasets available from Indian Institute of Technology Roorkee-Satellite Earth Station, India (IITR-SES) and from National Environmental Satellite Data and Information Service (NESDIS) website; were mainly used in this study. Data from other satellites thermal sensors like Moderate Resolution Imaging Spectroradiometer (MODIS) on board Terra/Aqua satellites, Special Sensor Microwave Imager (SSM/I) on board DMSP satellites (SSM/I-DMSP) were also used. NOAA-HRPT has a spatial resolution of 1.1 km and a temperature resolution of 0.5° C. LST time series maps were prepared for day and night data sets. Data calibration and temperature calculation is based on the Becker and Li (1990) split window algorithm, which uses the differential absorption effect in channel 4 and 5 of NOAA-AVHRR. A user specified temperature range for daytime as well as nighttime scenes was specified in each case. Temperature outside this range was masked. Cloud covers were delineated and avoided for any temperature calculation. Image co-registration and correction for different satellite view angles were not done since any image overlay analysis was not intended. Wherever, available air temperature data for previous and subsequent years along with the year of earthquake occurrence was collected and Temperature Variation Curves (TVCs) and Normal Curves were generated. A comparison has shown that the difference of LST and air temperature is within +/- 1° C. In the present study air temperature data has been used to support LST analysis. This means that just based on air temperature data such study of pre-earthquake thermal anomaly would not be possible logically from spatial point of view (Choudhury et al., 2006). 4.0 Observations Both nighttime and daytime datasets were utilized for the study and an almost consistent pattern in the temporal and spatial development of thermal IR anomaly was observed. The two case studies are here discussed in detail. 4.1 Bam earthquake, Iran (26 Dec 2003, Mw 6.6) Analyses of the daytime and nighttime datasets for the Bam showed a distinct anomaly in LST which appeared before the main shock (table 2 and figure 1). The temperature increase was about 5-7° C than the usual temperature of the region. At some places, the temperature was about 6-10° C higher than the normal temperature of the region in that period of the year. In the nighttime maps, it was seen that on 18 Dec 2003 there was a complete normal temperature regime in the region. The appearance of an intense thermal anomaly was seen around the earthquake epicenter near Bam on 21 Dec 2003 (data was not available on 19 and 20 Dec 2003 for the same time of acquisition). If this rise had started earlier and then reached its peak on 21 Dec, it could not be ascertained through the gap in the data. To bring continuity to this gap, data acquired on 18 and 19 Dec 2003 night at 16:30-18:15 UTC hours, with a difference of four hours, was analyzed. A thermally anomalous region had developed near the epicenter on 19 Dec 2003, which rose to the maximum amplitude of thermal anomaly on 21 Dec 2003. The temperature was around 7-13° C on 21 Dec 2003 (around 7-8° C higher than normal temperature in the region around that period of the year). On 22 Dec 2003, the temperature was less than this boost. The temperature was normal again on 23 Dec 2003. The earthquake came on 26 Dec 2003, five days after the boost in LST as observed through nighttime data. Daytime LST time series maps show that the rise in temperature started on 22 Dec 2003. The anomaly stayed on till 24 Dec 2003 (just two days before the earthquake). The normal temperature was around 22-25° C on 21 Dec 2003. On 24 Dec 2003, the temperature was around 29-32° C (about 7-10° C higher than usual temperature of the region around that period of the year). Analysis and similar processing of nighttime NOAA-AVHRR data of the year 2004 acquired at around the same time and on the same days as the 2003 data, showed that there was no such abnormal behavior of the LST on those days in that year. The anomalous region in the nighttime data on 21 Dec 2003 occupied an area of about 308,000 km2 and in the daytime LST map of 24 Dec 2003 covered an area of about 328,200 km2. The InSAR co-seismic surface deformation studied by Stramondo et al. (2005a, 2005b) of the same Bam earthquake, revealed the existence of an unmapped fault 4 km west of Bam fault, the only active tectonic feature in the epicentral area (figure 2). Analysis of the hypocenter distribution of aftershocks recorded by a temporal seismic network installed 6 weeks after the main shock revealed source of earthquake as the “Arg-e-Bam” fault (Nakamura et al., 2004). The unwrapped interferometric phase onto the SAR amplitude image displays the surface displacement pattern, extends to around 30 km in N-S and to around 20 km in the E-W direction (figure 2b). It was observed that co-seismic InSAR surface deformation filed is in concurrence with the pre-seismic thermal anomalous area developed to the east of the epicenter on the 26 Dec 2003. This analysis indicates a plausible relationship between the two and suggests that LST pre-earthquake phenomena are land related. 4.2 Bhuj Earthquake, India (26 Jan 2001, Mw 7.9) On 14 Jan 2001, S-W Gujarat, with respect to the surrounding region started to pick up an increase in temperature. Within a few days there was a spread and increase of the thermal anomaly. The area experiencing this anomaly had spread in a NW-SE direction. On 23 Jan 2001 there developed an anomaly, which had an area of around 300 km2, SE of earthquake epicenter (table 3 and figure 3).  Figure 1: Daytime NOAA-AVHRR LST time series map of Iran before and after the earthquake in Bam, Iran on 26 Dec 2003 (Choudhury, 2005). Red colour filled star symbol indicates the day of the earthquake, whereas without colour filled star signifies location of the impending earthquake epicenter of 26 Dec 2003.  Figure 2: A comparison of thermal anomaly field of 26 Dec 2003 derived from nighttime NOAA-AVHRR image (a) and surface displacement field derived by unwrapping interferometric phase onto the SAR amplitude image (b). The black line in (b) indicates the SAR inferred earthquake fault location. Red star indicates the location of Bam earthquake epicenter. Succeeding this boost, the increase started to wane out, after 23 Jan 2001, just three days before the earthquake of 26 Jan 2001. On 23 Jan, temperature was maximum between 28ºC - 31ºC in the anomalous region (about 5º C - 7º C higher than the normal temperature). On 28 Jan 2001, the anomaly disappeared and the region showed a normal temperature of around 24º C. In order to ascertain that the thermal anomaly discussed above was induced by the 26 Jan 2001 earthquake, same period satellite data from NOAA-AVHRR for the year 2003 of the same area were analyzed. Further, it was observed that during the January 2003 the thermal regime of the study area shows completely normal pattern. Available weekly average temperature data (between 1951-1980) of epicentral area and surrounding region clearly shows that in the third week, the temperature trend touches the lowest temperature as compared to second and fourth weeks of the above years (figure 4). Whereas, in the year 2001, a peak was observed in the third week (instead in the second or fourth weeks). This is in contrary to the observed trend for the past thirty years (1951-1980). This ground observation is in accord with the observed satellite thermal detection during January 2001. Table 2: Details of nighttime and daytime NOAA-AVHRR data (acquired by IITR-SES) of the year 2003 used to prepare LST time series maps as shown in figure 1 to study pre-earthquake thermal anomaly of 26 Dec 2003 Bam Earthquake.
5.0 Discussion and Conclusions The post-event investigation of pre-earthquake thermal anomalies by analyzing pre- and post-earthquake LST images reveals valuable information about the changes in the thermal IR regime of the affected area. The understanding of the buildup of pre-earthquake TIR anomalies and their satellite detection provides possibilities of a reliable and potential precursor. On the basis of this study we can conclude that earthquakes with moderate to strong magnitudes may be preceded by the detectable TIR anomaly. So far studied 23 recent past earthquakes have showed temporal, short-term TIR anomalies before the event. The analyses of time series LST maps for these earthquakes showed a 2-13°C rise in LST 1-10 days before the earthquake struck (table 4). It was also noticed that magnitude and focal depth play a vital role in intensity and spatial extent of the thermal anomaly. Higher earthquake magnitude and shallower focal depth are favorable conditions for the appearance of intense thermal anomaly with larger spatial extent and vice versa.  Figure 3: Daytime NOAA-AVHRR LST time series map of Gujarat, India before and after the earthquake in Bhuj, India on 26 Jan 2001 (Saraf and Choudhury, 2005a).  Figure 4: Weekly average temperature variations in the second, third and fourth weeks (Julian weeks) of the years from 1951 to 1980 (data source: IMD 1991) (Saraf and Choudhury, 2005a). Table 3: Details of nighttime and daytime NOAA-AVHRR data (acquired by IITR-SES) of the year 2001 used to prepare LST time series maps as shown in figure 3 to study pre-earthquake thermal anomaly of 26 Jan 2001 Bhuj Earthquake.
In case of cloud-free weather conditions this difference may be attributed to differential heating contrary to differential cooling. As such, the diurnal thermal regime may not be conformable at daytime or nighttime. But it was a regular observation that the anomalies appear first in nighttime thermal images. A prominent observation regarding the earthquakes of moderate magnitude, i.e. less than magnitude 6, is the appearance of a dual TIR peak in surface temperatures instead of single rise observed previously. In case of Kerman earthquake, Iran (21 Aug 2003) the first peak appears about ten days before the earthquake and second temperature peak relatively closer to the main shock i.e. 2-6 days (table 4). This may lead us to understand that perhaps the energy accumulated in the stressed rocks have been released sporadically in form of electromagnetic emission, apparent temperature increment or any other geophysical earthquake precursor, which in turn might reduce the magnitude of the main shock. Study of other earthquakes with prominent aftershock events reveals that the occurrence of aftershocks prevents the re-establishment of normal conditions even after the main event is over. Due to prevailing residual stresses, the epicenter and adjoining areas experience aftershocks and thus the disappearance of the IR anomaly takes longer time. The observation of common field for TIR anomaly and surface displacement of Bam is a significant contribution. The concurrences of both the fields provide enough reason to believe plausible relationship between surface deformations and appearance of the thermal IR anomaly. The observation substantiates the explanations for pre-earthquake TIR anomalies given above. This is an important contribution to the knowledge of pre-earthquake thermal IR anomaly phenomena and for developing it as reliable and potential earthquake precursors. Table 4: Characteristics of the short-term pre-earthquake thermal anomalies associated with the Iran earthquakes studied through NOAA-AVHRR and other thermal data sets.
Acknowledgement: We are greatly indebted to Department of Science and Technology (Seismological Divison) (now Ministry of Earth Sciences), New Delhi for financial assistance, National Institute of Oceanography (NIO), Goa, and India Meteorological Department (IMD), Ahmedabad for providing valuable data for this study. References
Page 1 of 1
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||