S43A-1034
Eruption seismicity of Bezymianny Volcano, Kamchatka, Russia
Seismic activity at Bezymianny volcano is currently being studied via a two-pronged approach. Real-time telemetered data from an increasingly dense short-period seismic network is analyzed by the Kamchatka Branch of Geophysical Services (KBGS; Russian Academy of Science Far East Division). This network is currently being supplemented by a campaign-style broadband array supported by the NSF Partners in Research and Education (PIRE) program. The combination of these networks is beginning to provide research-grade seismic data from Bezymianny's twice annual plinian eruptions. The St. Helens-like sector collapse and eruption of Bezymianny in 1956 (VEI 5) heralded the beginning of what has evolved into one of the most regular, and predictable, sequences of explosive eruptions in the world. Seismic data, as well as thermal remote sensing data, follow a repeatable pre-eruption sequence which has been exploited by KBGS to provide accurate forecasts days before impending eruptions. This remarkable pattern strongly suggests that the mechanics of recent Bezymianny eruptions are stable and long-lived. The implication of a robust, largely self-regulating system makes Bezymianny a preeminent target for discerning the mechanics of pre-eruption seismicity. Because Bezymianny is arguably the type example of a growing class of volcanoes that have experienced massive sector collapse followed by decades-long lava dome growth, the mechanics of Bezymianny may well be portable to numerous other volcanoes. http://gps.alaska.edu/PIRE/
S43A-1035
Small Repeating Earthquakes Found at Kusatsu-Shirane Volcano, Japan
Kusatsu-Shirane volcano is situated in the central part of Japan. Its major activity in the past 3000 years consists solely of recurrent phreatic explosions around the craters (Yugama, Mizugama and Karegama, etc). Recent explosions occurred in 1976, 1982, 1983 and 1989. Even now the volcano maintains fumarolic and seismic activity around the craters. We have monitored the seismic activity of the volcano with several stations including three borehole ones since 2001. Good signal-to-noise ratio data at borehole stations enable us to locate dominant earthquakes as small as M~ -2. Previous studies revealed two seismic clusters, one located under Yugama and Mizugama, and the other located to the south of these craters. It was identified at various tectonic settings, such as geothermal areas and subduction zones, that there exist repeating earthquakes (doublets and multiplets) which are repeatedly observed as similar waveforms. Such events are very useful to investigate small temporal changes in the crust which are caused by volcanic and/or tectonic activities, and can be an important tool to monitor volcanic activity. Therefore we search for seismic doublets and multiplets from very small earthquakes at Kusatsu-Shirane volcano. We analyze ~1600 earthquakes which occurred between July 2003 and May 2007. At each station, cross-correlation of seismic waveform is examined for all event pairs by using 2-second time window which includes P and S arrivals. Similar event pairs are picked up based on the maximum cross-correlation coefficient (≥ 0.8) at borehole stations and from these pairs are formed groups of repeating earthquakes if different pairs share the same event. We find three major groups (hereafter referred to as A, B and C) of repeating earthquakes which occur recurrently over the whole period and each group has more than 100 events. The groups A and B are found in the Yugama-Mizugama cluster. These groups are classified as high-frequency and low-frequency repeating earthquakes. Finally, the group C is found in the south cluster. The earthquakes are high-frequency events. These groups probably occur along the major magma-plumbing system under the volcano. We are also planning to incorporate results of geochemical observations such as fumarolic gas and lake water of the Yugama crater.
S43A-1036
A statistical study of volcano-tectonic, low frequency and hybrid earthquakes from Ethiopia
Volcano-tectonic, low-frequency, and hybrid earthquakes are anticipated near active volcanoes, but they are often overlooked in rift zones where dike intrusions may occur. A broadband seismic array in Ethiopia recorded seismicity from multiple dike injection events independently confirmed in geodetic data. We carry out statistical analyses of all events recorded on a 10-station array to automatically classify and characterize events over the 12-month deployment period. We correlate these events with deformation episodes recorded in GPS in InSAR data to understand the acoustic properties of the volcanic plumbing system beneath the rift zone.
S43A-1037
A Magmatic Origin for the 2007 Micro-Earthquake Swarms at Upptyppingar, Iceland?
Micro-seismicity is common within the rifting zones that intersect Iceland. However, between February and August 2007, anomalous swarms of tectonic earthquakes -- amounting to at least 3,350 epicentres -- have been detected over a 36 km2 area near to Upptyppingar (16.2° W; 65° N), which forms part of the Kverkfj~{o}ll volcano system, extending from the northern flank of the Vatnaj~{o}kull ice cap. Using results from the SIL seismic network, we summarise the spatial and temporal changes in ongoing seismicity that began in February 2007; additionally, we consider the processes underlying the escalated activity. Besides displaying spatial clustering, the Upptyppingar micro-earthquakes are noteworthy because (i) they concentrate at focal depths ranging from 15 to 18 km; (ii) the swarms comprise earthquakes <2 in magnitude; and (iii) several of the swarms originate at focal depths exceeding 18 km. Each swarm has been confined to a small surface area and focal depths have remained mostly consistent, both within and between swarms. Spatially, different parts of the affected region have exhibited seismicity at different times, with swarm sites alternating between distinct areas. Curiously, few earthquakes have been detected at shallow depths within the zone of unrest. A likely reason for the sustained clustering of earthquakes near to Upptyppingar is inflow of magma into the base of the crust; such an hypothesis would account for the depth, intensity, and duration of the swarms. Based on observations elsewhere in Iceland, a candidate explanation is the emplacement of a dyke. Interestingly, the up-surge in seismicity from the Upptyppingar region coincides with the ongoing formation of a 2,000 GL reservoir for water, located 21 km south-east of Upptyppingar. Nevertheless, the possibility of induced seismicity remains unclear. http://hraun.vedur.is/ja/englishweb/index.html
S43A-1038
Rockfalls at Augustine Volcano, Alaska: 2003-2006
Rockfalls, avalanches and landslides have been frequently recorded in seismic data at Augustine Volcano for many years. Typical years such as 2003 or 2004 had several dozen such events that were strong enough to trigger the automatic event detection system. Typical events lasted about 30 sec, had frequencies >6 Hz, and were strongest on summit stations, suggesting that they were rockfalls from the steep summit dome into the adjacent moat area. In 2005 both the rate and the occurrence pattern changed. Rockfall activity began in April 2006 and peaked in May and June, then continued through the fall and early winter. Overall there were more than 340 rockfalls in 2005, with both small and large events occurring. The 2005 rockfall activity increased at nearly the same time as earthquake activity and heating of the ground, suggesting that higher temperatures and steaming contributed to mechanical instabilities of the surface dome rocks. We examined relative amplitudes at station pairs and frequency contents to determine relative locations of the rockfalls by assuming that both higher amplitudes and higher frequencies are associated with events closer to a given station. When a low-light camera was installed at Augustine in January 2006 we were able to confirm these relations because there was a clear correlation between rockfalls, debris flows, and pyroclasic flows to the east (towards the camera) and high amplitudes and frequencies at east station AUE. Other events had high amplitudes and higher frequencies at west station AUW and no material was seen moving to the east. Still other events moved to the north and amplitudes were nearly the same at AUE and AUW. The systematic patterns in amplitude and frequency, verified by data from the low-light camera, make it possible to estimate mass flow in various directions using seismic data. Energy estimates of the rockfalls made from video images can be compared with energy estimates from magnitude-energy equations. The observer stations AUE and AUW show shifts in the frequency depending on whether the rockfalls are moving toward or away from them. Estimates of the seismic wave speeds from the rockfalls can be estimated using the Doppler equation, since the rockfalls are a moving frequency source. Also in progress is a program to estimate mass flow around the flanks of the volcano, using the amplitude ratios from stations around the volcano. The results from this work can be compared with geologic maps of deposits from the 2006 eruptions. The high rate of rockfalls in 2005 was also a new class of precursory signal that may be incorporated into long-term monitoring strategies at Augustine and elsewhere.
S43A-1039
Re-analysis of the largest 24 instrumentally recorded earthquakes since the beginning of the 20th Century
We have revised the size estimates, nucleation points, rupture areas and aftershock distributions of all the instrumentally recorded earthquakes with a seismic moment greater than or equal to 3.0E+28 dyne-cm, equivalent to a moment magnitude of 8.25. Re-analysis of these events was done in order for the United States Geological Survey (USGS) to publish a new series of global seismic maps based on the global seismicity compilation of Villasenor and Engdahl (2002). In addition, the USGS is using this information to provide better historical content and calibration information in its rapid response applications. We have compiled from the literature seismic moment estimates for all potential earthquakes with Mw greater or equal than 8.25, and we have selected a preferred value based on objective criteria. This has resulted in a total of 24 M8.25 or greater earthquakes in the time period 1905-2006. For these earthquakes we have then compiled focal mechanisms obtained using waveform modeling/inversion methods or inferred the mechanism from recent larger earthquakes that occurred in the vicinity of the event. We have relocated all the main shocks and aftershocks for which phase arrival time data are available from the ISS (International Seismological Summary) or ISC (International Seismological Centre). Relocation of the main event provides an estimate of the nucleation point, while the aftershock distribution is used to delineate the rupture zone. For earthquakes in the early part of the 20th century, rupture zones are also constrained by macroseismic data and surface effects (uplift or surface rupture). For more recent earthquakes we have constrained rupture estimates from finite-fault slip modeling. The 24 largest earthquakes analyzed in this study are all shallow-depth earthquakes. Nineteen of these earthquakes are best described as megathrust subduction events, 2 are outer-rise normal faulting events, 2 are intra-slab ruptures, and 1 is an intraplate strike-slip earthquake.
S43A-1040
Differential Energy Radiation from Two Earthquakes in Japan with Identical MW
Teleseismic studies have found that in general the radiated energy ES of intraplate strike-slip earthquakes is elevated significantly for a given rupture size (as measured by moment) relative to the energies radiated by thrust earthquakes at subduction zones. We verify that this phenomenon can be observed for regional (< 400 km) data by applying the spectral technique of Boatwright et al. (2002) for estimating radiated energies from S, Lg and surface waves to two earthquakes in Japan which had identical teleseismic MW but very different Me (energy magnitude). For the thrust-fault earthquake of 1996 Oct 19 offshore of Kyushu (MW 6.7), strong motion data from 154 K-net stations were used. A frequency dependent attenuation Q(f) = 230f 0.65 was derived with a resulting ES of 3.9e14 J. For the strike-slip Tottori earthquake of 2006 Oct 6 (MW 6.7), the attenuation Q(f) = 180 (f)0.6 was determined from 217 K-net stations with a resulting ES of 1.3e15 J. The attenuation functions derived for these earthquakes are comparable to those derived from other studies of earthquakes in Japan as well as central California for f > 1.0 Hz. The ES of the strike-slip earthquake is more than three times larger than that of the thrust earthquake. Commensurate with the difference in energies, the macroseismic effects as reported for the strike-slip earthquake were more severe and widespread than effects reported for the thrust earthquake. However, from teleseismic data the energy of the strike-slip earthquake is a factor of 10 larger than that of the thrust earthquake. For the strike-slip earthquake, the acceleration spectra of teleseismic and regional analyses overlap smoothly. Good regional-teleseismic overlap has been seen in analyses of other strike-slip earthquakes such as the Hector Mines event. In contrast, for the Kyushu earthquake, the teleseismic analysis appeared to underestimate energy at a band of frequencies about 1.0 Hz relative to the regional data. A similar difference between regional and teleseismic spectra has been observed for the North Ridge and San Simeon, California thrust events.
S43A-1041
Seismic Scaling Relations for the 2007 Off Mid Niigata Earthquake (Mw6.6) Sequence in Japan in Comparison with Two Other Earthquake (Mw6.6) Sequences
We carried out a spectral ratio analysis using the 2007 Off Mid Niigata prefecture earthquake (OMNE) sequence which occurred in the tectonic region similar to the 2004 Mid Niigata prefecture earthquake (MNE) sequence. The relationship of seismic moment (Mo) vs. corner frequency (fc) estimated for the OMNE sequence in the range of 3.4 ≤ Mw ≤ 6.6 is Mo \propto fc-3.87 which deviates from Mo \propto fc-3. The Mo - fc relation of the MNE sequence estimated previously was Mo \propto fc-3.41. Both of the source areas are located in the region where active thrust faulting, folding, and multi-planar faults are characteristic. However, the aftershock activity is different from each other, that is, the OMNE sequence consists of a small number of aftershocks while the MNE produced a large number of aftershocks including six large events with Mw ≥ 5.5. In the 2005 West Off Fukuoka prefecture earthquake sequence the Mo - fc relation is Mo \propto fc-3.06, and most of the aftershocks occurred on a single fault. There are low-velocity anomalies around the OMNE and MNE hypocenters determined in some seismic tomography models. The heterogeneous distribution of possible fluids may also be responsible for the deviation from the standard model in the scaling relations. The characteristics of lower fc's or stress drops of the small events of the OMNE sequence in 3.4 ≤ Mw ≤ 3.8 are similar to that of ¡Èoff-main fault¡É events of the MNE sequence. The difference (or variation) of the scaling relation for the small events can be interpreted in context with the event locations, i.e., whether they are on or off the main faults. The comparison of the scaling relationship between the on- and off-main fault events may have implications for seismogenic conditions that also results in different aftershock (or seismic) activity as described above.
S43A-1042
Hypocenter Relocations of Earthquakes in Central Southern Korea using the Double- Difference Method
The double-difference method is widely used for precise relocations of earthquakes to estimate the extent and attitude of seismogenic faults. In South Korea, it is very difficult to find the location and estimate the extent of seismogenic faults because large earthquakes are rare. Therefore, we applied the double-difference method to small earthquakes occurred in Central Southern Korea and try to find any evidence of well-defined fault planes. Latitude and longitude of earthquakes range from 35¢ªN to 37¢ªN and from 127¢ªE to 129¢ªE, respectively. The seismicity of this region is relatively higher and density of seismic stations is also higher than other regions in South Korea. We used 65 events in the magnitude (ML) 1.9-3.9 range occurred for the period from February 2001 to October 2007 in this region. We obtained the travel time data from 14 broad band and 42 short period seismic stations deployed by two Korean agencies, such as Korea Meteorological Administration (KMA) and Korea Institute of Geoscience and Mineral Resources (KIGAM). We determined P-wave travel times using manual phase picking and also travel time differences among P phases recorded at common station using waveform cross-correlation. The distribution of the earthquake relocations will be compared to the possible fault planes, which are constrained by regional moment tensor solutions.
S43A-1043
The Structure of the Kantishna Seismic Cluster, Central Alaska, Derived From Stress Tensor Inversions and Seismicity Trends
The Kantishna Cluster is an enigmatic and energetic cluster of earthquakes located in central Alaska, just to the northwest of Mt. McKinley/Denali and north of the Denali Fault. The Kantishna Cluster is located at a hub of tectonic activity including Bering Block rotation to the west, bookshelf faulting to the northeast, and rotation of southern Alaska due to Pacific plate convergence to the south. The Alaska Earthquake Information Center (AEIC) earthquake catalog locations show the cluster as an S-shaped cloud of earthquakes with no discernable structure. Double difference hypocentral relocations show that the cloud of earthquakes collapses into several distinct trends. Stress tensor inversions of events within the cluster show that the stress orientations vary systematically throughout the cluster. From stress tensor inversions, predominant faulting features can be determined for each trend in the cluster. Combining seismicity trends resolved from hypocentral relocations and faulting features determined from stress tensor inversions, it became possible to infer fault planes or planar structures in the region. The newly uncovered structures and a rotation of the maximum compressive stress throughout the cluster suggest that the Kantishna Cluster exists to accommodate interactions between all the tectonic structures surrounding the cluster.
S43A-1044
Seismotectonics of the Pitcairn Island 2001-2002 Earthquake Sequence
Continued investigation of the Pitcairn Island earthquake swarm in the southcentral Pacific Ocean is revealing new details about that remarkable sequence. The period of intense activity commenced in October 2001 and continued through January 2002, for which over 10,000 events have been cataloged with a battery of automatic signal detection algorithms. These algorithms include matched filter and waveform cross-correlation techniques that are coupled with statistical tests of the significance of the correlation coefficients. Relative amplitude scaling against a large master event (2001 November 23; 10:32:26.4 UTC; mb ~ 4.0) allows assignment of a magnitude value to smaller events. The size distribution of earthquakes calculated throughout the sequence appears to vary with time, with the slopes ranging as -1.3 ≥ b ≥ -3.0 on a scale of days. This variation may indicate the interaction of volcanic and tectonic processes. Cluster analysis with dendrograms suggests multiple distinct sources, consistent with previously reported results (Hyvernaud et al., 2002; Peyton et al., 2002). Although attempts at deriving moment tensors have not yet succeeded, modeling waveforms recorded at PTCN bracket the range of source depths as between 8 and 32 km.
S43A-1045
2006 Taiwan PingTung Aftershocks Locations and Preliminary Tomography from OBS Data
An array of 11 portable Ocean Bottom Seismometers (OBS) was deployed to collect aftershock sequences of two 6.9ML main-shocks occurred on December 26, 2006, off PingTung, Taiwan. One day after the main shocks, more than four hundreds aftershocks with magnitude larger than 2Mb were relocated from these OBSs records. According to the northern portion of the clustered aftershocks, the fault plane of the second main shock is trending NW-SE strike with a high angle westward dip. In contrast, the distribution of the southern portion of the aftershocks corresponds to a low angle westward dipping fault plane. Our preliminary tomographic results show that a high/low P-wave velocity boundary in NW-SE direction exist in the deeper offshore region and propagates to northeastward with increasing depth. It was supposed the high P-wave velocity zone can be associated with the subducting slab of the Eurasia plate.
S43A-1046
SEISMICITY AND PLATE GEOMETRY OF THE MENDOCINO TRIPLE JUNCTION REGION, NORTHERN CALIFORNIA (1986-2006)
The Mendocino triple junction is an unstable triple junction located at the convergence of the North American (NAP), Pacific, and Gorda lithospheric plates near Cape Mendocino, northern California. Thirty years of seismic data (1976 - 2006) was analyzed to characterize plate geometry and stress orientations affecting the region. A simple 3-D model of plate geometry was developed to separate earthquake events occurring in the NAP or Gorda plate. Upper- hemisphere focal mechanism solutions were hand plotted for select earthquake events in the study area. NAP solutions north of the MTJ are primarily reverse mechanisms. P and T axes for these solutions indicate north-northeast (~N45E) directed pressure consistent with geodetic and geologic data. NAP solutions south of the MTJ are strike-slip solutions consistent with northwest striking faults of the northern San Andreas fault zone. Gorda plate solutions east of the MTJ are primarily normal fault solutions indicating downslab tension in the subducting plate. East-west cross section plots indicate a very dense cloud of seismicity at the MTJ and demonstrate a shallow dip angle (~8° - 12°) for the top of the Gorda plate to longitude 123.5 W. East of this longitude dip angle increases and Gorda plate seismicity phases out. North-south cross sections show a clear gap in seismicity coincident with an interpreted southern extent of the Gorda plate at latitude 40.25 N as well as the aseismic nature of both the Pacific plate and northern San Andreas fault proper.
S43A-1047
Scanning for Unusual Seismicity in the Mendocino Triple Junction Region
Anomalous seismic activity with wide-ranging behavior has been detected in the vicinity of the Mendocino Triple Junction (MTJ) and offshore transform faults. Among those unusual earthquakes are non-volcanic tremors, repeating earthquakes and slow-rupture or low-stress-drop earthquakes. These unusual events together with ‘typical' earthquakes provide clues regarding the mechanics of faulting in the offshore region. We have compiled a catalog of unusual earthquakes looking for relationships between them, the fault structures and ‘typical' earthquakes. We present broadband observations of these events over a multiple year period ending in 2007 as well as their characteristics that allow one to detect and recognize them. One difficulty in the study of seismicity of the region is that events located far offshore may go undetected or they may have poor locations with large uncertainty. Another difficulty is that there is a class of events that have either low stress drop or are have slow rupture processes that make detection difficult. To improve our monitoring capability in the region we have implemented a low-frequency continuous waveform scanning method (Tajima et al., 2002 and Kawakatsu, 1998) to detect, and locate events in the offshore region, as well as compute the seismic moment tensor. Using waveforms of a few known slow earthquakes as references, we also perform a cross-correlation analysis over multiple years of bandpass filtered continuous seismic data recorded at several broadband stations in northern California with the goal of detecting unknown slow-rupture event similar in waveform to known slow events. We will show results of these wavefield scanning efforts in the Mendocino transform fault. Finally, the continuous scanning method we are implementing offers improved response time for rapid characterization of earthquake and tsunami hazard from offshore earthquakes.
S43A-1048
Seismic Documentation for Rock Damage and Heal on the San Andreas Fault Involved in the 2004 M6 Parkfield Earthquake
After the M6 Parkfield earthquake that occurred on 28 September 2004, we deployed a dense seismic array at the same sites as used in our experiment in the fall of 2002. The measurements using moving-window cross- correlation of waveforms for the repeated explosions and microearthquakes recorded in 2002 and 2004 show a decrease in shear velocity of at least ~2.5% within a ~200-m-wide zone across the San Andreas main fault trace most likely owing to co-seismic damage of fault rocks caused by dynamic rupture in this M6 earthquake. The width of the damage zone characterized by larger velocity changes is consistent with the low-velocity waveguide model on the SAF near Parkfield derived from fault-zone trapped waves [Li et al., 2004]. The estimated ratio between the P and S wave traveltime changes is 0.57 within the rupture zone and ~0.65 in the surrounding rocks, indicating wetter cracks within the damaged fault zone, probably due to the ground water percolating into the cracks opened in the mainshock. The measurements of traveltime changes for repeated aftershocks in 21 clusters, with a total of ~130 events, located at different depths along the rupture in 2004 show that the maximum shear velocity increased by ~1.2% within the damage zone in 3.5 months starting a week after the mainshock, indicating that the fault heals in the post-seismic stage due to the closure of cracks in the damaged rock. The data recorded at a seismograph installed in the SAFOD mainhole passing the San Andreas fault zone at ~3-km depths for repeated aftershocks in December of 2004 and later show that seismic velocities within the damage zone were changed by ~0.3% in a month, but no changes were registered at seismographs installed in the vertical pilot borehole drilled ~1.8 km away from the main fault trace for the same repeated events. We find that the healing rate is logarithmically decreasing through time with greater healing rate in the earlier stage after the mainshock. The magnitude of fault damage and healing varies along the rupture zone and with depth, but is most prominent at depths above ~7 km and roughly correlating with the slip distribution on the SAF in the 2004 M6 Parkfield earthquake [Johanson et al, 2006]. Observations of fault zone damage and healing associated with this M6 event illuminate the faulting-healing progression on an active fault in the major earthquake, in general consistent with previous observations of velocity evolution owing to damage and healing for Lander and Hector Mine earthquakes [Li et al., 1998; Vidale and Li, 2003]. However, the magnitude of damage and healing observed near Parkfield on the SAF is smaller than those on the Landers and Hector Mine rupture zones, probably related to the smaller magnitude mainshock, and smaller slip, and possibly differences in stress drop, pore-pressure, and rock type.
S43A-1049
The 1914 Reno Earthquakes–Major Events within the Reno Metropolitan Area
The 1914 Reno earthquakes are instructive events that serve as distinct warnings of future potential seismic damage. The earthquakes occurred in two major groups, one in February and one in April. The largest February event occurred mid-morning on the 18th and frightened people enough to cause them to rush into the streets in Reno, Sparks, and Virginia City. This earthquake caused significant nonstructural damage in Reno. The largest event (~M6.3 ±0.3) occurred on April 24th and followed the third largest event (a foreshock) by 28 hours. The main earthquake was a nighttime event and was felt from Berkeley, CA to Winnemucca, NV. Again people ran to the streets and there was nonstructural damage, but this time it was reported as far away as the Great Valley in California. The first detailed intensity maps for these earthquakes are based on 106 accounts we have gathered, and were used to estimate magnitudes. Some accounts hint at the possibility that two or more earthquake sources may have been active during this sequence; however, the main events appear to be located beneath Reno.
S43A-1050
Preliminary Use of the Seismo-Lineament Analysis Method (SLAM) to Investigate Seismogenic Faulting in the Grand Canyon Area, Northern Arizona
This is a progress report on our application of the Seismo-Lineament Analysis Method (SLAM) to the eastern Grand Canyon area of northern Arizona. SLAM is a new integrated method for identifying potentially seismogenic faults using earthquake focal-mechanism solutions, geomorphic analysis and field work. There are two nodal planes associated with any double-couple focal-mechanism solution, one of which is thought to coincide with the fault that produced the earthquake; the slip vector is normal to the other (auxiliary) plane. When no uncertainty in the orientation of the fault-plane solution is reported, we use the reported vertical and horizontal uncertainties in the focal location to define a tabular uncertainty volume whose orientation coincides with that of the fault-plane solution. The intersection of the uncertainty volume and the ground surface (represented by the DEM) is termed a seismo-lineament. An image of the DEM surface is illuminated perpendicular to the strike of the seismo- lineament to accentuate geomorphic features within the seismo-lineament that may be related to seismogenic faulting. This evaluation of structural geomorphology is repeated for several different azimuths and elevations of illumination. A map is compiled that includes possible geomorphic indicators of faulting as well as previously mapped faults within each seismo-lineament, constituting a set of hypotheses for the possible location of seismogenic fault segments that must be evaluated through fieldwork. A fault observed in the field that is located within a seismo-lineament, and that has an orientation and slip characteristics that are statistically compatible with the fault-plane solution, is considered potentially seismogenic. We compiled a digital elevation model (DEM) of the Grand Canyon area from published data sets. We used earthquake focal-mechanism solutions produced by David Brumbaugh (2005, BSSA, v. 95, p. 1561-1566) for five M > 3.5 events reported between 1989 and 1995. Fieldwork to test our hypotheses is planned for the spring of 2008. Based on a preliminary analysis, the northwest-trending Muav, Tipoff, Cremation and Phantom Faults are all located within northwest-trending seismo-lineaments associated with recorded earthquakes that are characterized by normal (slightly oblique) focal mechanism solutions. This is consistent with Brumbaugh's suggestion that present-day seismicity in the eastern Grand Canyon is concentrated on faults with a northwest trend.
S43A-1051
The Recent Seismicity Near Tabouk
Tabouk is a very populated historic city that was jolted by a 5.2 earthquake in June of 2004. Fortunately no damage has been reported. The seismic activity attracted the attention of residence as well as scientists and authorities in the region. The earthquake is located in a neogene/Quaternary volcanic region named Harat ar Raha about 140 km south west of Tabouk. Not far from the Red Sea and the Gulf of Aqaba, Tabouk is at the north western edge of the Arabian Shield. The main shock was followed by small magnitude (less than 4) aftershocks. Moreover the region remained active (many events with a magnitudes less than 2 has been recorded) every now and then until now. According to historic reports the city and the region surrounding it has been affected by a large magnitude earthquake in March of 1068 that caused about 20000 deaths in the region. The goal of this study is to map the seismicity and understand the source of the activity. The results would hopefully have an impact on hazard mitigation in the region. Realizing that more seismic stations were needed to better monitor the activities and locate the small events, we deployed 5 temporary short period and broadband seismic stations to the region. We have also collected waveform data from 23 broadband stations that are part of the National Seismic Network previously run by King Abdulaziz City for Science and Technology (KACST). Our early results show that the historic event location does not match with the current seismicity. In addition we have been able to present a new and more accurate seismicity map for the study region. This project has been fully funded by KACST and is planned to last until the end of 2008.
S43A-1052
Earthquake Locations in Southeastern Spain Obtained Using the COMPLOC Software
We present preliminary results of relocating over 9500 events in southeastern Spain from 1998 to 2006 using the COMPLOC earthquake location software (Lin and Shearer, 2006). This algorithm uses a robust misfit measure and improves the relative location accuracy among nearby earthquakes by computing source-specific station terms. COMPLOC has been tested mainly in southern California but is available for use in other regions. Because of the relatively sparse station coverage in southern Spain, we modified the programs to use regional Pn and Sn arrivals, in addition to crustal P and S picks. Determining the best reference 1-D velocity model for the region is challenging because nearby stations are lacking for most of the events and there are large tradeoffs between the event locations and the velocity model. Our preferred velocity model locates most of the seismicity between 5 and 15 km depth and yields locations that are reasonably robust with respect to changes in the maximum epicentral distance. Our locations exhibit greater clustering of seismicity than is seen in the standard catalog, which may permit improved identification of faults.
S43A-1053
Microseismicity Studies in Northern Baja California: The Sierra Juárez Fault System
The Sierra Juarez is one of the major, well defined, and active fault systems of northern Baja California. During two months of 2002, we installed 30 seismological stations (digital, three-components, Reftek instruments) between latitudes 31.6º N and 32.2º N, surrounding the most active part of this system as well as the SE segment of the San Miguel fault and the region in between. Almost half of the stations were installed in the Laguna Salada basin, located East of Sierra Juarez ranges and 1500 m below them. Observations resulted in 4200 high-quality hypocenter and ~500 focal mechanism determinations; magnitudes and rupture planes are still to be determined. For locating we use the Nava and Brune (1982) seismic structure, complemented with station residuals which are small and negative for stations located in the Sierra ranges. For stations installed in the Laguna Salada basin, residuals vary between 0.30s and -0.15s, with the exception of three sites where mean residuals reach -.50s. Seismic activity occurs either aligned (SE segment of the San Miguel fault) or in small clusters with radii ~1.5 km (elsewhere). Predominant depths are around 10 km with a secondary maximum at 5 km. Focal mechanism solutions show a consistent pattern which is common for all northern Baja California, with predominant strike-slip (a nodal plane striking in a NW-SE direction) and normal (T-axes mostly in EW direction) solutions. This pattern is interpreted as a transtensive regime consisting of strike-slip faults intercalated with extension zones; this pattern seems to be repeated at various scales. Also discussed are other details, like the interpretation of normal faulting in the scarp separating the Sierra Juarez ranges from the Laguna Salada basin, the dip of the fault planes, and interpretation of travel time residuals
S43A-1054
Stress Field and Seismicity in the Basin of Mexico
Mexico City is located in the basin of Mexico, inside the so called Trans-Mexican Volcanic Belt. The region in general and the basin in particular, is characterized by local low magnitude seismicity (Mc < 4.0) that may represent a risk to the city due of the nearness from epicenters. We can distinguish three main areas of local activity: 1).- surrounding the old basin of Texcoco lake, 2)- Chalco and 3)- Juchitepec – Milpa Alta outside Mexico City; the rest of the basin presents lower seismic activity. We recorded and located 336 earthquakes with digital seismograms between 1996 and 2007. From them, just 23 focal mechanisms could be evaluated because of low magnitude that creates recording problems in the seismological networks and high frequency background noise. The focal mechanisms are mainly strike–slip and dip–slip (normal) faulting. We used three different techniques (when possible) to calculate the focal mechanisms: simple and composite first motion focal mechanism, Hash's S/P amplitude rate focal mechanism and time domain moment tensor inversion using broadband three components seismograms. The final goal is to find the local and regional stress field for the whole basin.
S43A-1055
Microearthquake Observations in a 7-level Vertical Seismic Array in the TCDP Borehole, Taiwan
In order to obtain in-situ information on slip zones of the 1999 Chi-Chi earthquake, the Taiwan Chelungpu-fault drilling project (TCDP) drilled two vertical boreholes (A, B) and a branch hole (C) through the fault where a displacement of 12 m had occurred. The TCDP hole A is 2 km deep, and a slip zone was identified at a depth of 1111 m. Hole B (with side track, hole C) is 1.3 km deep with an identified slip zone at 1138 m. In July 2006, a 7- level vertical borehole seismic array (TCDP BHS) was installed in hole A covering a depth from 946 m to 1274 m with 50- 60 m depth intervals. For this layout, three seismometers were placed in the hanging wall and footwall, respectively. The forth one is located at the depth of 1110.28 m, close to the identified slip zone. Microearthquakes with magnitude down to -0.5 were detected by the TCDP BHS. A temporary seismic array with 10 short period seismometers around the TCDP drill site was also installed to incorporate with the TCDP BHS for the precise locations of the microearthquakes. A real-time location software (MIMO) (Oye and Roth, 2003) was used to automatically determine P- and S-wave onset times, incidence and azimuth angles and locations of the microearthquakes. Regardless of the large co-seismic slip of 12 m at the drill site during the 1999 Chi-Chi earthquake, our preliminary studies do not show any close-by seismicity near the drill site after almost 8 years since the large earthquake happened. The microearthquakes clustered at a depth of 8-10 km, where the 30 degree dipping of the Chelungpu thrust fault becomes flat to a decollement of the Taiwan fold-and-thrust tectonic structure. As a continuous GPS survey did not observe post-slip at the large slip region, and as no seismicity was observed near the drill site, we suggest that the thrust belt above the decollement during the interseismic period seems to be locked. A Fluid Injection Test (FIT), pumping high pressure fluid into hole B and C with hole A as observation well was carried out at the TCDP boreholes in November 2006, and January, March and April, 2007 to improve the understanding of the fault permeability (Ma et al., AGU 2007 abstract). We use the monitoring data from the TCDP BHS recorded during the FIT to examine potential seismicity changes associated with FIT and the observation of potentially triggered events. The source parameters of the microearthquakes were also investigated with regard to the understanding of scaling relationships between small and large earthquakes within earthquake physics.
S43A-1056
Cataloguing Seismic Waveform Properties Recorded With a 3D Network in a Gold Mine in South Africa
The SAVUKA gold mine is located in the northwestern edge of the Witwatersrand basin, a Late Archean (3.07- 2.71~Ga) intracratonic basin in South Africa that hosts the largest known gold-uranium-pyrite ore deposits in the world. Seismic events related to the mine activity span several orders of magnitude through a variety of sources that include mine blasts, pillar collapses, and faulting events. These events are systematically recorded and catalogued through an in-mine, 3D seismic network consisting of 20, three-component, short-period stations with natural frequencies ranging between 4.5 and 28.0~Hz and deployed as deep as ~3.5 km. After 5 months of seismic monitoring of the mine, we have been able to assemble a database of over 6000 events spanning magnitudes in the -2.5 < ML < 4.4 range. The potential of this unique data set for characterizing the detailed seismic properties of the basin and studying source properties of non-double couple events is explored through simple, first-pass analysis on the recorded waveforms. Moreover, the in-mine network is complemented by a small array of 4 broadband stations interspaced ~10~km apart on the surface of the mine, and by a number of AfricaArray stations in South Africa and neighboring countries located at regional distances (50- 1000~km) from the mine. The largest mine-induced events are clearly recorded at distances as far away from the mine as 450~km and provide a unique opportunity for studying the regional propagation of seismic phases as well as the structure of the cratonic crust underlying the basin.
S43A-1057
Recalibrating ML for the California Integrated Seismic Network
Since Richter (1935) and Gutenberg and Richter (1942) developed the local magnitude scale using records from Wood-Anderson seismographs, it has been used to describe event sizes in the catalogs of both northern and southern California (NC and SC). With each change in instrumentation and addition of a station, careful calibration procedures were necessary to ensure catalog continuity, although different amplitude decay functions (logA0) have been used in each region. In the past 10 years, many broadband stations and strong motion stations have been added to the networks in NC and SC, and efforts are underway to unify earthquake reporting for the state. We have collected data from 180 earthquakes in all of California, from nearly 1000 horizontal channels (described by station-network-channel-location, or SNCL) from the AZ, BK, CI and NC networks. Synthetic Wood-Anderson amplitudes were measured from each channel's trace by decovolving the instrument response and convolving the response of a Wood-Anderson seismograph (Uhrhammer et al, 1996). For each event, differences for all SNCL pairs were calculated, giving a dataset with more than 5 million observations. The differences were simultaneously inverted to determine a correction for each SNCL, dML, and a logA0 that is valid statewide. To ensure consistency with past magnitudes determined in NC and SC the inversion was constrained by two factors: (1) logA0(100 km) = -3; (2) dMLs for station BKS sum to zero. We have produced a set of dMLs and a logA0 function for distances from 1 to 500 km which should be valid in the entire state. We are validating the new values with existing events to ensure the consistency with current magnitude determinations.
S43A-1058
Analysis of Modern Techniques for Nuclear-test Yield Determination of NTS Events Using Data From the Leo Brady Seismic Network
Nuclear test detection is a challenging, but important task for treaty verification. Many techniques have been developed to discriminate between an explosion and an earthquake and if an explosion is detected, to determine its yield. Sandia National Laboratories (SNL) has maintained the Leo Brady Seismic Network (LBSN) since 1960 to record nuclear tests at the Nevada Test Site (NTS), providing a unique data set for yield determination. The LBSN is comprised of five permanent stations surrounding the NTS at regional distances, and data (in digital from post 1983) exists for almost all tests. Modern seismic data processing techniques can be used with this data to apply new methods to better determine the seismic yield. Using mb(Lg) we found that, when compared to published yields, our estimates were low for events over 100 kilotons (kt) and near the published value for events under 40 kt. We are currently measuring seismic-phase amplitudes, examining body- and surface-wave spectra and using seismic waveform modeling techniques to determine the seismic yield of NTS explosions using the waveforms from the LBSN.
S43A-1059
The Enhancement of the Stations ALMAR and ERMAR for Signal Detection and Seismic Phase Identification at Regional and Teleseismic Monitoring
This study is focused enhancement of the Mongolian borehole seismic array stations ALMAR and ERMAR. We determined most effective filter range for detection processing of signals for regional and teleseismic events and systematic deviation of azimuth and slowness values from theoretical ones for various regional and teleseismic phases for several seismic zones. In addition, we construct attenuation parameter this stations using amplitude and period measurements. In this study, we selected several regional and teleseismic events from IDC and seismic catalogue of RCAG, which are detected both ALMAR and ERMAR stations. We specially choose events large enough to be well located, so that we can expect to measure the errors in azimuth and slowness associated with the array analysis. In order to determine most effective filters for regional and teleseismic detection, we manually tune F-K analysis F-statistic parameter and noise, event spectrum comparison. For all events, we applied different filters for each phases and determine most suitable filters for each phases. Then we made several statistical analyses given results to be determining average applicable filter ranges. To be determining theoretical azimuth and slowness we used IASPEI model and TAUP software in MATLAB. To determine deviation of azimuth from theoretical ones for various teleseismic and regional phases, first, we calculate Delta azimuth for all events and plotted in geographical map, then we determine seismic zones, which are give azimuth deviation greater than 2 degrees. For example in case of ALMAR station: We found most effective filter range of teleseismic events for P phase is 0.3-1.5hz and regional events Pn, Pg and Lg phases 0.9-0.4hz, 0.8-3.5hz and 0.5-4.0hz respectively. For the biggest azimuth deviation found 280°-20° degree with value 6.1±4.6, 6.08±3.7 and 5.95±3.8 for Pn, Pg and Lg phases respectively.
S43A-1060
Large Earthquake Repeat on Normal Faults: Insights from dense in-situ 36Cl Exposure Dating of Limestone Fault Scarps, Central Apennines, Italy
The only chance to learn how major earthquakes have repeated in the past on major active faults is to analyze their surface geological record, if any. We analyze such a record on 11 active normal faults to seek identifying, dating and measuring the large earthquake ruptures that have broken these faults in the last 10-20 kyrs, and determine how these major ruptures have followed in space and time on each fault and from one fault to another. As large earthquakes repeat on a normal fault, the fault plane is progressively exhumed and exposed at the free air, forming an escarpment at the surface. Provided that this escarpment is preserved from erosion, its surface holds the complete record of the successive ruptures (and possible aseismic slip) as they have broken the ground surface. We have started to analyze such a record on 11 neighboring, likely interacting active faults in the Fucino area, Central Italy, where seismic activity can be devastating (1915, M7 Avezzano earthquake, 30 000 casualties). Faults offset limestone rocks and form several hundred meters high cumulative escarpments, whose youngest parts (10-20 kyrs) are well preserved in the form of 10-20 m high, steep scarps running along the fault lengths (10-20 km). The Holocene seismic slip history of the faults can be recovered from base-to-top continuous in-situ 36Cl exposure dating of the limestone scarps (Benedetti et al. 2002, 2003; Palumbo et al. 2004). To reach that objective, we have sampled the faults in two different ways: two faults (Magnola and Roccapreturo) were sampled at several, regularly spaced sites along their length, so that to recover the earthquake slip variability in both space and time. Nine other faults (Fiamigniano, Campo-Felice, Velino, Tre- Monti, Trasacco, Parasano, San Sebastiano, Castel di Ieri, Roccacasale) were sampled at one single spot along their length, so that to examine the possible earthquake interactions within the entire fault system. Doing so, we have collected one thousand samples (60-100 sample per site). We are now in the process of measuring the in situ 36Cl concentrations of these samples. Preliminary results will be shown, in comparison with previous data coming from paleoseismology and catalogues of historical earthquakes.
S43A-1061
Identifying Past Earthquakes on an Active Normal Fault (Magnola, Italy) from the Chemical Analysis of its Exhumed Carbonate Fault Plane
A normal fault scarp formed by repeated strong earthquakes is made of a series of rupture zones having been exposed, thus weathered, over significantly different time spans. We show that such differential weathering can be detected from the chemical analysis of the fault scarp rocks, and its signature used as a base to recover the past earthquake history (in the last 10-15 ka) of the fault. We focus on the Magnola normal fault, Central Italy, whose Holocene seismic slip history has been previously determined from in situ 36Cl surface exposure dating of the limestone fault scarp surface (Palumbo et al., 2004). Five major earthquakes were found to have occurred over the last 12 ka, with slips of 1.5-3 m and recurrence times of 0.7-4.8 ka. We analyze the major and trace element concentrations in 15 carbonate samples collected from base to top of the 10m-high Magnola Holocene scarp, at the emplacement of the Palumbo et al. previous sampling. We find that most element concentrations (70 %) decrease up-dip along the scarp, at a rate averaging 5%/m or 4.5%/ka. That decrease is attributed to both leaching and dissolution/recrystallization of purer calcite, together increasing with exposure time. Superimposed to the up-dip concentration decay, narrow zones of peak-concentrations (concentrations are increased by up to 100%) are found that coincide with the transition zones separating the successive earthquake ruptures. We suggest that those peak-concentrations result from calcite enrichment of the scarp sections which remained stuck in the impurity-doped, acidic upper soil during the quiescence periods that separated the earthquakes. The Rare Earth-Yttrium Elements (REE-Y) are among those most significantly enriched at the earthquake transition zones, which makes them the best chemical markers of the past seismic events. We propose a first-order numerical model that includes the two observed phenomena: a constant concentration decay rate in the exposed scarp rocks, and a constant enrichment concentration in the very upper soil. The model reproduces well the observations. Thus, provided that a limestone fault scarp is densely sampled so that to precisely determine the shape of the concentration-scarp height curves for a few REE-Y elements, the model might be inverted to recover, through iterative adjustments, the best-fitting earthquake slip exhumation history on the fault. Only a few rock exposure dating would then be needed to calibrate that history.
S43A-1062
Paleoseismology and Geomorphology of the Northern Calico Fault: Testing for Temporally Variable Slip Rate and Earthquake Clustering
The Calico-Blackwater is the longest fault system in the Mojave section of the Eastern California shear zone (ECSZ), and recent studies (Oskin et al., 2006a) indicate a long-term slip rate of 1.8±0.3 mm/yr for the Calico fault. Although this is the fastest slip rate among the six major faults that comprise the Mojave section of the ECSZ, the rupture history of the Calico fault remains unknown. In order to better understand the spatial and temporal pattern of seismic activity on the Calico fault, we excavated paleoseismic trenches across a desert playa in Newberry Springs, California (~30 km east of Barstow). We also measured small-scale offsets of geomorphic features along a 10 km stretch of the fault extending southward from the trench site. The trenches exposed fluvial, lacustrine, and playa sediments, consisting of pebble gravel and coarse-grained sands, silts, and clays, respectively. Several of these strata also displayed well-developed carbonate soils which help us establish rough constraints on the age of deposition. Preliminary field analysis of soil development suggests that the section exposed in the trenches encompasses latest Pleistocene through Holocene sediments (dating from ~15 or 20 ka). We see evidence for four surface ruptures during this time interval. Two of the ruptures appear to be latest Pleistocene in age; the penultimate event is latest Pleistocene or early Holocene in age and the most recent event (MRE) occurred during mid- or latest-Holocene time. In order to determine the absolute timing of all four ruptures, we have collected over two dozen samples from stratigraphic units throughout the trenches for optically stimulated luminescence dating. Our paleoseismic observations are consistent with geomorphic evidence from displaced alluvial surfaces cut by the Calico fault. Field observations of alluvial surfaces interpreted to be latest Pleistocene to early Holocene in age reveal only ~5.5 m of right-lateral offset of incised channels. Smaller offsets of bar-and-swale topography suggest that offset in the MRE along the northern Calico fault was only ~1.5-2 m. The occurrence of only two Holocene surface ruptures, coupled with the geomorphic evidence for only ~5.5 m of total slip during Holocene time, is in apparent conflict with the relatively rapid (1.8±0.3 mm/yr) long-term slip rate measured by Oskin et al. (2006a) over the past ~60,000 years. This discrepancy may result from: (1) temporal variability in slip rate across faults within the Mojave section of the ECSZ with slip rates on the Calico fault decreasing during the Holocene; or (2) the absence of a latest Holocene rupture event on the Calico fault, with the possibility of occurrence of a near-future earthquake as part of the ongoing ECSZ seismic cluster. We note, however, that even if a large-displacement event (e.g., 5 m of slip) were to occur tomorrow, the Holocene slip rate of the Calico fault would still only be ~1 mm/yr.
S43A-1063
The Energetics of Gravity Driven Faulting
Faulting can result from either of two different mechanisms. These involve fundamentally different energetics. In displacement-bounded faulting, locked-in elastic strain energy is transformed into seismic waves plus work done in the fault zone. Elastic rebound is an example of displacement-bounded faulting. In force-driven faulting, the forces that create the stress on the fault supply work or energy to the faulting process. Half of this energy is transformed into seismic waves plus work done in the fault zone and half goes into an increase in locked-in elastic strain. In displacement-bounded faulting the locked-in elastic strain drives slip on the fault. In force-driven faulting it stops slip on the fault. Tectonic stress is reasonably attributed to gravity acting on topography and the Earth's lateral density variations. This includes the thermal convection that ultimately drives plate tectonics. The gravity collapse seismic mechanism assumes the fault fails and slips in direct response to the gravitational tectonic stress. Gravity collapse is an example of force-driven faulting. In the simplest case, energy that is released from the gravitational potential of the topography and internal stress-causing density variations is equally split between the seismic waves plus work done in the fault zone and the increase in locked-in elastic strain. The release of gravitational potential energy requires a change in the Earth's density distribution. Gravitational body forces are solely dependent on density so a change in the density distribution requires a change in the body forces. This implies the existence of volumetric body-force displacements. The volumetric body-force displacements are in addition to displacements generated by slip on the fault. They must exist if gravity participates in the energetics of the faulting process. From the perspective of gravitational tectonics, the gravity collapse mechanism is direct and simple. The related mechanics are a little more subtle. If gravity is not deliberately and explicitly included in an earthquake model, then gravity is locked out of the energetics of the model. The earthquake model (but not necessarily the physical reality) is then elastic rebound.
S43A-1064
Analysis of Aftershock Activity in Stable Continental Regions: Implications for Aftershock Forecasting after Strong Earthquakes in the CEUS
Statistical forecasts of the probabilities of future aftershocks after a large earthquake have become routine in California, an area where the average Omori-law properties of aftershock sequences have been determined from a large number of aftershock sequences. Unfortunately, there are not enough strong earthquakes with aftershock sequences in eastern North America for robust determinations of the average aftershock properties for that region. Thus, for the application forecasting in eastern North America, the aftershock properties from earthquakes in stable continental regions (SCRs) globally have been determined under the assumption that the aftershock behavior in all SCRs is comparable. The aftershock activity from 28 SCR earthquakes of Mw 6.0 and greater has been compiled, and the Omori-law parameters of this data set have been determined. Some SCR mainshocks have been followed by more active aftershock sequences, while some have been followed by relatively inactive aftershock sequences. Although the variation in SCR aftershock activity from one event to another is quite large, the average Omori-law aftershock properties are similar to those determined for California. These properties can be used as a basis for statistical forecasts of aftershock activity after strong earthquakes in eastern North America.
S43A-1065
The Ground Tilt in the 1999 Chi-Chi, Taiwan Earthquake
The baseline of a strong-motion record is very sensitive to the leveling change of the accelerometer. A tilting angle x (in radian) will cause a baseline drift about 980 sin(x) cm/s/s in the horizontal component and about 980(1- cos(x)) cm/s/s. in the vertical component. The effect of this baseline drift can be further magnified in a displacement waveform. Although the tilting is not the only source of baseline drifts, the baseline drift due to the tilting can be easily identified and separated from other types of baseline drifts because the tilting always appears in middle of a record and happens during or after the large shaking. The length of the surface rupture in the 1999 Chi-Chi Taiwan earthquake is about 80km with large ground motions; the maximum recorded ground velocity is over 3m/s, and the maximum ground displacement is over 10m. Some recordings near the surface rupture show significant baseline drift due to the ground tilt. In this study, we measure the ground tilt due to the Chi-Chi earthquake at 167 strong-motion stations. Among them, several stations are only few tens meters away from the surface rupture. As expected, the baseline drift of vertical component is smaller than that of the horizontal component. In general, the ground tilt is small no matter the fact that the permanent displacements are large at hanging-wall stations; the maximum tilting angel is about 0.5 degree while the most stations are less than 0.02 degree.