Scientific & Technical Reports - The 2002 Denali Fault earthquake
Note: As of Friday, new updates will be posted to separate pages. Look here for links.
originally compiled by Mary Lou Zoback, USGS, Menlo Park
(last updated 07-November-2002 4:40 pm MST)
Note: This update begins with a general summary of the present state of knowledge, and then follows with more detailed scientific observations, preliminary results, and a cursory summary of current field deployments and modeling efforts.
One of the largest recorded earthquakes to strike the U.S. mainland struck central Alaska on Sunday, November 3, at 1:12 PM local time. The epicenter of this magnitude 7.9 earthquake was approximately 75 miles (135 km) south of Fairbanks and 176 miles (283 km) north of Anchorage (USGS National Earthquake Information Center-NEIC location: 63.520N, 147.530W; Alaska Earthquake information Center (Univ. of Alaska)-AEIC location: 63.5175N, 147.4440W). It was preceded by a magnitude 6.7 foreshock in the early morning of October 23, 2002. This earlier earthquake and its zone of associated aftershocks were located slightly to the west of the Sunday's quake. Figure 1 and Figure 2 give two views of the location of these two earthquakes and some of their aftershocks. The 1964 M9.2 Good Friday earthquake caused considerable damage in southern Alaska, however, it was centered offshore in Prince William Sound.
The image files above have some place names added, see basic figures on web: Figure 1 from NEIC; Fig. 2 from AEIC, http://www.aeic.alaska.edu/Seis/Denali_Fault_2002/
The earthquake resulted from slip on the Denali fault, an arcuate strike-slip fault that stretches over 700 km across the State of Alaska and extends southeastward into Canada. (see Figure 2). The Denali fault is one of the longest continental faults in the world and rivals California's San Andreas Fault in size. The eastern part of the fault shows about 400 km of right-lateral strike-slip displacement since early Tertiary time. Offset surficial deposits and seismicity indicate that the fault has remained active to the present, but the November 3 earthquake is the largest seismic event so far recorded on the fault. The November 3 quake was larger than the 1906 San Francisco earthquake, which had a magnitude of 7.8.
Analysis of the teleseismic waveforms indicate that this very large event, like many other of this size, was a complex one. Modeling by M. Kikuchi, and Y. Yamanaka (Earthquake Research Institute, Univ. Tokyo) suggests that M7.9 event is a doublet, with an ~M7.2 thrust earthquake initially (a large earthquake in its own right), followed ~10-15 sec later by an ~M7.8 strike slip event that ruptured to the east. Their modeling also indicates a peak displacement of 12 m along the strike-slip fault plane at depth.
Aftershock locations and geologic observations also indicate that the earthquake ruptured to the east, toward the Canadian border, for at least 300 km (to 142.58W according to geologic observations, see below). Preliminary field observations suggest dominantly right-lateral movement with a relatively small thrust component, generally north side up. On Monday, November 4, Rod Combellick and Melanie Werdon of the Alaskan geological survey (DGGS) observed the fault from the air and found no evidence of surface rupture in epicentral area or to the west (based on the preliminary earthquake locations). On Monday and Tuesday, Peter Haeussler (USGS, Anchorage) and Patty Craw (DGGS) flew along the Denali fault east of the epicentral area. They reported that the fault offsets increased to the east, and at approximately 144.5° W longitude the rupture jumped onto the Totschunda fault, consistent with the aftershock locations. They measured a maximum right-lateral offset of 6.9 m at the Tok Cutoff Rd, and saw, from the air, an offset of 12-15 m little bit to the east of there-this needs to be checked out on the ground. More details on offset observations are given below.
The earthquake ruptured across the Trans-Alaska Alyeska pipeline (TAPS). Although some supports for the pipeline were knocked out by the quake, causing the pipeline to sag; it did not break and alarm systems worked as designed. See attached aerial photo taken by Rod Combellick, DGGS,on November 4 (Figure 3) where the pipeline crosses the fault, looking west. Note that the pipeline is supported by rails on which the line can move freely in the event of fault offset. Workers are building a temporary foundation for a support that slid of the end of one of the rails. Note the transverse crack on the Richardson Highway in lower left. The fault runs from upper right to lower left, although it is not clearly visible here.
The pipeline was shutdown about 2:00 PM on Sunday as a precautionary measure. U.S. Geological Survey seismologists and geologists, serving on a Federal task force, were instrumental in insuring that the Alaska Pipeline was designed and built to withstand the effects of a magnitude 8.0 earthquake with up to 20 feet of slip across it. Although these design standards were considered to be excessively conservative at the time, the USGS guidance proved to be on target, and the resilience of the pipeline to Sunday’s fault rupture is a testament to the importance of science in engineering design and hazard mitigation .
The earthquake was felt throughout southern Alaska. As of Wednesday, November 6, 2785 Alaskans had reported shaking levels to the USGS Community Internet Intensity map (Did You Feel It? Map) ranging from weak, Intensity III, to very strong, Intensity VII, see http://pasadena.wr.usgs.gov/shake/ak/STORE/X22614036/ciim_display.html ).
PLEASE NOTE: IF YOU ARE PLANNING ANY FIELD DEPLOYMENTS TO STUDY THE EARTHQUAKE, WE REQUEST THAT YOU PLEASE CONTACT DONNA EBERHART-PHILLIPS, USGS Anchorage, and make her aware of your plans. She is in contact frequent contact with DGGS and UAF, hence knows of their plans as well. Contact information for Donna:
Donna Eberhart-Phillips, USGS Anchorage
More detailed information follows under the following headings.
The USGS is contracting for aerial photos of the rupture to be flown beginning Wednesday, November 6. It is critical to collect this valuable dataset before snow blankets the region and before ephemeral features, such as on glaciers, disappear. Plans are for 1:6,000 scale color stereo photographs, 60% overlap, single swath, uncontrolled. Ultimate coverage depends on funds available.
The DGGS, UAF, USGS, and Kerry Sieh (Caltech) and Charlie Rubin (E Washington State U) are cooperating on ground investigations. Kerry, Charlie and George Plafker (USGS retired, George created a strip map for the Denali fault in 1976) are flying up on Wednesday, November 6 and plan on 3 days of field work using helicopters.
USGS Menlo Park and USGS Golden are sending geologistst to Alaska to assist DGGS and others in the field reconnaissance of the fault. They are on hold, pending availability of funds. Two geologists from Menlo Park will begin reconnaissance tomorrow on Nov 7.
The GIS group in Menlo Park is currently digitizing the fault traces from Plafker's Denali strip maps done in 1976 at 1:63,360 (true inch to a mile). These files will hopefully be available in the next few days.
(made 11/4 by Rod Combellick and Melanie Werdon, DGGS, and on 11/4-5/02 by Peter Haeussler, USGS Anchorage, and Patty Crowe, DGGS. Summary below based on reports provided by Rod Combellick and Donna Eberhart-Phillips, based on phone conversations with Peter, liquefaction information from Peter posted 11/7)
- On 11/4 Patty Craw and Peter Haessler began their reconnaissance at the Richardson Highway and TAPS (pipeline), near the middle of the rupture area, and worked east to the vicinity of Mentasta Pass, where they were turned back by weather. In addition to the observations below, they tried to follow the rupture for a little ways west of the Richardson Highway, toward the epicenter of the 7.9 shock, but lost sight of the trace.
- Melanie Werdon and Rod Combellick flew the fault in a fixed wing aircraft from Cantwell to the Susitna glacier where we were turned back by weather. They saw a great many rock falls and snow slides but no clear evidence of surface rupture, even though this was the region of both the October 23 foreshock and November 3 main shock. They had to stay rather high because of severe turbulence, and most likely missed seeing the surface rupture as a result.
- The Richardson Hwy offered the best set of piercing points for measuring offsets. The three painted lines on the highway were offset 265, 255, and 257 cm. There was also vertical motion, 62 cm with North side up. North side up vertical offsets were found in other places, as well.
- They could not see the fault trace in the unconsolidated sediments of the Delta River. However, the fault showed really well as a break across several glaciers, including the Black Rapids, Canwell, and others. Shear cracks and tension cracks were seen in the ice, all indicating right-lateral shear. A lack of piercing points made measurement of the offset difficult to impossible on the glaciers.
- There were some very large landslides west of the Richardson Hwy. Peter estimated that one involved ~0.25 km**3 of material. This slide went down the slope, across a valley, and part way up the opposite slope.
- The pipeline was damaged where the fault ruptured across it. It appears that when they built the pipeline, they mislocated the fault. The pipeline crossing has a number of rails on the ground to allow the pipeline to slide and shift laterally. However, the fault trace crossed the pipeline around the southernmost of these rails rather than in the middle of that section.
- A highly linear stream located at 144 04.039' W exhibited 5.2 meters of right-lateral offset.
- In many places, they observed anelastic deformation away from the fault trace. In addition to the offset at the fault, when you step back you can see a "drag" of distributed deformation across the fault zone. In general, this zone of distributed shear could be 60 m or more wide. This means that the surface offsets Peter mentioned are likely to be minimum slip for the surface. No estimates of the magnitude of the distributed shear were reported initially, but it was enough to be clearly visible.
- At Tok Highway there was 6.9 m of right-lateral offset and around 0.5 m of vertical offset, N side up.
- At approximately 144.5W the rupture jumped on to the Totschunda fault. The furthest SE extent of the rupture on the Totschunda is at 62 degrees 20.0 minutes and 142 degrees 34.7 minutes. Peter measured an offest of 65 cm RL movement and 1.2 m of thrust movement, south up (near the eastern end of the rupture, 62 20.05N 142 35.75W). The villages of Nabesna and Slana are near the trace of the Totschunda.
- From Peter Haeussler: Report from
Bill Perkins and Frank Wuttig of Shannon
and Wilson in
Fairbanks last night 11/6/02. (Shannon and Wilson link goes to their web site for this earthquake)
More observations of liquefaction at the N end of Fielding Lake
Northway has extensive liquefaction. Boils in the ditches surrounding the
airstrip, cracking and lateral spreading of the airstrip. Boils all along
the 1.25 mile long runway. A 7 ft diameter, 4 foot deep sinkhole in one
A lodge owner in Northway reported water spurting 4 feet into the air.
There was differential movement of the foundation of a lodge.
There was a telephone pole that sunk 3 feet into the ground leaving the
supporting guy wires slack.
There were frozen sand blows.
E of Tok there was liquefaction along the highway. At mile 1258 there was a
S of Mentasta Lodge, there was spreading 4 ft vertical and 6 feet lateral.
The bridge abutment at the Slana River had moved due to liquefaction.
Alaska State Department of Transportation damage photos
Sigrún Hreinsdóttir and Evelyn Price from the Geophysical Institute at University of Alaska, Fairbanks observed sand blows, cracks, and "fountain features" on or next to the Northway Road, East of Tok, Alaska (general area 62.98N, 141.9W, see their photos at (http://www.aeic.alaska.edu/input/sigrun/sprunga2/northway/northway.html). They found cracks in 11 major areas along the road. They also reported evidence of water/mud/sand spraying up from the ground, including a baseball field that looked “like frozen geyser area.
Bill Perkins from Shannon & Wilson, Seattle and the Shannon & Wilson Fairbanks office have been visiting sites damaged during M7.9 earthquake. Photos from their post-earthquake reconnaissance are just coming in, and can be found on their web sit: http://clients.shanwil.net/project.php?projectid=Fairbanks_Quake_2002). This site will be updated frequently.
NSF is sending Nick Sitar (UC Berkeley) and Rob Kayen (USGS) to observe liquefaction features.
Randy Gibson and Ed Harp (USGS Golden) and David Kiefer (USGS Menlo Park) are in Anchorage and hope to use fix-winged aircraft to observe landslide features associated with the earthquake beginning Thursday, Nov. 7.
The location of aftershocks has been slow because of the shear volume of events overwhelming the staff at both AEIC and NEIC. As shown on Figure 2, the aftershock define an east to SE-trending fault plane and suggest rupture for at least 300 km along strike from the epicenter.
The earthquake was recorded by more than 30 strong motion stations in Anchorage: strong motions stations installed and maintained by Niren Biswas from UAF, 8 new ANSS stations, and several USGS NSMP stations. A total of 39 strong motion stations are run by NSMP in Alaska, most likely recorded the event. This includes 3 instrumented buildings in Anchorage: Hilton Hotel, Anchorage Regional Hospital, and the BP building. Some of these data are currently available on their web page: http://nsmp.wr.usgs.gov. The rest are being retrieved from the instruments.
In addition, 5 strong motion instruments (Guralps) and 4 broadbands with Reftek digitizers were installed in the epicentral region on 10/24 or 10/25 by Roger Hansen's group from UAF. As of Wednesday morning, Nov. 6, the data have not been retrieved, owing to some technical problems. However, when available these recordings will provide critical information about near-fault ground motions. UAF is deploying 20 IRIS PASCAL refteks. The USGS working in conjunction with UAF will be installing at least 3 refteks and 5-10 strong motion instruments, all with accelerometer sensors.
Strong motion stations at 6 Alyeska Pipeline pumping stations recorded the event, the closest station was less than 3 miles from the fault in an area where the surface offset is about 2.6 m right-lateral and about 0.6 m vertical. These data will probably be made available through UAF and the USGS.
Greg Anderson, USGS Pasadena, has done a preliminary model for Coulomb stress transfer from the 23 Oct M6.7 Nenana Mountain. foreshock to the 3 Nov M7.9 Denali Park event. His initial results indicate that the M6.7 event increased the stress in the vicinity of the 3 November mainshock, hence encouraging failure (increase of ~0.1-0.2 MPa (1-2 bars) of Coulomb stress ). He used a dominantly strike-slip source model for the 6.7 event (rake=162°, from Harvard CMT) and considered both a thrust and strike-slip fault plane for the initiation of the M7.9 event. He warns that these results are very preliminary, the source model he used for the M6.7 event is basic, a better one is neededto do a better job. Also, if the hypocentral location are not very accurate (errors on the order of 5 km or more) than the results may change significantly.
The geodetic community is working together to assure that anticipated large post-seismic transients will be well recorded. Jeff Freymuller and students from UAF had already deployed a number of continuous GPS stations in the vicinity of the October 23 epicenter. They are continuing to receive GPS units from UNAVCO and actively deploying them in the field
On Wednesday, USGS Menlo Park is sending 3 people to monitor postseismic deformation at four sites along the Denali rupture, two along the Richardson Highway near Paxson (one on either side of the fault) and two on the Tok Cutoff Highway near Mentasta Pass (one on either side of the fault). Ideally they would like the stations to be 15 to 20 km from the fault to capture the postseismic relaxation, which according to modeling by Fred Pollitz, USGS Menlo Park, is best measured within 30 km of the fault.
This relaxation should be in opposite directions on either side of the fault. Thus taking the differential displacement between stations on either side of the fault should both amplify the signal and decrease the noise associated with determining absolute position (e.g., relative to Fairbanks).
Ben Pauk of USGS Anchorage was unsuccessful in deploying 4 GPS on benchmarks installed in 1991 near Paxson and reoccupied only a couple years ago. One benchmark has disappeared, apparently the victim of road maintenance. Severe ice on the road precluded Ben from getting to the other 3 benchmarks. He drove on to University of Alaska in Fairbanks with the 4 GPS receivers to add to the pool there..
The USGS has a campaign geodetic profile across the Totschunda fault. The ~50 km long profile runs roughly perpendicular to the fault, at about 61.75 N, 142.0 W. Currently the AEIC aftershock locations and Peter's observations extend the rupture to about 142.5 W, so very close to this profile. This profile was surveyed in the mid-1980's by geodelite, it is a reoccupation of a 1906 triangulation boundary survey. Lisowski et al. (JGR, 1987, v. 92, p. 11, 522-560) reported a small amount of nearly pure right-lateral strain accumulation across the fault over this 80 year period. This profile is in a fairly remote region, not easily accessible now, but certainly should be resurveyed next summer.
Keith Howard, USGS Menlo Park, has spoken with Simon Hook at JPL who
is requesting ASTER images of the earthquake rupture area. The ASTER
satellite collects hyper-spectral imagery, which in the visible
wavelength provides stereo coverage with a resolution of about 15 m.
The swath width is 60 km. It also takes lower resolution data in the
thermal infrared range. It is Keith's understanding that the
satellite passes over this area daily around 10:30 AM. These images
will be very useful for landslide mapping.
Rick Saltus provided this:
Regional gravity and aeromagnetic data are sensitive to three-dimensional variations in physical properties (density and magnetic susceptibility) of subsurface rock units. Both data sets show an interesting change in trends near the epicentral region. A significant gravity gradient, indicating a shallow crustal boundary between dense rocks to the northwest and lower density rocks to the southeast, trends eastnortheast-westsouthwest, passes just to the north of Cantwell, and wraps around sharply to the northwest further east of the epicenter. To the west of Cantwell, this gravity gradient curves to the south and follows the Totschunda fault. The physical property boundary reflected by the gravity gradient appears to be related to fault patterns in the region. A distinctive zone of shallow-source, high-amplitude, aeromagnetic highs passes through Cantwell and closely follows the direction of the gravity gradient in that region. To the east, an en-echelon magnetic low has the same trend and follows the Totschunda fault. In the epicentral region a complex aeromagnetic low appears to reflect crossing east-west and northeast-southeast trends. These linear magnetic lows could be reflecting weathering zones along crustal faults. Interactions between fluids and rock in these zones could have destroyed magnetic minerals such as magnetite. Discrete aeromagnetic highs that parallel fault trends in the region could be caused by magmatic intrusions that have been emplaced preferentially in the fault zones.
Zhong Lu, USGS Anchorage, Chuck Wicks, USGS Menlo Park, and Tim Wright, Oxford University, have looked into InSAR coverage available to study the two earthquakes. There are numerous pre-event images, the satellite made a pas,s and is making many passes in the
next month, between the foreshock and the aftershock. It is hoped that this post earthquake pass will occur before snow blankets the region.
They are planning to look (along with many others undoubtedly) at three main issues: 1)co-seismic deformation on Denali fault; 2) deformation over the neighboring faults; and 3) deformation of volcanoes due to the earthquake. Zhong Lu has ordered some scenes from the Alaska SAR Facility to get a sense on radar coherence over the area.
From Tim Wright:
In terms of future acquisitions, we have high hopes for two overpasses on
Friday and another two on Monday. Whether we get any results depends on 1.
if the satellite orbital parameters are suitable; 2. what the ground cover
is like (how much trees and snow etc...).
Some of the photos show areas with little snow, and some have a lot. For
InSAR, it would be useful to have information in the field on the amount
of snow there is about at the moment, and where that snow is (e.g. if
there was less snow in the east then we could focus our attentions there,
or if heavy snow comes in on Thursday then we will know not to expect much
from Friday's acquisitions).
The likelihood is, however, that we won't get decent coseismic
interferograms until next summer, when the snow has melted. In the mean
time, we have ordered images to test for coherence.
The quake also created a ripple effect thousands of miles away. The passing seismic waves triggered a series of microearthquakes (M<3) about 2,000 miles away in at the Geysers geothermal area in northern California and at Yellowstone National Park in Wyoming. Mammoth Mountain, Coso,
and possiblly the south moat of Long Valley caldera also responded with bursts of seismicity. In the New Orleans area more than 3,000 miles away residents saw water slosh about as a result of the quake’s awesome power. Similar observations were recorded in Seattle, Minnesota, and Louisiana.