July/August 2004

Surface Faulting: A New Paradigm for the Pacific Northwest

The past twenty years have witnessed three major paradigm shifts in our understanding of earthquake hazards in the Pacific Northwest. The first paradigm shift was the recognition in the 1980's that the Cascadia subduction zone is capable of great earthquakes. The locked part of the Cascadia subduction zone lacks instrumental seismicity, but geological evidence shows that it can produce an earthquake of MW 9, larger than any expected for California south of Cape Mendocino. In response to the geological evidence, building codes have been upgraded, tsunami inundation maps have been prepared for coastal areas from northern California to northern Washington, and earthquake hazard maps have been constructed for northwestern cities, including Victoria, British Columbia. Earthquake awareness has increased among the general public and within many private companies.

The second paradigm shift, the alignment of seismic provisions in building codes to reflect greatly increased scientific understanding, has been subtle but of overarching societal importance. This alignment began in the early 1990's when Washington, under the leadership of the Structural Engineers Association of Washington, expanded the Uniform Building Code seismic zone III to include the Mount St. Helens seismic zone, and Oregon adopted a greatly revamped designation of seismic zone maps. This trend toward direct use of scientific results in the building code has accelerated in the last two years, with the State of Washington adopting the International Building Code (IBC). The IBC uses the U.S. Geological Survey (USGS) probabilistic hazard map as its basis, thus aligning current best science with engineering practice. Oregon may follow suit; currently 32 states have adopted the IBC. The IBC literally brings USGS hazard calculations onto the desktops of structural engineers.

Despite the paradigm shifts related to Cascadia earthquakes and the increased reliance on scientific hazard assessments, crustal earthquakes in the Pacific Northwest have remained a problem. In part this is because the record of crustal earthquakes is sparse. Several medium-size earthquakes within the North American crust accounted for nearly all of the losses in Oregon, including the 1936 Milton-Freewater earthquake, the 1993 Scotts Mills earthquake, and the 1993 Klamath Falls earthquakes. Of these, one of the Klamath Falls events was the largest, M 6. A crustal earthquake of estimated M 6.8 in 1872 struck a thinly settled part of eastern Washington. None of these crustal earthquakes had intensities higher than MMI VIII, although the 1872 event had a limited number of intensity reports.

In the early 1990's, geoscientists, using coastal uplift records, recognized that the Seattle Fault, striking approximately east-west for 80 km from the city of Bremerton across the Puget Lowland beneath Seattle and Bellevue, was the source of a large earthquake of close to magnitude 7 accompanied by a tsunami in Puget Sound about a thousand years ago. At the time, however, no surface exposures of this fault were known, and it appeared that even though a repeat earthquake would produce high intensities, the "millennium earthquake" might have been a single, rare event.

None of the historical crustal earthquakes had surface rupture, and west of the Cascades, only one crustal fault with Holocene surface rupture had been found north of the California border. None of the faults in the Puget Sound region could be defined clearly enough that geoscientists and engineers could offer guidance to local government on how best to regulate building construction and land use in crustal fault zones. The Final Environmental Impact Statement (FEIS) for a proposed wastewater treatment plant north of Seattle illustrates the current standard used in consulting practice for siting large projects. This FEIS, issued in late 2003, states that "[b]ecause seismic hazards associated with ground rupture have historically been difficult to assess in the Puget lowlands, regulatory agencies have not included fault rupture in their hazard maps." Hazard analyses commonly consider only a "[r]andom crustal event that could occur in the upper 20 miles anywhere in the region."

In contrast to California, deposition and erosion related to the Cordilleran continental glacier strongly modified the landscape in the Puget Lowland of Washington and the Fraser River delta of British Columbia as recently as 14,000 years ago. Any fault scarps that might have formed since the glaciers melted away are masked by dense forest and underbrush. Now, the dense veil of forest vegetation is being parted by a new method of aerial mapping called LiDAR (Light Detection And Ranging) or Airborne Laser Swath Mapping, in which laser beams from a light aircraft penetrate the forest canopy and image the ground beneath.

The first indication of the power of LiDAR imaging occurred when the Kitsap County Public Utility District commissioned a LiDAR survey of Bainbridge Island, directly across Puget Sound from the city of Seattle. The LiDAR survey of this densely forested island revealed an east-west scarp parallel to the blind Seattle Fault but with the south side down, opposite to the presumed sense of displacement on the Seattle Fault. This scarp, named the Toe Jam Hill Fault, was trenched by Alan Nelson, Brian Sherrod, and several colleagues with USGS and other organizations, revealing evidence for three and possibly four surface ruptures on this fault between 2,500 and 1,000 years ago.

The discovery of the Toe Jam Hill Fault led to the establishment of the Puget Sound LiDAR Consortium, funded by partnerships between USGS, NASA, local government agencies, and other parties in western Washington. LiDAR surveys commissioned by this consortium have revealed more than a dozen scarps that might have formed during postglacial faulting. The list of faults for which trenching studies have documented at least one episode of surface rupture includes the Seattle Fault system, the Tacoma Fault, the Frigid Creek Fault, the Utsalady Point Fault, and the Canyon River Fault. The list of potential fault scarps waiting for field study is at least twice as long, but with only one full-time USGS geologist available to slog through the underbrush and verify the LiDAR scarp interpretations, progress will be slow.

Two recent fault studies have shown the increasing sophistication of USGS in jointly interpreting aeromagnetic surveys and LiDAR images. For many years, geophysical imaging suggested the existence of the Tacoma Fault, but without surface exposure, its exact location was unknown. Aeromagnetic mapping combined with shallow crustal seismic imaging from the Seismic Hazard Investigations in Puget Sound (SHIPS) experiments identified a narrow band of deformed rock west of Tacoma where the fault might be located. LiDAR revealed a possible fault scarp coincident with the aeromagnetic and velocity anomalies. Subsequent trenching showed that the scarp imaged by LiDAR is a late Holocene trace of the Tacoma Fault. Although geophysical data show an extension of this fault eastward into densely populated areas, the presence of surface faulting there has not been established from LiDAR or geological studies. LiDAR images for Tacoma should be available sometime in 2004.

A second recent use of LiDAR and aeromagnetic data is on the possible southeastern extension of the southern Whidbey Island Fault in the northern Puget Lowland. This zone of faulting, comprising several strands, was mapped across Whidbey Island by interpolation of faults imaged by marine seismic surveys. LiDAR images of southern Whidbey Island did not reveal young fault scarps, but differential elevation changes across one strand of this fault in southern Whidbey Island were interpreted by Harvey Kelsey of Humboldt State University as evidence for an earthquake 2,800-3,200 years ago.

In the summer of 2003, USGS scientists performed a detailed analysis of aeromagnetic data along the possible southeastern extension of the southern Whidbey Island Fault. They found an aeromagnetic lineament about 20 km long that appears to be the extension of one of the strands of this fault. This lineament, however, is less prominent than others in southern Snohomish County that appear to be more likely candidates for an extension of the southern Whidbey Island Fault. LiDAR imagery became available for a portion of the aeromagnetic study area in the summer of 2003, and at least one possible late Holocene scarp coincides with one of the aeromagnetic anomalies. Whether this scarp is related to tectonic, glacial, or other causes will be tested with trenching later this year.

Evidence for Holocene crustal earthquakes in the Puget Lowland is supported by GPS measurements under the Pacific Northwest Geodetic Array consortium (PANGA) and the Western Canada Deformation Array. Stéphane Mazzotti and his colleagues at the Pacific Geoscience Centre in Sidney, British Columbia found evidence that southern Washington state is moving northward toward the Coast Mountains of British Columbia at a rate of 2 to 4 mm/yr. This motion is compressing the Puget Sound region nearly as fast as the GPS-documented contraction across the Los Angeles Basin between the San Gabriel Mountains and the Peninsular Ranges-California Continental Borderland. Although the tectonic setting is different in the Pacific Northwest, the lesson from the Los Angeles area is that basin contraction has produced damaging urban earthquakes in 1971, 1987, 1991, and 1994, the largest having a moment magnitude of 6.7. All of these earthquakes were accompanied by secondary ground failure, and one was accompanied by surface rupture.

For people living in the urban corridor between Eugene, Oregon and Vancouver, British Columbia, the LiDAR results from Puget Sound are sobering. Urban crustal faults, if they produce relatively frequent earthquakes every thousand years or so, dominate earthquake hazard calculations used for most building codes and design practice. Their identification brings strong ground motion and the possibility of surface rupture, with all of the expected effects, into urban environments. It is very likely that the third paradigm, the importance of characterizing urban crustal faults, is now beginning to unfold in western Washington and northwestern Oregon.

It is clear that LiDAR surveys can rapidly produce inventories of possible Holocene surface faults. Completing LiDAR surveys in northern Puget Sound and the Willamette Valley, and across the Vancouver-Victoria, British Columbia urban area should be a top priority. With fault inventories in hand, a more complete assessment of ground motions and potential surface rupture would be available to help guide societal discussion regarding both existing and future development in urban crustal fault zones.

The importance of completing LiDAR surveys quickly to allow an urban fault inventory is evident by looking at damaging earthquakes elsewhere. Crustal faulting in the Puget Lowland could subject structures to MMI intensities as high as X and horizontal peak ground accelerations of 1 g, as strong as experienced in the 1994 Northridge and 1995 Kobe earthquakes. An illustration of the importance of surface rupture was provided by the 1999 Chi-Chi, Taiwan earthquake, which resulted in heavy loss of life and property due to surface rupture, especially in Fengyuan City, near the north end of the fault rupture.

Is there surface faulting elsewhere in the urbanized Pacific Northwest? Portland has a problem similar to Puget Sound's in that late Quaternary surface processes, in this case the Missoula floods, reshaped large portions of the landscape as recently as 12,000 years ago. The Portland Hills Fault extends through downtown Portland. High-resolution seismic surveys and a backhoe trench provide evidence suggestive of Holocene surface deformation, but as yet there is no documented Holocene surface rupture. Rick Blakely of USGS used aeromagnetic surveys to locate the Portland Hills Fault as well as the Mount Angel Fault (the probable source fault for the 1993 Scotts Mills earthquake), although the aeromagnetic models of these faults are more ambiguous than those in Puget Sound. USGS, in partnership with the city of Portland, completed a small pilot LiDAR survey across the Portland Hills fault in early 2004. LiDAR surveys in the Vancouver metropolitan area and on Vancouver Island, the site of a M 7.3 crustal earthquake in 1946, are clearly overdue.

The discovery of Holocene surface rupture in western Washington indicates that the hazard from shallow crustal faulting--high-intensity shaking and surface rupture--is a major hazard to the Northwest, as it is in California. Earthquake scientists have a responsibility to safeguard the public by pushing to complete our evaluation of crustal faulting hazards quickly and by working to move this understanding into state and local practices that advance earthquake hazard mitigation.

Robert S. Yeats
Earth Consultants International
1654 NW Crest Place
Corvallis, OR 97330-1812
Department of Geosciences
Oregon State University
104 Wilkinson Hall
Corvallis, OR 97331-5506

Craig Weaver
U.S. Geological Survey
University of Washington
Mail Stop 351310
Seattle, WA 98195-1310

To send a letter to the editor regarding this opinion or to write your own opinion, contact the SRL editor by e-mail.



Posted: 23 July 2005