top of page

Clastic Dikes in Eastern Washington: Formed by Lateral Spreading?

Lateral spreads. A network of lateral spread fractures in pavement formed by an earthquake. Peru 1970. USGS photo archive.

Lateral Spreading: A proposed origin for sheeted clastic dikes in Eastern Washington

Earthquake-triggered lateral spreading is invoked to explain the many thousands of sheeted Pleistocene clastic dikes in south-central Washington (Cooley et al., 1996; Neill et al., 1997; Pogue, 1998). Previous articles have not provided a coherent model, but I imagine the explanation would go something like this: Seismic shaking triggers consolidation, opens tension cracks, and causes sliding of blocks down low angle ramps. The subvertical cracks are filled with rubbly material collapsed from crack walls. Spreading occurs similarly in bedrock so rubble-filled cracks (unconsolidated sediment) and gouge-filled joints (bedrock) formed at the same time in different materials.

In this article, I review the evidence for a lateral spread origin for the dikes. The passive infilling of tension cracks is inconsistent with the shape of the dikes, characteristics of their fills, or their regional distribution.

What is Lateral Spreading?

The U.S. Geological Survey's Fact Sheet 2004-3072 (USGS, 2004) tells us that lateral spreading is a specific type of low-angle landsliding that involves extension, translation, and liquefaction. Lateral spreads,

...occur on very gentle slopes or flat terrain...the dominant mode of movement is lateral extension accompanied by...tensile fractures. The failure is caused by liquefaction...usually triggered by rapid ground motion, such as that experienced during an earthquake...Lateral spreading in fine-grained materials on shallow slopes is usually progressive. The failure starts suddenly in a small area and spreads rapidly.

According to Varnes (1978), lateral spreading involves,

Movements may involve fracturing and extension of coherent material...owing to liquefaction or plastic flow of subjacent material. The coherent upper units may subside, translate, rotate, or disintegrate, or they may liquefy and flow.

Types of landslides. Ten different types of slides, flows, topples, avalanches, and creep according to USGS (2004). Tensile cracks appear in nine of the ten diagrams. The fractures are strongly parallel and oriented perpendicular to the direction of slip. The figure depicts tensile fractures in nine of ten landslide types. Lateral spreading is not unique to one type.

Unique or Gradational?

The ten styles of mass wasting are not themselves end members, but share certain aspects in common; one type commonly grades into another. For example, "block slides", "rock falls", and "topples" largely describe the same thing, except the size of the blocks varies and maybe the dip of the beds. Likewise, "debris flows" and "debris avalanches" vary in the amount of sediment, water, and organic material they contain. Debris flows and debris avalanches often occur together. Whether you call a slope failure a "lateral spread" or a "creep" can come down the rate of its movement downslope. Slow movement is typically called creep. Fast, a spread. Many landslides involve more than one process, thus are "complex" according to the Fact Sheet.

Lateral Spreads: The geological stem cell

Lateral spreads are the geological equivalent of a stem cell in biology. They grow with time and assume one form or another if permitted to develop. Hear me out.

In most landslides, tension gashes (ground fissures) appear first, often at the location of the soon-to-be head scarp. Down-slope translation (extension) causes the fractures to form in sediment, rock, or soil. The fractures warn us of things to come. With time and continued movement, they will grow and merge at depth to form a basal sliding surface. Tensile cracks (not unique to any slide type) will evolve into a scarp from which the slide mass will separate and become something we name - a topple, a slide, a flow, a creep. Geologists wait until mass wasting develops for awhile before we give the thing a name because the name identifies the dominant process responsible. Luckily, landslides generally proceed quickly, so we rarely have to wait long.

Before and after. Lateral spreading extends the upper unit (stratified sand and silt). Tensile fractures separate coherent, stratified blocks. How well this model fits dike-bearing strata in Eastern Washington (i.e., Touchet Beds) is unclear. Some basic questions must be asked: Did liquefaction occur at a low level in the Touchet Beds? Can we see the base of the translated blocks? Are sheeted fills in the dikes consistent with liquefaction below and brittle cracking above as shown in the diagram? What is the sedimentary record of strong shaking in the region over the past 2 million years? Did lateral spreading occur in the same way everywhere the dikes are found - that is, does it explain identical dikes that occur throughout >30,000 km2 region? Where else on Earth do we find analogous structures: vertically-sheeted, wedge-shaped sand-silt dikes formed by earthquake-triggered lateral spreading?

What Are Proponents of Lateral Spreading Really Saying?

What are we actually saying when we say, "these clastic dikes formed by lateral spreading" (i.e., Pogue, 1998)? First, that liquefaction was involved. Second, that sliding occurred on a liquefied basal plane (a thin body of fluidized sediment). Third, that all Touchet-type clastic dikes are products of a single mechanism.

A few field-testable questions (and answers) come to mind:

Q: Is there evidence of intense or widespread liquefaction or surface-vented sand in the Touchet Beds of Eastern Washington, northeastern Oregon, and west-central Idaho?

A: No.

Q: Is a basal sliding surface present in Touchet Beds that contain clastic dikes?

A: No.

Q: Does lateral spreading, which by Varnes's definition requires liquefaction and a slide plane, explain the presence of identical sheeted dikes in both the unconsolidated sediments and in bedrock?

A: Not well.

Q: Lateral spreading and diking were triggered across a vast region, but mass wasting was not?

A: No.

Three alternatives. All three scenarios assume a near-level ground surface, which comports with known locations of the largest clastic dikes (basin centers). A free face cliff several meters high is needed in order for spreading to occur. Valley bottoms have little slope and without something like an incised channel, slide blocks would have nowhere to go. Scenario A.) Channel incision creates a free face and removes lateral support allowing blocks to separate and slide laterally. The open fractures between blocks fill with collapsed sediment to be come dikes. The trigger is rapid incision. Scenario B.) Earthquake-triggered liquefaction causes a wet sandy layer at depth to consolidate and expel jets of sediment and pore water toward the surface. Sand slurry is vented at the surface as sand blows. Dikes are sourced from an previously-deposited bed (in situ, undeformed) remobilized by shaking. Scenario C.) A large vertical load is rapidly imposed on the ground surface triggering hydraulic fracturing in the substrate. The thin fractures immediately fill with sediment supplied from the base of the flood. Fractures are propped by the sandy fill, becoming dikes. Pressure-volume cycling (repeated crack and fill) produces vertical sheeting. Deformation occurs along discrete fractures; the surrounding host material remains undeformed. Repeated flood-load cycles create composite clastic dikes as new dikes sourced in younger flood sediments follow weakness pathways established by older dikes.

Typical Touchet-type dike at Burlingame. Do large exposures like Burlingame Canyon show evidence of lateral spreading or sand blow activity? Missoula flood rhythmites R1 through R7 deposited by separate floods are intruded by a typical Touchet-type clastic dike. The dike originates at the base of R7 (within the rhythmite stack) and descends through several underlying beds. It does not crosscut the entire the section, rather it formed during one flood (flood that deposited R7) in the middle of the Missoula flood period (18-14 ka). Other dikes descend from older and younger levels in the stack. The dike cuts a clean path through the host sediment and does not follow a rubble zone between laterally-translated blocks. Bedding contacts are not offset, tilted, or laterally spread across the dike. No low angle sliding surfaces are present in the outcrop. Both branches taper downward to a point. The sediment that fills the dike was not supplied by a liquefied sandy layer at depth (a bed below R1) as would be the case if it fed a sand blow (i.e., Fuller, 1912; Obermeier, 1998), rather the source of the dike is clear. It begins at the base of rhythmite R7. Widening at the top of the dike is a small sag, not the truncated edifice of a sand blow. Bedding at the top of the dike grades smoothly upward into R7. Flute casts on the interior faces of the dike's walls provide clear evidence of downward infilling. This example exhibits the characteristics typical of sheeted Touchet-type dikes found throughout the region.

Not your fault. This dike is unfaulted. There is no bedding-parallel offset here. This dike-sill-dike geometry is its original, undeformed shape. Alternating dike and sill segments results from fluid-driven fracturing (hydraulic fracture) in materials with different properties (grainsize, porosity, permeability). The dike exploits the most efficient pathway available, diking across less-permeable silty layers (tan), paralleling sandy, ihgh-permeability ones (gray). Routing changes occur at the centimeter scale with variations in bedding. Is this dike form explained by lateral spreading? Note sheeting in the dike segments is vertical while sheeting in sill segments is horizontal. Can you visualize dike and sill segments in 3D (think: disc-shaped fracture fronts)? We're not seeing its complete geometry in the 2D cutface. Hellsgate Recreation Area at Lewiston, ID. More info on this exposure here:

Landslides Are Local

By their very nature, landslides comprise a relatively small portion of Earth's surface. Geologists get excited about really big landslides because they are really big. If big slides were the norm, nobody would bat an eye at the way Anchorage slipped into the sea in 1964. No one would care about the Heart Mountain Detachment.

If lateral spreading is the origin of the clastic dikes in Eastern Washington, then the dike network constitutes one of largest landslide complexes on Earth. The trigger for such a slide would require an enormous amount of energy - far greater than even a full rip Cascadia quake could generate. The dikes show clear evidence of repeated injection, which means diking was triggered many times - but only during the Pleistocene and only within Ice Age floodways. No Touchet-type dikes originate in Tertiary or Holocene deposits, though the Cascadia subduction zone and various faults inboard of the margin have been active for millions of years. The sheeted dikes are not found beyond the margins of the routes followed by Ice Age megafloods (Cooley, 2020).

Mass Wasting Disrupts Bedding

Slides and flows destroy depositional layering. Liquefaction is effective at homogenizing bedded sandy materials. In fact, sandy layers that have been liquefied often appear structureless; some geologists call them "homogenites". If the dikes are products of lateral spreading, delicate bedding and fine scale laminae in dike fills and in the unconsolidated sediments that host them would be disrupted. Rubbly zones of between translated blocks would be numerous, especially where the dikes are numerous. Disrupted zones would form a halo around at least some dikes, as tensile fracturing must precede diking.

Flat-lying and mostly undisturbed. Touchet Beds at Burlingame Canyon near Lowden in the Walla Walla Valley are intruded by several large, slender dikes. The dikes cut cleanly through bedding, descending and pinching as they go. Deformation at Burlingame, other than the dikes, is meager. There are very few faults and no evidence of liquefaction. No sand blow edifices have ever been reported here or elsewhere in the Touchet Beds. If the dikes are taken to be lateral spread structures, the total amount of lateral extension is minor, equal to the cumulative width of the dike fills. Since the dikes are not oriented parallel to one another and the spread direction is unclear (no free face apparent), the total amount of spreading amount would actually be some fraction of cumulative width (not true width). It would seem strange to interpret the Burlingame exposure as "a set of coherent, shifted, stratified blocks separated by vertical clastic dikes occupying tensile fractures". Where is the liquefied sliding surface required by lateral spreading as defined by USGS and Varnes? Photo taken in 1978 by an unnamed photographer, Washington Geological Survey archive.

Lateral Spreads Produce Tilted Blocks

Landslides often produce jumbled, chaotic deposits. Deposits produced by lateral spreads are decidedly less jumbled, consisting of slightly-tilted, but coherent blocks. After more than a century of investigation, no such pattern has been found in any outcrop (Jenkins, 1925; Black, 1979; Cooley, 2020).

Landslides Are Messy

If lateral spreading occurred at a regional scale, diking would be just one effect, one type of evidence left behind. A whole suite of other mass wasting structures and deposits would accompany the dikes. However, the Touchet Beds are neither jumbled nor chaotic. Some warping (open folds) and some faulting (meter-scale listric normal faults) are observed in certain locations (Cooley et al., 1996), but overall the Touchet Beds remain largely undeformed.

Pristine preservation. Fine laminae and delicate primary sedimentary structures in the fills of clastic dikes are nearly always perfectly preserved. Back-filled rodent burrows (originally round), common in Touchet Beds, have not all been flattened by compaction. Adjacent dikes are not sympathetically folded nor are their margins uniformly crenulated. Gentle, open folds and some saggin in slackwater rhythmite sections is common, but wholesale homogenization and loss of original bedding has never been observed by the author or reported by others. My photos from Pine Creek, OR.

Where's the Free Face?

The dikes form polygonal networks near the center of the Pasco Basin, visible in aerial photos. Numerous reports discuss this same location on the Hanford plain off Army Loop Road (Newcomb, 1962; Grolier and Bingham, 1978; Lillie et al., 1978; Black, 1979; Silver and Pogue, 1997; Fecht et al., 1999; Johnson et al., 1999; Murray et al., 2003; Gee and Ward, 2006). The site is today a low-relief valley bottom and was the same during the Pleistocene, when the dikes formed. The Hanford plain accumulated coarse, sandy sediment delivered by all Ice Age floods that drained through Wallula Gap. Incised channels are absent in the remnant Pleistocene topography. Borehole studies make no mention of buried valley fills or incised channels beneath the Hanford plain. If lateral spreads translated blocks toward a free face, where was the free face in the center of the Pasco Basin?

Dike polygons. A polygonal network of clastic dikes appears in aerial photos of the Army Loop Road onthe Hanford Site, Pasco Basin, WA. Vegetation was burned away by wildfire several years ago, revealing the dikes. Google Earth photo.

What Does the Literature Say?

Geology textbooks identify lateral spreads as on of several common responses of wet sediment to seismic shaking. Lateral spreads are not, however, associated with clastic dikes, much less sheeted dikes. Remarkably few examples of sheeted clastic dikes are found in journal articles written over the past 150 years. I've reviewed >300 articles published since 1830s that discuss clastic dikes. To my knowledge, not a single article that describes sheeted, wedge-shaped dikes formed by earthquake-triggered lateral spreading. Sheeted dikes do occur where the ground surface was overridden by glaciers, lahars, and debris flows.

Where Is Clear Evidence of Lateral Spreading in Eastern Washington?

Lateral spreading coincident with diking is clear along the banks of the Upper Columbia River (Lake Roosevelt) north of the Spokane River. The varved beds exposed in high shoreline bluffs near Hunters and Inchelium were deposited on the bottom of Glacial Lake Columbia during the late Pleistocene. Large, wedge-shaped, single-fill dikes (unsheeted) descend for 10-20m through the section, terminating just above a highly-strained zone in the same material. Textures in the basal shear zone resemble those in high grade metamorphic rocks like mylonite. The dikes, up to 60cm wide at their tops, appear to meet the USGS/Varnes criteria for lateral spreading.

How it works up north. Lateral spreading at Hunters and Inchelium went something like this. To learn more about the dike at Hunters, see my other post:

Dikes at Hunters. Dikes descend many meters through shoreline bluffs along Lake Roosevelt.

Single-fill structures. Dikes at Hunters, WA are not vertically-sheeted like those in the Touchet Beds. These formed near a free face that laterally spread and partially toppled into an open channel. They contain sandy-gravelly material supplied by glacial outwash that overlies the lake beds and armors the terrace surface above. Cobbles up to grapefruit size are contained in dike fills.

Basal shear zone. Mylonite-like shear fabric occurs in a bedding-parallel zone near the base of the exposure, below where dikes pinch out. The deformed material is unconsolidated silt and clay.

Shear zone in context. Flat-lying varve sets occur above the shear zone and beneath the swirled zone below it. The dark-colored lake-bottom clays represent background sedimentaton (varves) that alternates with light-colored silts. Silts are the coarsest grain size here and likely represent the plume of suspended sediment and some portion of the associated bottom-hugging density flow. Flood currents moved up the valley out of the main body of Glacial Lake Columbia. Hunters is the distal, northern limit of Missoula backflooding in the Columbia Valley (Hanson and Clague, 2016, Fig. 1). The Hunters area would be a great place for an MS student with a good kayak interested in subaqueous deformation structures to base out of. The place is peaceful and beautiful. Good campsites abound. The local tavern attracts a cast of colorful characters. Discoveries would be made.

Peaceful Hunters. O.P. Jenkins' photo of ice-sculpted terrain at Hunters, WA c. 1921. WGS photo archive No. 00525.


Lateral spreading is a type of landsliding commonly observed in tide flats, recent alluvial fills, or deltaic deposits subjected to shaking by strong earthquakes. The desire to attribute sheeted clastic dikes in Missoula Flood rhythmites to lateral spreading is understandable. Extension certainly did occur with diking, but spreading wasn't the dominant mechanism. Field evidence provides little support for the hypothesis; I've changed my mind since my undergraduate thesis work in the 1990s (Cooley, 1996; Coole et al., 1996). The dikes number in the hundreds of thousands and 99.9% of them have identical characteristics. They were not produced by repeated mass wasting of any sort. Different types of slides, slumps, and flows would have occurred with spreads, each leaving behind its own unique evidence. Fine laminae and delicate primary structures are pristinely preserved in both dikes and rhythmites today, more than 10,000 years later. Diking occurred along discrete fractures; dikes did not passively fill voids or fissures, nor infill rubble zones between slide blocks. An explanation consistent with many aspects of dike morphology, distribution, and age is detailed in Cooley (2020) and in 'Sheeted Clastic Dikes in the Megaflood Region', a post on this website. Flood-load triggered, fluid-driven fractures were rapidly opened forces generated by glacial outburst floods and rapidly filled by sediments sourced in circulating sediment-laden currents of the same floods. Flooding, loading, fracturing, and diking were coincident. Touchet-type dikes are flood injectites (naturally-propped hydraulic fractures), not liquefaction structures (sand blows) or lateral spread fractures.


Work on clastic dikes in Eastern Washington has suffered from too much speculation and too little data. College faculty are the primary offenders. Reports on the dikes published over the past century are nearly devoid of data. Sweeping conclusions sans data has never been acceptable in Geology, but somehow is acceptable for clastic dikes.

I've bucked the trend by actually doing work. I've collected data on thousands of dikes at >500 localities (Cooley, 2020) in WA, OR, and ID. I've ventured into Montana, British Columbia, and Alaska as well. While certainly inadequate, mine is a substantial dataset and one representative of the entire region thoughout which the dikes are found (>30,000 km2). It is by far the best information available.

Those who favor a lateral spread origin for the dikes either need to redefine the term, create a new term for their purposes, or explain the absence of a liquefied sliding surface at classic localities such as Touchet, Lewiston, Starbuck, Cecil, Warden, and Latah Creek. Data supporting a lateral spread origin has not been presented to date, despite a considerable amount of professorly roadside arm waving. Observing the same outcrops year after year and professing to "just know" how the dikes formed sounds a lot like belief, not science. Simply pointing to the Horse Heaven Hills and saying, "There's a big fault right there!" doesn't cut it any longer. Claiming the dikes are "multigenetic" structures indicates you know nothing about which you speak.

Yes, we all continue to learn. But some of us have done real work and have actually solved a few things. Acknowledge it. Be honest about the extent of your own work. Embrace relevant phenomena not addressed by your own education (i.e., fluid-driven fracture). Keep current. Be a lifelong learner. Wasn't that what they taught us there amongst the the ivy, the brick, and the flowering dogwoods?

Your comments are always welcome:


Cooley, S.W., 2020, Sheeted clastic dikes in the megaflood region of Washington, Oregon, Idaho, and Montana, Northwest Geology, v. 49, p. 1-17,

Cooley, S.W., 1996, Timing and emplacement of clastic dikes in the Touchet Beds of south-central Washington, BA thesis, Whitman College, 37 pgs.

Cooley, S.W.; Pidduck, B.K.; Pogue, K.R., 1996, Mechanism and timing of emplacement of clastic dikes in the Touchet Beds of the Walla Walla Valley, GSA Cordilleran Section Abstracts with Programs, v. 28, p. 57

Hanson, M.A., Clague, J.J., 2016, Record of glacial Lake Missoula floods in glacial Lake Columbia, Washington, Quaternary Science Reviews, v. 133, p. 62-76

Obermeier, S.F., 1998, Liquefaction evidence for strong earthquakes of Holocene and latest Pleistocene ages in the states of Indiana and Illinois, USA, Engineering Geology, v. 50, p. 227-254

Pogue, K., 1998, Earthquake-generated(?) structures in Missoula Flood slackwater sediments (Touchet Beds) of southeastern Washington [abstract], Geological Society of America Annual Meeting Abstracts with Programs, Session 174, Abstract T43

U.S. Geological Survey, 2004, Fact Sheet 2004-3072, Types of landslides,


Last 50 Posts
All Posts by Month
    bottom of page