Landslides in Seattle are caused by a combination of steep slopes (topography), glacial and post-glacial soils (stratigraphy), and a pronounced wet winter season (typically November through March).  It requires the interaction of all three to create landsliding in the city.  With the exception of coastal California, Seattle suffers more damage from landslides than most other large cities in the United States.  

Seattle is comprised of a series of linear ridges and broad plateaus with intervening river valleys and linear depressions that were shaped by the last glacial ice to reach this area.  To the south of the Lake Washington Ship Canal, ridges and swales dominate the landscape.  The major hills that crest at about 450 to 500 feet are Magnolia, Queen Anne Hill, Capitol Hill, Beacon Hill, West Seattle, and Mount Baker Ridge.  They are separated by Interbay, Lake Union, the Duwamish River Valley, Rainier Valley, and Elliott Bay.  Not all of the swales are water-filled.  Some are naturally filled with glacial and nonglacial sediments and others are modified with artificial fill.  With the exception of Longfellow Creek in West Seattle, the ground surface is drained by short and steep streams. 

In the area north of the Lake Washington Ship Canal, the ground surface is a broad undulating plain.  The ground rises up to the north gradually from the ship canal, nearly reaching elevation 500 feet near the north city boundary.  It is broken by depressions, such as Green Lake, Haller Lake, and Bitter Lake.  It has also been incised by Pipers Creek on the west and Thornton Creek on the east. 

As shown on a topographic relief map of Seattle, Figure A-1 (Appendix A, Volume 2), the ridges and plateaus are surrounded on all sides by steep slopes.  These slopes range in inclination from about 25 to 90 degrees with the horizontal.  In general, the steeper slopes are those that border the shoreline of Puget Sound, particularly the rare ones that are not protected from wave erosion.  The only remaining unprotected bluffs in the city are in Discovery Park and a short section of shoreline at the south end of Magnolia.  Elsewhere, the shoreline is armored or otherwise protected by individual short bulkheads or by long bulkhead/embankments, such as the Burlington Northern Santa Fe Railroad (BNSF), north of the ship canal. 

It is on these steep slopes that surround the ridges and plateaus of Seattle that Seattle's landslides recur on a regular basis.  This process is particularly evident in the retreat of bluffs since the disappearance of the last glacial ice from the Seattle area about 13,500 years ago.  It has been estimated, based on marine charts showing change in submarine topography, that the Puget Sound bluffs in Seattle have retreated at a rate of about 75 feet per century (Galster and Laprade, 1991).  The rate was undoubtedly much greater in the first few millennia following glacial retreat; however, it is equally obvious that slope instability is still very active.

Seattle is underlain by bedrock of Tertiary age, glacial and interglacial soil deposits of the Pleistocene Epoch (2 million to 10,000 years ago), and nonglacial soil deposits of the Holocene Epoch (present-day).  However, soils deposited during the most recent glaciation of the central Puget Lowland dominate the surface and subsurface geologic conditions in Seattle.  These rock and soil deposits are very completely interwoven by repeated sequences of deposition and erosion.  It is clear that each of the major ridges or uplands has a unique stratigraphic system. 

Bedrock, consisting primarily of sandstone and siltstone, outcrops sporadically to the south of the Seattle Fault (see Figure A-3); however, no bedrock outcrops to the north of this fault within the city limits because it is buried by 1,000 to 3,000 feet of glacial and nonglacial sediments.  The bedrock does not play a significant role in the landslide history of Seattle.  The only major area of bedrock instability occurred east of Boeing Field where large excavations were made for Interstate 5 (I-5) in the 1960s. 

2.2.2  Pre-Vashon Deposits

Older nonglacial and glacial soils (pre-Vashon Stade) are present within the downtown business district and in the cores and flanks of most of the hillsides.  However, these older soils have not produced much landsliding.  Pre-Vashon glaciomarine deposits (Possession Drift) underlie the downtown business district, Beacon Hill, and Mount Baker Ridge.  They are intermixed and chaotically stratified clayey till, glaciolacustrine silt/clay, and sand.  Locally throughout the city, these deposits are overlain by a variety of sediments of the Olympia interglacial period.  These sediments include sand, silt and clay layers with scattered organic fragments, peat pockets, and thin interbeds of gravel. 

2.2.3  Vashon Glacial Deposits

The primary geologic units that are involved with landsliding in Seattle are those that were laid down during the Vashon Stade of Fraser Glaciation, between about 17,000 and 13,500 years ago (Waldron, 1962; Booth, 1987).  Together, the four members (Lawton Clay, Esperance Sand, Vashon Till, and Vashon recessional outwash) comprise most of the ridges and uplands in the Seattle area (Figure 1-1).  The lowest three members were overridden by approximately 3,000 feet of Vashon Stade ice.  Recessional outwash was not overridden by this ice.

The Lawton Clay, a glaciolacustrine deposit, was laid down in a lake that formed as the glacial ice advanced southward from British Columbia and blocked the Strait of Juan de Fuca.  The unit consists of laminated and massive silty clay and clayey silt with scattered fine sand lenses.  It is hard, from having been glacially overridden.  Because of its hard condition and fine-grained consistency, it is relatively impervious such that groundwater tends to perch on top of its upper surface.  The Lawton Clay is typically interbedded with sands of the overlying Esperance Sand near the contact of the two units.

The Esperance Sand is a glaciofluvial advance outwash that was deposited by streams issuing from the Vashon glacier as it advanced southward.  It is comprised chiefly of fine to medium sand that is locally gravelly.  Locally, it also contains silt layers and pockets and discontinuous layers of gravel.  It is very dense and pervious with groundwater normally flowing freely through this soil.

Vashon Till (lodgment till) was deposited beneath the Vashon Stade ice.  Also known locally as "hardpan," it is normally a gravelly, silty sand or a gravelly, sandy silt with scattered cobbles and boulders.  It is very dense (one of the most compact soils in the world) and relatively impervious.  Water infiltrating through overlying deposits normally perches on top of the till; however, locally the till contains pervious water-bearing zones.

Vashon recessional outwash was deposited by streams issuing from the Vashon glacier as it receded or wasted.  It is a slightly silty to silty sand with scattered gravel, and is the deposit that is found at the ground surface on most of the uplands.  It is relatively pervious and loose to medium dense, not having been glacially overridden.  Precipitation commonly infiltrates readily through this deposit and then perches on top of the Vashon Till.

2.2.4  Holocene Deposits

Holocene (post-glacial) deposits are ubiquitous throughout Seattle.  They include alluvium, beach deposits, depression fillings, and colluvium.  These soils have not been glacially overridden.  As such, they are typically loose to medium dense or soft to stiff.  Alluvium is deposited along the major rivers and creeks, such as the Duwamish River and Longfellow and Pipers and Thornton Creeks.  It is comprised of loose sand and gravel that is normally wet; however, because of its low slope gradient, it is not normally related to landsliding.

The beach deposits found along the shorelines are normally not landslide-prone because of a lack of relief.  Landslides do deliver material to the shoreline that contribute to forming beach deposits.

Depression fillings commonly consist of soft clay and silt and organic materials, such as peat.  They accumulate in low spots on the ground surface.  They are normally found on the upland ridges and plateaus, although they can be included in the same areas as river alluvium.  Depression fillings are not normally associated with landsliding.

Colluvium is very commonly associated with landsliding.  Colluvium is the loose to medium dense or soft to stiff soil that mantles the sides and toes of slopes throughout the city.  Because it was deposited by gravity processes such as soil creep, surficial sloughing, landsliding, and slope wash, grain-size can vary from clay and silt to boulder-size.  The mode of accumulation ranges from slow creep (the imperceptible movement of only inches per year or less) to catastrophic landslides.  Soil creep in the upper few feet of soil on a slope is commonly reflected in the bowing of trees on the slope.  Colluvium is normally moist to wet, especially during the rainy season. 

Another category of Holocene soil is fill placed by humans.  Fill soils vary widely in grain-size, location, presence of debris, and size.  Although many new fills have been compacted and engineered, most older fills were just dumped in place or nominally run over with the spreading equipment.  These fills can be particularly unstable where they have been placed on or in close proximity to a steep slope. 

In addition to topography and stratigraphy, groundwater is the other factor that plays a significant role in the generation of landslides in Seattle.  In spite of loose soils on steep hillsides, landslides very rarely occur in the dry summer months, although this sometimes happens under unusual conditions.  It is the water pressures that build up in the ground, usually during the pronounced wet-weather season, that nearly always trigger the slide event.  The source of water for any individual landslide can be natural or influenced in some way by human activity or a combination of these two factors.  This section presents the various ways in which the groundwater interacts with the geologic units in Seattle.

The groundwater found closest to the ground surface is that perched atop the Vashon Till (Figure 1-1).  In this case, precipitation or water related to human activities, such as improper drainage, infiltrates down through recessional outwash until it encounters the top of the Vashon Till.  The perched water may flow until it emerges in a pond, creek, or a steep slope, where it forms a spring.  Water at this contact normally dries up in the spring or summer and does not reestablish itself until the winter months.  Because the contact between the recessional outwash and the Vashon Till is shallow (normally less than 10 feet deep), it is normally reachable with a backhoe for dewatering.

On the sides of hills, the undisturbed glacial soils are covered with a rind of colluvium.  Water is commonly able to penetrate the semi-pervious colluvium because it is relatively loose and contains some fraction of sand; however, it cannot infiltrate easily into the very dense or hard underlying glacial soils.  The water, therefore, travels along the inclined contact between the two materials of different permeability.  Water at this contact also normally dries up in the spring or summer and does not reestablish itself until the winter months.  Because the contact between the colluvium and the underlying undisturbed soil is shallow (normally less than 10 feet deep), it is generally reachable with a backhoe for dewatering.

The most prevalent groundwater aquifer in the Seattle area is the Esperance Sand.  Precipitation infiltrates through "windows" or cracks in the Vashon Till and continues vertically down into the Esperance Sand until it encounters the top of the underlying Lawton Clay.  Owing to the low permeability of the Lawton Clay, the groundwater perches on the clay and then moves laterally, eventually saturating near-surface colluvium and/or emerging in a spring on a hillside.  Because of the residual lag travel time of this water, many of these springs are perennial.  They are the most prolific springs throughout the city.  It is fairly easy to trace the level of the sand/clay contact by locating the springs on a hillside.  The source of water for an individual spring or group of springs is very difficult to define, as it probably has a large regional contributing area uphill from the spring.

The key stratigraphic marker for landslide location is the contact between the Esperance Sand and the Lawton Clay (Figure A-4).  It is commonly termed "The Contact."   In Landslides in Seattle, Tubbs concluded that "the landslide typically occurred along the trace of the contact between the Esperance Sand and either the Lawton Clay or pre-Lawton sediments."   No experiences or collected data in the past 24 years have changed that conclusion. 

Three main sources were used to develop the historical database for the assessment of landslide hazards in the City of Seattle (City).  The primary source of landslide information was Seattle Transportation Department (SEATRAN), which has landslide files dating to 1890.  However, good records were not kept before the 1960s, when the landslide file was started by Mr. Finney of the Seattle Engineering Department.  These files consist of several types of information including memorandums by field inspectors, court documents, photos, maps, subsurface data, geotechnical reports, mitigation plans, and cost estimates.  The accuracy of the City landslide files is dependent on several factors.  These factors include available staff levels, the amount of damaged or missing information, and, most importantly, the degree of landslide reporting by the public.  In general, SEATRAN's files record landslides that primarily affected rights-of-way or utilities; not private properties. 

The second source of information was the Department of Design Construction and Land Use (DCLU), which has maintained a landslide database since 1986.  This database includes the reported locations of particular landslides with a brief description regarding the structural integrity of the affected structures.  The DCLU landslide database contains a relatively complete record of landslides that occurred during severe storm periods.  However, the database is relatively incomplete for other periods.  Furthermore, the landslide dates included in the DCLU database reflect the time of inspection rather than the initiation of ground movement.  Therefore, when the failure date could not be determined from the DCLU files, we assumed an approximate failure date that was close to a previous storm event. 

The third source of information for the landslide database was the Shannon & Wilson, Inc. (Shannon & Wilson) files.  Shannon & Wilson has maintained project files since 1954.  The inventory includes projects that pertain to ground displacement performed by the company.

The input data needed for the Seattle Landslide Study is subdivided into six main groups:

The following is a detailed description of the data comprising each of the attributes in the landslide database, which is presented in Appendix E.  The Appendix also contains a legend that defines abbreviations and provides additional explanations for each data field.

Record Number

The Record Number field represents the unique identifying number for each documented landslide in the database.  Each landslide was assigned a sequential record number when it was entered into the database.  After data processing (refer to Section 3.3), several landslides were omitted (duplications, etc.) and, therefore, the record numbers are not in a continuous series.

Location

The Location field consists of an address representing one or more of the following:  the address of the person or persons reporting the incident, the address of the closest property to the event, the address of a property affected by the event, or an approximate address specifically used for plotting on the Geographic Information System (GIS).  Locations of landslides reported by BNSF along the railroad right-of-way are referenced by Milepost number (e.g., MP 8.5).

Date

The Date field contains the approximate date of initiation of ground displacement.  In cases where the exact date was not known, we estimated precipitation year.  This was accomplished by assigning the first day of January as the date.  A higher percentage of older landslides were assigned this date because of the paucity of information in the older files.  Note that many landslides occurred during the New Year's storm of 1997.  The January 1, 1997, dates for these landslides are accurate.  A note in the Comments field or the Date Confidence field includes the 1997 landslides where the exact date is not known.  A precipitation year is defined for this study as beginning on July 1st and ending on June 30th of the following year. 

File Number

The File Number field contains the source's file number where the information was obtained.  DCLU file numbers consist of  "J#", "J##", and "J###" (where "#" represents a number, e.g., J21) for events from 1986 to 1996 and "96-97 storm" for landslides occurring during the 1996-1997 winter.  Other file numbers are as follows:

Shannon & Wilson project files consist of "S&W."  Landslide events discovered during field reconnaissance for this landslide study consist of "Field." Landslides reported by sources other than the City or Shannon & Wilson consist of "Citizen."  All other file numbers represent SEATRAN file locations.

Consultant Report

The Consultant Report field shows a letter code if an engineering consulting company prepared a report regarding the landslide. 

Field Checked

Field Checked shows the confidence in whether or not the landslide is properly located on the map.  Landslides that are accurately located on the map and field checked are "True."  Landslide locations shown on the map that were not found during field reconnaissance are marked "False."    

Date Confidence

The Date Confidence field is "True" where the date is believed to be accurate and "False" when the date may be approximate.  Where no determination could be made, the Date Confidence field was left blank.

3.2.2   Landslide Characteristics

Slope Height

The Slope Height field is an estimate of the approximate elevation difference, in feet, between the headscarp and the toe of the slide, as estimated from historical records and field verification.  Differences between these estimates and actual conditions may exist. 

Landslide Type

The Landslide Type field consists of a general classification of each landslide, even though more than one classification may have been involved at a specific location.  The classification of landslide type recorded in the database was the predominant type based on our interpretation of the records and our site visits.  There are four general landslide-type possibilities:  high bluff peeloff (HBP), shallow colluvial (SC), deep-seated (DS), and groundwater blowout (BO).  Please refer to Section 4.0 for detailed descriptions of landslide types. 

Debris Flow

The Debris Flow field is "Y" if a debris flow with runout generally longer than 50 feet occurred and "N" if a debris flow did not occur or had a short runout.  Where no determination could be made, the field was left blank. 

Size

The Size field represents the approximate aerial extent of the ground displacement.  Landslides covering an area greater than 10,000 ft² are denoted with an "L" and those equal to or less than 10,000 ft² are denoted with a "S."  Note that some large landslides may cover a small area, but displace a large volume of material because of their depth.  This type of landslide is not represented because of the difficulty in estimating depth to the slide plane.  Differences between these estimates and actual conditions may exist.

Vegetation

The Vegetation field describes the vegetative ground cover contained within the landslide margins based on file pictures and/or field reconnaissance.  There are four vegetation type possibilities:  brush (B), wooded (T), sparse cover or bare ground (S), or grass (G).  If even one tree was contained within the slide margins, the landslide was designated with a "T."  When there were no pictures in the file and no determination could be made, the field was left blank. 

Topography

The Topography field describes the approximate average slope angle.  If the slope angle is greater than 40 percent, it was described as a steep slope (SS).  Moderate slope (MS) was used for slope angles less than or equal to 40 percent. 

3.2.3   Stratigraphy (Geology)

The next four fields of the attribute table indicate the geologic units involved in the ground displacement.  The units are ordered, as they would appear in the headscarp, typically from youngest to oldest (top to bottom).  The five designations used in this study were:  fill (HF), colluvium (HC), glacial till (QT), glacial outwash sand (QS), and lacustrine clay/silt (QC).  The list of geologic units involved in a particular event is estimated based on the type of landslide, the geographic and topographic location, and any subsurface information disclosed in the file.  The geologic units were not field verified and differences between these estimates and actual conditions may exist.

3.2.4   Landslide Trigger Mechanisms

All four of these fields reflect documentation in files or reports regarding the trigger mechanism of ground displacement.  These fields do not reflect the degree of contribution, nor do they necessarily represent a professional evaluation.  Many landslides documented in the SEATRAN files are claims to the City and may contain some degree of bias.

Natural

The Natural field describes the trigger mechanism as being natural (Y) or human (N).  For example, precipitation is considered a natural trigger, whereas pipe breaks and excessive lawn watering are not.  Blank spaces indicate that no determination could be made.

Groundwater and Surface Water

The next two fields, Groundwater and Surface Water, indicate when groundwater or surface water may have been the possible triggering mechanism of the event.  A "Y" indicates that groundwater and/or surface water may have triggered the landslide and an "N" indicates that these two trigger mechanisms probably were not involved.  Improperly directed surface water by private parties and naturally occurring surface water were not differentiated in this field.  Blank spaces indicate that no determination could be made.

Fill and/or Cut

The Fill and/or Cut field indicates when filling and/or cutting may have triggered the landslide event.  Landslides resulting from inadequate shoring of an excavation, for example, were denoted by a "Y." An "N" indicates that filling and/or cutting was not involved in triggering the landslide.  Blank spaces indicate that no determination could be made.

3.2.5   Roads and Public Utility Impact

All four of these fields reflect documentation in files or reports regarding the effect of roads and underground public utilities on ground displacement.  These fields do not reflect the degree of contribution, nor do they necessarily represent a professional evaluation.  Many landslides documented in the SEATRAN files are claims to the City and may contain some degree of bias. 

Road Cut and/or Fill

The Road Cut and/or Fill field is similar in nature to the fill and/or cut field, but only pertains to public roads.  Filling at the top of the slope is denoted by an "F", cutting near the toe of a slope is denoted by a "C", and in cases where both filling and cutting were factors in triggering the event, an "FC" was entered.  Blank spaces indicate that no determination could be made.

Surface Drainage

The Surface Drainage field refers to the effect(s) of City-maintained drainage systems on ground failure.  Landslides that may be affected by City-maintained drainage systems are denoted by a "Y" and landslides that may not be affected are indicated by an "N."  Blank spaces indicate no determination could be made.   

Pipe Leak

The Pipe Leak field refers to the presence of additional water introduced to a landslide as the result of a pipe leak or pipe rupture.  A "Y" indicates the presence of a pipe leak or rupture and an "N" indicates no involvement.  No differentiation was made between a pipe break resulting from ground displacement and ground displacement resulting from a pipe break.  Blank spaces indicate no determination could be made.   

Trench Fill

The Trench Fill field indicates the presence of trenches serving as conduits for groundwater that possibly contributed to instability.  A "Y" indicates possible involvement and an "N" indicates no involvement.  Blank spaces indicate no determination could be made. 

3.2.6   Damage and Repair (Mitigation)

This group of fields pertains to landslide mitigation.  It is important to note that the degree of damage or the type of mitigation does not necessarily refer to the address listed in the location field. 

Damage

The Damage field is a numeric field referring to the degree of damage caused by a particular event.  A value of "3" is equivalent to a "red tag" or severe damage to the property; a "2" is equivalent to a "yellow tag" or moderate damage; a "1" is equivalent to a "green tag" or some temporary damage; and a "0" indicates no observable significant damage to property.  Blank spaces indicate no determination could be made. 

Repair Type

The Repair Type field reflects repairs described in the files, reports, and/or observed in the field during field reconnaissance.  Refer to the Legend in Appendix E for a description of the repair types.  To the extent possible, we included only repairs that were constructed.  However, some landslide records did not include enough information to determine if the repair was actually built.  These cases may be included.  Furthermore, the type of mitigation described in the table does not distinguish between structurally engineered repairs and non-engineered repairs.  

Repair Effective

The Repair Effective field indicates whether the repair was effective in preventing further ground movement at that particular location thus far.  It is not a warranty that the repair will remain stable.  If repaired and no subsequent ground movement was documented, a "Y" was entered.  If the repair subsequently failed regardless of design, an "N" was entered.  If the integrity of the repair is questionable regardless of a failure, an "N?" was entered.  If a repair was not necessary, no repair was accomplished, or if there was no information regarding repairs, the field was left blank.

Upon completion of the data collection phase of the study, duplicate addresses with the same failure date were eliminated.  The data was plotted on a GIS using ArcView 3.0a desktop GIS software.  Several City ArcInfo GIS coverages were loaned to Shannon & Wilson for purposes of map reproduction, discrete address-based geocoding, and analysis of the occurrence of landslides.  These City ArcInfo coverages included, but were not limited to, orthophoto-derived topography, parcels, parks, streets, utilities (which includes water mains and laterals, and drainage and wastewater mains and laterals), buildings, potential slide areas, and steep slopes. 

Each landslide data point was automatically geocoded using the location field and plotted on the center of the parcel.  Note that each landslide is represented as a single point, regardless of size.  The landslide locations were then edited based on the comments in the files regarding location, direction of sliding, type of damage, and/or existing topography.  The point indicates the center of the point of landslide initiation (middle of the headscarp).  Duplicate landslide events were eliminated based on date and their spatial relationship.  For example, a property owner at the top of the slope may have reported a landslide to SEATRAN, while another property owner at the base of the slope may have contacted the DCLU because the same landslide affected their residence.  Shannon & Wilson initially field checked landslides within the three study areas (West Seattle, Magnolia/Queen Anne, and Madrona) for accuracy.  Shannon & Wilson subsequently field checked the balance of the landslides in the database for the analysis in Part 4.  The map symbol was then moved to the proper location or deleted when necessary.  The maps that portray the landslide locations are presented in the Map Folio, Volume 2. 

This section of the study discusses particular aspects of landsliding, including the types of landslides that occur in the city, the timing of landsliding, chronic landslide areas within the city and their characteristics, the causes or contributing factors of landsliding, and the coincidence of landslides with existing potential landslide areas and steep slope areas.  To support this discussion, a series of figures and maps will be presented.  The maps are presented in a separate volume (Volume 2) because of their size.

In evaluating the landslide data compiled for this study, most of the landslides were found to fit into four generalized types.  Those types, together with the figure numbers that illustrate a schematic profile view of each type, are as follows:

There are various combinations of these generalized landslide types, as one type of mechanism may lead to another during the sliding, or the slide may be complex, exhibiting different modes of failure in different portions of the slide.  Landslides involving fill material were classified as shallow colluvial landslides.  The following sections describe each landslide type in greater detail.

High bluff peeloffs (Figure 1-2) occur on the face of near-vertical bluffs where vegetation is absent or sparse.  The soil at and near the bluff face, which has been loosened by the forces of weathering (freezing, thawing, root-wedging), slabs off or slides when it becomes wet during periods of heavy rainfall.  This type of landslide commonly occurs following a period of freezing weather.  Sometimes seepage from more pervious soils, such as recessional outwash, at the top of the bluff, or runoff over the edge of the bluff contributes to this type of instability.  Also, water-bearing layers in the steep bluff could contribute to saturation of the face soils.  Normally, the thickness of soil that slides off the face is only a few feet.  The soil that comes off the bluff may or may not slide for a considerable distance, depending on the water content of the soil and the angle of the slope below the bluff.  Alternative names for this type of landslide are earth fall and blockfall.

4.1.2   Groundwater Blowout

A profile of a groundwater blowout landslide is shown on Figure 1-3.  This type of slide occurs where a pervious soil (sand) overlies a lower permeability soil (clay or silt).  Groundwater collects in the pervious soil and becomes perched on the underlying, relatively impervious soil.  The lower permeability soil could be either a relatively thin silt or clay layer or a thick stratum of silt and clay.  Seepage travels to the slope face immediately above the contact with the under­lying, relatively impervious zone and causes instability where the sand essentially blows out and flows downslope (runout).  Because of this blowout, the upper portion of the slope becomes undermined and also fails.  Groundwater is more important in the development of this slide type than direct infiltration of precipitation and is commonly found at "The Contact," Figure A-4.  Nevertheless, this type of slide normally takes place during or shortly after periods of heavy precipitation because of the added water near the spring exit.  It should be noted that this mechanism for causing landslides (seepage at pervious/impervious contact) was probably involved in a number of slides that were categorized as shallow colluvial landslides in the database table and landslide maps.  This categorization would occur where there was a lack of detailed data on a landslide, particularly in the older records.

4.1.3   Deep-Seated Landslides

In the database table, those landslides that were identified or estimated to involve a depth of movement greater than an estimated 6 to 10 feet were categorized as deep-seated (Figure 1-4).  These landslides may involve higher density, in-place soil as well as colluvial soil.  This type of slide normally consists of the block movement of soils where a mass of soil slides downhill on a failure surface that is often arc-shaped.  Sometimes the surface of rupture parallels the ground surface.  As blocks of soil move downhill, a setdown of the ground surface occurs at the upper edge of the blocks, thus forming a slide scarp.  Such movement is commonly progressive; that is, a lower block of soil moves first, which takes away lateral restraint for higher blocks that, in turn, slide.

The deep-seated landslide is initiated by water coming into the slide mass, which takes place either from rising groundwater levels, direct infiltration of heavy precipitation, surface runoff, saturation by the discharge or leaking of pipes into or onto slope soils, or a combination of these sources of water.  Where the soils subject to movement are relatively pervious, such as sand and/or gravel, the movement normally occurs rather abruptly (within minutes or hours).  Where the soils are silt or clay, movements usually occur gradually, over days, weeks, or even months.

4.1.4   Shallow Colluvial (Skin Slide)

Shallow colluvial landslides occur when loose, heterogeneous soils on a steep slope become saturated and slide (Figure 1-5).  The term "skin slide" is sometimes applied to this slide type because a relatively thin depth of soil is normally involved.  They generally consist of rapid movements of the saturated soils, and commonly act like a thick fluid, flowing or running out over a considerable distance.  In the database, they are noted as "debris flows" when the runout generally exceeded 50 feet.  The saturation of soils that causes shallow colluvial landslides takes place by infiltration of surface runoff, direct infiltration of precipitation, groundwater seepage, discharge from pipes, or a combination of these sources of water.

Figure 1-6 illustrates a relatively shallow slide involving fill material.  This type of landslide was categorized as shallow colluvial landslide in the database table and on landslide maps.  If fill is placed at the top or the side of a slope without compaction and suitable drainage provisions (surface and subsurface), instability is likely inevitable. 

The timing of landslides is dependent on precipitation at three different scales.  These are total annual rainfall, monthly rainfall, and a single storm event.  The longest scale is that of annual rainfall.  Although heavier rainfall can occur in years widely spaced or consecutively, the citywide pattern of landsliding, as shown on Figure 1-7, indicates that about every decade a higher than average amount of mass wasting (landsliding) occurs.  For purposes of this study, the rainfall (landslide) year has been designated from July to June.  In this way, all of the rainy winter season is tied together statistically. 

The most notable landslide winters were 1933/34, 1955/56, 1959/60, 1960/61, 1966/67, 1968/69, 1971/72, 1973/74, 1985/86, 1995/96, and 1996/97.  Of these eleven winters, three produced particularly large numbers of landslides:  1933/34, 1985/86, and 1996/97.  The damage incurred during the winter of 1933/34 was responsible for the formation of the Works Progress Administration (WPA) drainage program in Seattle, administrated through the Seattle Engineering Department.  The number of landslides during that winter may have been comparable to the 1985/86 or 1996/97 winters had there been comparable development in the city.

A map showing the distribution of landslides by decade is presented on Figure A-5 (refer to Figure A-3 for area locations described below).  Figure A-5 indicates some interesting trends in the incidence of landsliding and/or changes in the recording of the landsliding.  Many of the landslides on the west side of the Beacon Hill (east of Interstate 5 [I-5]) are older, probably because many of the larger landslides were stabilized by the construction of the freeway in the 1960s.  The newer slides in this area are mostly smaller shallow colluvial landslides.  This reduction in severity of landsliding illustrates how the construction of a major public works project can increase the stability of a hillside; the reason being that buttressing and drainage were widely incorporated into the project.

Two areas that appear to be dominated by new (post-1960) landslides are the Burke-Gilman Trail and Inverness.  These locations may be prone to increasing numbers of landslides because the area had been only sparsely developed prior to 1950.  However, it is also possible that older slides in this area were not reported.

The areas that have large numbers of landslides dispersed through time are chronic slide areas and they include:  Beach Drive S.W., Alki, Pigeon Point, Madrona, Rainier Avenue S.E., Interlaken, Lakeview Boulevard, North Capitol Hill, Laurelhurst, East Queen Anne, Southwest Queen Anne, Southwest Magnolia, and Northwest Seattle.

An above-average winter rainfall punctuated by a large heavy storm on January 18, 1986, led to a rash of shallow colluvial slides throughout the city (refer to Figure A-6 for severe storm events).  The most disastrous storm was the Holiday Storm of December 29, 1996, through January 2, 1997, during which heavy and prolonged rain melted an accumulation of about 12 inches of snow in a two-day period.  The water equivalent of the snow and direct precipitation caused a total of 8.35 inches of water (as measured at SeaTac Airport) to run off and infiltrate the ground from December 29 to January 2.

Two other time periods were important, but did not produce landsliding on the level of 1933/34, 1986/87, and 1996/97.  First, the landslides that occurred in the winter of 1971/72 were the basis for the statistical and geologic conclusions drawn by Tubbs in Landslides in Seattle.  This publication was one of the major factors used as the basis for establishing the boundaries of the landslide-prone areas in the city.  Then during the winter of 1995/96, the Northwest experienced a record winter-long rainfall; a four-month period from November through February.  This exceptionally wet winter was a major contributing factor of numerous deep-seated landslides throughout the region, including Seattle.

A recent study (U.S. Army Corps of Engineers, 1997) of the relationship between landslide frequency and precipitation indicates that the most extensive landslide activity is related most closely to a 3-day storm event (lower bound of about 3.8 inches of precipitation) with an 8-year return interval.  This statistic approximates a rule of thumb that has been used in the Seattle geotechnical professional community for many years.  That rule of thumb says that landslides are likely to initiate whenever there is more than 2 inches of rain in one day or 3 inches in 2 days.

The time of year in which landslides occur is very closely related to the precipitation regime in Seattle, as shown on Figure 1-8 (sheet1  sheet 2).  The overwhelming number of slides occurs in January (45 percent); however, the landslide season typically encompasses a four-month interval: December through March (86 percent).  Although November normally has more precipitation than March, it is likely that a certain threshold of antecedent groundwater is necessary to trigger landslides.  In summary, although landslides are most likely to happen in January and February, it is not uncommon for landsliding to occur in December and March.  However, for planning purposes, the landslide season could begin as early as November and end as late as April.  Although slides can occur in the other months, the probability is low.  Only 7 percent of the landslides in the database occurred during May through October, outside of the normal 6‑month landslide season.  These landslides are often not related to the normal factors that contribute to landsliding (precipitation, steep slope, high groundwater table).  Examples include such things as overwatering, pipe breaks, and excavation slope failures. 

The following sections discuss the distribution of landslides that have occurred in Seattle.  These discussions and associated maps show the historical aspects of landsliding by decade, landslides that occurred without human influence, and the different types.  The 1,326 landslides contained in the City database are presented on Figure A-5, where they are shown by decade.  Figure A-7 shows the same events by type of landslide.  The citywide map (Figure A-3) shows 22 specific areas in the city that have experienced landslides.  They are as follows:

There are many other scattered areas of landsliding and singular landslides in the city; however, the areas listed above are those where densities and frequencies are the greatest.  Five areas in particular appear to have the highest density of landslides:  Southwest Magnolia, Southwest Queen Anne, Madrona, Interlaken, and Alki.  The pattern for natural landslides (slide records that did not indicate human influence), as shown on Figure A-8, mimics the general map of landslide locations.  Except for Southwest Queen Anne, the most dense areas of natural landslides appear to be the same as those that have the highest density, considering all landslides.  This is no coincidence because those areas that are naturally unstable are more likely to continue unstable behavior when human activity disturbs the area than relatively stable areas.  It has long been recognized that disturbance of the ground surface and improper drainage increases the frequency of landsliding.

The chart on Figure 1-9 indicates that only about 13 percent of the landslides recorded citywide were totally natural.  For three percent of the events, there was not enough data to categorize if the slide was natural or influenced by human activity.  About 84 percent of the landslides were determined to have some factor of human influence.  This is consistent with other studies and estimates, including the estimated 80 percent in Landslides in Seattle, 1974.

Four maps (Figures A-9 through A-12) present the distribution of each of the four landslide types.  In addition, a chart presented on Figure 1-10 indicates the percentage of each of the types of slides.  The majority citywide (68 percent) were shallow colluvial slides, followed by deep-seated landslides at 20 percent.  High bluff peeloff  (3 percent) and groundwater blowout (6 percent) landslides were small percentages.  A combination of shallow colluvial and deep-seated landslides accounted for 88 percent of the total landslides.  Three percent of the landslides could not be categorized because of insufficient information in the records.

The distribution of high bluff peeloff landslides is shown on Figure A-9.  This type of slope instability (illustrated on Figure 1-2) only occurs on precipitous cliffs, which are normally comprised of till or sand.  Many, but not all, of these landslides are naturally triggered.  They are either associated with the headscarps of deep-seated landslides or old sea bluffs that formed prior to the construction of shoreline protection.  Such high bluffs are found in only a few locations:  Perkins Lane, Northwest Seattle, and Southwest Queen Anne.

Groundwater blowout landslides (illustrated on Figure 1-3) are spread around the city, as shown on Figure A-10.  They are a direct indicator of the sand/clay contact, where high groundwater pressures commonly exist.  This landslide type may be more common than indicated by this database, but they are difficult to discern from the older records.  Only those slides where an engineer or geologist noted the characteristics of this type of slide were placed in this category.  Some of these slides are natural because they are related to groundwater, which is more likely than not from natural sources.  Groundwater blowout landslides occurred in Northwest Seattle, Southwest Magnolia, Southwest Queen Anne, Alki, and West Beacon Hill (I-5). 

The locations of deep-seated landslides are shown on Figure A-11.  They are located in significant numbers in the following areas:  Southwest Magnolia, Northwest and Southwest Queen Anne, East Queen Anne, Alki, Admiral Way, West Beacon Hill (I-5), Interlaken, Madrona, and Pigeon Point.  Because deep-seated landslides (illustrated on Figure 1-4) are dependent on regionally recharged groundwater, these slides are mostly natural.  This type of slide commonly encompasses several properties and sometimes one or more city blocks.  

As shown on Figure 1-10 and Figure A-12, shallow colluvial landsliding is the most prevalent and widespread type in Seattle.  The areas with the highest densities of shallow colluvial land­slides include Burke-Gilman Trail, Laurelhurst, Madrona, Rainier Avenue S.E., Alki, Beach Drive S.W., East Queen Anne, Southwest Queen Anne, Southwest Magnolia, and 47th Avenue S.W.  Although the distribution of this type of slide (illustrated on Figures 1-5 and 1-6) indicates that they follow the overall pattern of landslides in the city, they often occur outside of the areas where natural slides occur because a shallow colluvial slide is the type of landslide most likely to be caused by human activity.

In its two most basic elements, a landslide can be categorized as natural or human influenced.  Virtually all landslides in Seattle occur where natural factors are conducive to landsliding, but many are also influenced by human activity.  It is normally difficult to discern the percentages of contribution between these two elements to landsliding.  Likewise, it is very difficult to assign percentage of contribution among the many human-influenced contributing factors in a landslide.  The natural factors that contribute to landsliding (geologic conditions, topography, freezing and thawing, heavy or prolonged precipitation, and natural groundwater seepage, among others) are conditions to be accepted.  For sites or areas located on or near slopes, there is always a risk that instability can occur.  Engineering solutions are generally available to reduce the risks to acceptable levels of safety; however, there will always be risks.  As shown on Figure 1-9, totally natural slides only comprise about 13 percent of the total number of slides in the records reviewed.  Their distribution is shown on Figure A-8.  In general, totally natural slides are most likely deep-seated, groundwater blowout, or high bluff peeloff landslides.  Deep-seated slides and groundwater blowout are influenced most often by regional groundwater sources.  High bluff peeloffs are in mostly inaccessible locations that are not susceptible to human disturbance.

Some factor of human influence was reported in 84 percent of the landslides citywide in Seattle.  The implications of this are that there are measures that can be implemented by the City and by private property owners to reduce the risk of damage to public and private properties.  Some of the human reasons that contribute to landsliding in Seattle include improper drainage/subdrainage, broken or leaking pipes, excavation at the toe of a slope, fill placement at the top or side of a slope, imprudent cutting of vegetation, and the lack of maintenance of drainage facilities or vegetative cover.  All of these factors have been chronicled in the files that were reviewed for this landslide inventory and study.

The percentage of reported landslides with some factor of human influence noted in this study may be high with respect to the total number of landslides that have taken place within the City.  The actual percentage of landslides with such human influence may actually be lower.  The reasons for this are as follow:

    Only the reported slides were included in the database that provided the basis for this study.

    Many of the reported landslides were those where people were making claims usually against the City.

    The reported landslides were generally in developed areas, and totally  natural landslides in other areas may not have been reported.

Nevertheless, property owners and developers need to realize that human influences can be significant contributing factors to instability.  It is thus imperative that competent professional advice be obtained to reduce the risks of landsliding and damage.  On the other hand, the influence of geology in this area must also be recognized as a significant contributing factor in landslides since, with or without some factor of human influence, most of the instability occurs on the steep slopes that surround Seattle's ridges, near the sand-clay contact, or at other locations where adverse soil layering and groundwater conditions are present. 

The City of Seattle presently regulates public and private development in environmentally-critical areas by requiring special standards for design and construction in potential slide areas (and known slide areas) and steep slope areas.  Potential and known slide areas are defined by historical landslides and by a zone encircling many of the hills and ridges based on the sand/clay contact as shown in Tubbs' Landslides in Seattle, 1974.  Steep slopes are defined as slopes steeper than 40 percent, with a rise exceeding 10 vertical feet.  These restricted areas are shown on maps prepared by the DCLU.  If a proposed new development is within one of these zones, geotechnical evaluations must be completed to obtain a permit for construction. 

Some of the benefits of accurately delineated potential slide areas in Seattle are for zoning, administration of construction permits, notification for landslide education and public meetings, and emergency notification.

Some of the pitfalls for such accurate mapping may be complaints from property owners who did not want to be included within a restrictive zone, and the variation in accuracy of the data and geologic conditions on an individual property.

A common theme of homeowners who object to being included in a potential slide area is that their property values would be diminished.  It has been our experience that property values can be reduced temporarily for one or two years when landsliding is active on a property; however, upon remediation of the instability, the property values revert again to the same or higher value as before the landslide occurred.  Most of the property that is in a potential slide area is also view property that has risen steadily in value, unless an individual property is impacted by a landslide without suitable remediation.  Therefore, it is beneficial for a public agency to accurately map and regulate construction in such sensitive lands to assist the public to prevent unwarranted reduction of property values.  Knowledge of the potential landslide risk and education of property owners (including public owners) are the most effective methods to maintain property value in the long-term.

Of the total number of landslides in the database, 58 percent were within existing mapped potential slide areas and 76 percent were within the existing mapped steep slope areas.  Note that 24 percent of the landslides occurred on slopes flatter than 40 percent.  This shows the need for improving the potential slide area mapping.  The percentage of landslides within either the steep slope or existing potential slide areas was 88.  Nine areas in Seattle were identified where clusters of slides were outside of the potential slide areas.  Figure A-13 is a map showing the landslides in Seattle in relationship to the potential slide areas.  Those areas where significant groups of slides were outside the designated potential slide areas include:  Interlaken, North Capitol Hill, Laurelhurst, Shilshole (south end of Northwest Seattle), the hillside west of Burke-Gilman Trail, Seola Beach, Rainier Avenue S.E., Mount Baker Ridge (south of I-90, at south end of Madrona), Admiral Way S.W. (southern end), Alki (high elevations), and 47th Avenue S.W.  As shown on a map of landslides in relationship with steep slope areas (Figure A-14), the outliers are primarily scattered, isolated occurrences, but five areas have concentrations of slides outside the steep slope areas:  Burke-Gilman Trail, Northwest Queen Anne, Admiral Way, east of Lincoln Park (in Beach Drive S.W. area), and the east side of Beacon Hill (Cheasty Boulevard S.). 

A subset of shallow colluvial and groundwater blowout landslides is debris flows, those landslides defined herein as those that flowed more than 50 feet beyond the toe of the steep slope on which they fail.  These slides are significant in that their runout zones do not normally confine themselves to the current potential slide areas, as presently mapped.  As shown on Figure A-15, they are scattered around the city in the chronic landslide areas. 

In our opinion, the above discussion points out the need for further definition of the hazard zones based on landslide prone characteristics and on runout zones that could impact downstream properties.  Please refer to Part 5 for an evaluation of the existing Potential Slide Areas.