A dissertation submitted in partial fulfillment of the requirements for the degree of







Copyright by





University of Washington






Chairman of the Supervisory Committee: Professor A. L. Washburn
Department of Geological Sciences and Quaternary Research Center






Influence on landsliding during 1971-1972

While the spatial distribution of landsliding in Seattle during the winter of 1971-1972 was related to certain geologic factors, the temporal distribution appears to have been the result of climatic factors. The specific dates of movement are known for 29 of the 47 landslides included in Appendix 1. The relationship between the dates of landsliding and daily precipitation is shown in figure 22; for landslides having more than one date of known movement only the first date given in Appendix 1 is illustrated. Nearly 75 percent of the landslides occurred on two of the three days during the winter of 1971-1972 having more than 1.5 inches of precipitation and nearly 90 percent of the landslides occurred on three of the four days having over an inch of precipitation.

Tubbs (1974b) discussed possible reasons for the apparent lack of landsliding on January 20, despite nearly two inches of precipitation. Because the several months prior to that date had experienced slightly below-normal precipitation, the suggestion was made that cumulative as well as short-term precipitation might influence landsliding. It was also suggested that a period of sub-freezing weather after January 20 may have increased the subsequent susceptibility to landsliding. General conclusions were drawn concerning the climatic conditions under which future episodes of widespread landsliding could be expected in Seattle, but because the landslides considered in that paper were restricted to a single year it was not possible to adequately develop a predictive model based on climatic factors.

Influence on landsliding during 1932-1972

To facilitate analysis of the relationship of landsliding to annual and cumulative precipitation, the period from August 1, 1932 to July 31, 1972 was divided into 40 precipitation years, each beginning on August 1 and ending on July 31. These dates were chosen because the mean daily precipitation in Seattle is at its minimum in late July and early August and at its maximum in December and January (figure 23). The annual precipitation and number of landslides listed in Appendix 2 for each of the precipitation years are shown in figure 24. There is a significant linear relationship between landsliding and annual precipitation (figure 25), with more landslides occurring in years having high annual precipitation, but the low correlation coefficient and the large time intervals involved limit the usefulness of the diagram for predictive purposes.

In consideration of the apparent influences of daily and cumulative precipitation on landsliding during the winter of 1971-1972, an analysis was conducted of the relationships between cumulative precipitation, short-term precipitation, and the landslides included in Appendix 2 for which the specific dates of occurrence are known. The cumulative precipitation was calculated for each date of reported landsliding, as was the precipitation during five intervals (of length n = 1 to n = 5 days) including the day of the landslide and the (n - 1) preceding days. The number of reported landslides is plotted against cumulative and short-term precipitation in figures 26 through 30. Cumulative precipitation is plotted in 1.0-inch intervals and short-term precipitation is plotted in 0.1-inch intervals. Combinations of cumulative and short-term precipitation for which more than one date of landsliding was reported are plotted as the greatest number of landslides reported for any of the dates rather than as the sum of the number of landslides reported for all the dates.

Several aspects of these figures require explanation. The general clustering of landslide dates toward the center of the cumulative precipitation range appears to be the result of the annual distribution of periods of intense rainfall. Relatively few dates having two or more landslides are shown at cumulative precipitation values of less than about 25 inches, but this is at least partly the effect of inadequate data; several occurrences of two or more landslides at cumulative precipitation values of <20 inches are listed in Appendix 2 but are not plotted in figures 26 through 30 because the dates of occurrence are known only to within two days. Although dates having two or more landslides tend to involve more short-term precipitation than dates having no landslides or only one landslide, the two fields almost completely overlap. There is no obvious relationship between cumulative precipitation and the amount of short-term precipitation necessary to produce widespread landsliding.

These results differ somewhat from those of Nilsen and Turner. They reported a more distinct separation of the conditions under which larger and smaller numbers of landslides occurred in urban areas of Contra Costa County, California, and suggested that less storm-period precipitation was required to produce widespread landsliding as cumulative precipitation increased.

There are several differences, however, between the precipitation parameters used by Nilsen and Turner and those used in this study. Nilsen and Turner used storm-period precipitation, rather than precipitation during intervals of constant length, as the measure of short-term precipitation. Storms were defined as any number of consecutive days having measurable precipitation, and storm periods as periods including storms spaced less than four days apart. Although useful in their study area, these definitions are not useful in Seattle because much of a typical Seattle winter would fit the definition of a single storm period.

Nilsen and Turner attributed all landslides occurring during or immediately following a storm period to that period. However, because the length of a storm period is variable and the amount of storm-period precipitation is partly a function of the length of the period, storm periods receiving greater total precipitation are generally longer and can be expected to include a correspondingly greater number of landslides. Figures 26 through 30 avoid this effect by treating each day as a separate event and relating the number of landslides occurring on that day to the record of previous precipitation, both short term and cumulative. Individually, the figures depict the effect of precipitation intensity and exclude the effect of duration. The durational effects are implicit in the differences between the figures.

The cumulative precipitation expressed in figures 26 through 30 includes the immediately preceding short-term precipitation; this facilitates comparison between the figures because it allows each landslide date to be plotted in the same column in all five figures. Nilsen and Turner, in a similar figure, did not include storm-period precipitation in the cumulative precipitation, but this difference does not account for the differing results of the two studies.

Some of the difference between the results of this study and those of Nilsen and Turner may be due to differences in the nature of landsliding in the two areas. The landslides they described in the San Francisco Bay region generally involved ancient landslide deposits of considerable thickness, while landsliding in the Seattle area usually involves a relatively small thickness of regolith. To the extent that landsliding in both areas is related to the degree of saturation of the potential slide mass, this difference might cause landsliding in Seattle to be more susceptible to relatively short periods of intense precipitation.

The relationships between landsliding and short-term precipitation in Seattle are expressed in figures 31 through 35. The average number of reported landslides per day is based on the landslides included in Appendix 2 for which the specific dates of occurrence are known. Precipitation is plotted in 0.2-inch intervals except where larger intervals are necessary to include at least one landslide (thus allowing the data points to be plotted on semi-log paper); the intervals are indicated for each point.

The relationships can be represented by the regression lines shown in figures 31 through 35. The general form of the equations for these lines is

where L is the average number of reported landslides per day, P is the short-term precipitation (in inches), and a and b are constants. All the relationships are significant at the .01 level and their correlation coefficients range from .92 to .97. The correlation coefficients are highest for the relationships involving one-day and two-day precipitation (figures 31 and 32), and progressively decrease with longer periods. Landsliding in Seattle appears to be most responsive to periods of intense precipitation but short duration. The correlation coefficient for one-day precipitation (.95) is slightly lower than that for two-day precipitation (.97), but this effect is probably due to the larger fraction of one-day precipitation that may occur subsequent to landsliding.

Predictive implications

The regression lines shown in figures 31 through 35 are plotted in figure 36, which has applications in landslide prediction and in the analysis of historic landsliding. The accuracy of figure 36 depends on the constancy of the relationships between precipitation and landsliding and between landsliding and the number of reported landslides.

The relationships expressed in figure 36 can be used to evaluate the influence on landsliding of possible freeze-thaw effects such as decreased strength or increased infiltration capacity. During the 40 precipitation years considered in this paper there were 35 intervals of one day or longer during which the maximum air temperature did not exceed 0 degrees C. The minimum air temperature during these intervals was generally considerably lower, suggesting the possibility of soil freezing; unfortunately, soil temperature data is not available for the 40-year period. Sub-freezing air temperature is necessary, although not sufficient, to produce sub-freezing soil temperature, and in the absence of soil temperature data the specific landslide dates listed in Appendix 2 were compared to the dates of sub-freezing air temperature. Figure 37 shows the cumulative sum of reported landslides during the 10-day periods following each of the 35 intervals of sub-freezing temperature, and the cumulative sum that could be expected on the basis of the relationship between two-day precipitation and landsliding shown in figure 36. (In cases where two or more intervals of sub-freezing temperature occurred within a 10-day period, reported and expected landslides were attributed to only the most recent interval.) The number of reported landslides following intervals of sub-freezing temperature does not significantly differ from the number that could be expected on the basis of short-term precipitation. Freeze-thaw effects do not appear to be an important factor affecting landsliding in Seattle.

Figure 36 predicts the number of landslides expected to be reported in The Seattle Times and for which the specific date of occurrence is reported. In order to predict the number of landslides that actually occur on a particular day as a result of an episode of short-term precipitation, it is necessary to multiply the number given in figure 36 by a scale factor. A reasonable scale factor for predicting the number of landslides that cause significant damage is 10, and a reasonable factor for predicting the total number of landslides is 20. The scale factor for damaging landslides was estimated by multiplying

damaging slides

Federal slides

reported slides



Federal slides

reported slides

specific reported slides

where the first term is the ratio of the number of damaging landslides during the winter of 1971-1972 to the number of landslides included in the Federal disaster assistance records (approximately 1.25:1), the second term is the ratio of the number of landslides included in the Federal disaster assistance records to the number of landslides during the winter of 1971-1972 reported in The Seattle Times (approximately 5:1), and the third term is the ratio of the number of landslides reported in The Seattle Times to the number of those landslides for which the specific date of occurrence was reported (approximately 1.5:1). The scale factor for total landslides was estimated in the same way, except that the ratio of the total number of landslides during the winter of 1971-1972 to the number of landslides included in the Federal disaster assistance records (approximately 2.5:1) was substituted for the first term. The uncertainties in these approximations are sufficiently large that such scale factors were not incorporated into figure 36; instead, the expected number of reported landslides is presented as an index of temporal landslide hazard.



Considerable landscape modification accompanies urbanization, and can be expected to influence both the spatial and temporal distribution of landsliding. More than 80 percent of the landslides included in Appendix 1 involved one or more human factors that may have contributed to landsliding (table 4). The large number of landslides involving human factors may be partly due to the bias of the data, as discussed under Data sources, and may also be partly due to recognizing possible human influences in landslides that would have occurred regardless of the human factors. It is usually more difficult to assess the importance of possible human influences than to determine the geologic and climatic causes of landsliding.

Diversion of water onto (and into) a slope was the most common human factor, noted in more than 40 percent of the landslides. The water was usually the result of runoff from roofs and paved areas, but other sources were occasionally involved. Steepening of slopes by excavation was also recognized in over 40 percent of the landslides. This can contribute to sliding either by the removal of lateral support, often resulting in immediate failure, or by the creation of unnaturally steep slopes upon which debris slides are likely at some future date. Only one landslide listed in Appendix 1 was of the former type, but the latter variety was quite common. The placing of artificial fill upon a slope can contribute to landsliding, especially on steep slopes underlain by an impermeable substrate; over 30 percent of the landslides involved some fill. Finally, about 10 percent of the landslides were associated with retaining wall failures, due to inadequate design, construction, or maintenance.

Although human factors can result in occasional departures from the previously described relationships of landsliding to topography, stratigraphy, and precipitation, they generally reinforce those relationships. The diversion of water onto a slope, for instance, usually occurs during periods of intense precipitation. The creation of artificially steep slopes is most conducive to landsliding in areas underlain by an impermeable substrate. Artificial fill is most susceptible to landsliding during periods of intense precipitation on steep slopes underlain by an impermeable substrate. The widespread involvement of human factors does not greatly diminish the predictive value of the relationships described in this paper.



Landsliding in Seattle is related to certain geologic, climatic, and human factors. Landslides typically occur in areas sloping >15 percent and underlain by either the Lawton Clay or pre-Vashon sediments; they are often associated with the contact between those units and the overlying Esperance Sand. On the basis of these relationships, a slope-stability map was constructed that predicts the spatial distribution of landsliding (plate I). The temporal distribution of landsliding in Seattle is primarily a function of short-term precipitation. Precipitation data can thus be used to estimate daily landslide hazard (figure 36). Human influences affect both the spatial and temporal occurrence of landsliding, and cause occasional departures from these predictions.



 American Commission on Stratigraphic Nomenclature, 1961, Code of Stratigraphic Nomenclature: Am. Assoc. Petroleum Geologists Bull., v. 45, p. 645-665.

Armstrong, J.E., Crandell, D.R., Easterbrook, D.J., and Noble, J.B., 1965, Late Pleistocene stratigraphy and chronology in southwestern British Columbia and northwestern Washington: Geol. Soc. America Bull., v. 76, p. 321-330.

Biederman, D.D., 1967, Recent sea-level change in the Pacific Northwest: Masters research paper, Univ. Washington, 24 p.

Church, P.E., 1974, Some precipitation characteristics of Seattle: Weatherwise, v. 27, p. 244-251.

Crandell, D.R., 1965, The glacial history of western Washington and Oregon, in The Quaternary of the United States: Princeton, N.J., Princeton Univ. Press, 1965, p. 341-353.

Crandell, D.R., Mullineaux, D.R., and Waldron, H.H., 1958, Pleistocene sequence in southeastern part of the Puget Sound lowland, Washington: Amer. Jour. Sci., v. 256, p. 384-397.

_____, 1965, Age and origin of the Puget Sound trough in western Washington: U.S. Geol. Survey Prof. Paper 525-B, p. B132-B136.

Easterbrook, D.J., Crandell, D.R., and Leopold, E.B., 1967, Pre-Olympia Pleistocene stratigraphy and chronology in the central Puget Lowland, Washington: Geol. Soc. America Bull., v. 78, p. 13-20.

Hall, J.B., and Othberg, K.L., 1974, Thickness of unconsolidated sediments, Puget Lowland, Washington: Wash. State Div. Geology and Earth Resources Geologic Map GM-12.

Henkel, D.J., 1967, Local geology and the stability of natural slopes: Jour. A.S.C.E. Soil Mech. and Found. Div., v. 93, p. 437-446.

Hsu, K.J., 1975, Catastrophic debris streams (sturtzstroms) generated by rockfalls: Geol. Soc. America Bull., v. 86, p. 129-140.

Johnson, A.M., 1970, Physical Processes in Geology: San Francisco, Freeman, Cooper and Company, 577 p.

Liesch, B.A., Price, C.E., and Walters, K.L., 1963, Geology and ground-water resources of northwestern King County, Washington: Wash. Div. Water Resources Water Supply Bull. 20, 241 p.

Mackin, J.H., Mullineaux, D.R., and Stark, W.J., 1950, Glacial geology of Seattle: Univ. of Washington, The Trend in Engineering, v. 2, no. 4, p. 19-21.

Miller, R.D., 1973, Map showing relative slope stability in part of west-central King County, Washington: U.S. Geol. Survey Misc. Geol. Inv. Map I-852A.

Mullineaux, D.R., 1961, Geology of the Renton, Auburn, and Black Diamond quadrangles, Washington: Ph.D. thesis, Univ. Washington, 202 p.

Mullineaux, D.R., Nichols, T.C., and Speirer, R.A., 1964, A zone of montmorillonitic weathered clay in Pleistocene deposits at Seattle, Washington: U.S. Geol. Survey Prof. Paper 501-D, p. D99-D103.

Mullineaux, D.R., Waldron, H.H., and Rubin, Meyer, 1965, Stratigraphy and chronology of late interglacial and early Vashon glacial time in the Seattle area, Washington: U.S. Geol. Survey Bull. 1194-0, 10 p.

Nilsen, T.H., and Turner, B.L., The influence of rainfall and ancient landslide deposits on recent landslides (1950-1971) in urban areas of Contra Costa County, California: U.S. Geol. Survey Bull. 1388, [in press].

Palladino, D.J., 1971, Slope failures in an over-consolidated clay, Seattle, Washington: Ph.D. thesis, Univ. Illinois, 188 p.

Stark, W.J., and Mullineaux, D.R., 1950, The glacial geology of the city of Seattle: M.S. thesis, Univ. Washington, 87 p.

Taylor, D.W., 1948, Fundamentals of Soil Mechanics: New York, John Wiley and Sons, 700 p.

Tubbs, D.W., 1974a, Landslides and associated damage during early 1972 in part of west-central King County, Washington: U.S. Geol. Survey Misc. Geol. Inv. Map I-852B.

_____, 1974b, Landslides in Seattle: Wash. Div. of Geol. and Earth Resources Information Circular 52, 15 p.

Tubbs, D.W., and Frederick, J.E., 1974, Two techniques for identifying areas of landslide hazard in Seattle [abs]: Program and Abstracts, Northwest Scientific Assoc., p. 34.

Tubbs, D.W., Dunne, T., and Sternberg, R.W., 1974, Earth sciences component, in Discovery Park Inventory and Natural History Report: Univ. of Washington Institute for Environmental Studies, p. 15-46.

Vames, D.J., 1958, Landslide types and processes, chap. 3 of Landslides and engineering practice: Natl. Research Council, Highway Research Board Spec. Report 29, p. 20-47.

Waldron, H.H., 1967, Geologic map of the Duwamish Head quadrangle. King and Kitsap Counties, Washington: U.S. Geol. Survey Geol. Quad. Map GQ-706.

Waldron, H.H., Liesch, B.A., Mullineaux, D.R., and Crandell, D.R., 1962, Preliminary geologic map of Seattle and vicinity, Washington: U.S. Geol. Survey Misc. Geol. Inv. Map 1-354.



Appendix 1 includes the 47 landslides that occurred in Seattle during the winter of 1971-1972 and are included in the Federal disaster assistance records.

Location of each landslide is given according to the street name and block number. The slides are listed in roughly counter-clockwise order beginning in the southeast comer of the city.

Dates of initial movement are given as precisely as can be determined from newspaper accounts, Seattle Engineering Department records, and discussions with property owners and neighbors. Dates of major subsequent movement are also noted where such information is available. Dates of occurrence that cannot be determined to within two days are listed as indefinite.

Classification is based on the scheme presented by Varnes (1958). Where the type of landslide listed is followed by an arrow and the name of another type of landslide, the slide is believed to have originated as the first type and subsequently developed into the second type.

Slide material refers to the material that moved (i.e. the material above the failure surface).

Substrate refers to the stratigraphic units underlying the regolith. Except for landslides in which the failure surface was totally within the regolith, substrate refers to the material immediately below the failure surface.

Causes listed include only the specific geologic and human factors discussed in the text that are believed to have had contributed to the landslides. No attempt has been made to list the general causes common to all the slides, nor are climatic factors included.




Appendix 2 includes all landslides occurring in Seattle between August 1, 1932 and July 31, 1972 that were reported in The Seattle Times and for which the dates of initial movement are known to within two days. Landslides known to have occurred on specific dates are listed as separate entries from those for which the dates of occurrence are only known to within two days. Where movement was reported at the same location on two or more dates during a precipitation year (from August 1 to July 31) only the earlier movement is listed. The total number of landslides in Appendix 2 is 160.

Return to Tubbs Geosciences' Library