NA4-Seattle, Washington, USA

HOME

Four years of soil moisture and pore pressure observations at sites near Seattle demonstrate spatial, seasonal, and short-term variations in soil wetness and pore pressure and the association of shallow landslide occurrence with times of high soil wetness (degree of saturation exceeds 60–80%; Baum et al. 2005; Godt et al. 2009). Winter storms in 1934, 1972, 1986, 1990, 1996, 1997, and 2001 have triggered tens to hundreds of landslides (Tubbs 1974; Laprade et al. 2000; Chleborad 2000, 2003).In 1999, the USGS began a project to identify precipitation thresholds that might be used to forecast the occurrence of landslides in Seattle.

1 The setting

Landslide alerts or advisories for Seattle, WA began in 2003 during research to develop landslide-warning thresholds for the Seattle area and evolved into an informal, experimental warning system. The USGS issued informal landslide advisories to city officials in connection with two storms in 2003, one storm in 2004, and one storm in 2005. Since 2006, NWS (National Weather Service) has issued landslide alerts based on USGS tracking of rainfall conditions relative to the thresholds.

The Seattle prototype landslide early warning system consists of several components, including real-time monitoring of rainfall, soil moisture, and pore pressure; NWS quantitative precipitation forecasts; automated tracking of rainfall relative to the two thresholds; and a decision algorithm, that guide the use of the thresholds to determine alert level (Chleborad et al. 2008). The USGS computes threshold levels for rainfall at NWS gages at Seattle-Tacoma Airport near Seattle and posts the results on its website and measures the pore pressure at depths of 20, 50, and 100 cm (measured by tensiometers), soil water content at depths ranging from 20 to 200 cm at the USGS field monitoring site near Edmonds, WA (Fig. 32).

Figure 32: Map showing the city of Seattle and locations of historical landslides (small open circles, see Laprade et al. 2000), locations of the National Weather Service rain gage at the Seattle-Tacoma International Airport (in red), and the USGS field monitoring site near Edmonds, WA (in green).

2 The modelling

The EWM uses different thresholds: a cumulative rainfall threshold (CT), a rainfall intensity–duration threshold (ID) and antecedent water index (AWI). Chleborad’s (2000, 2003) cumulative rainfall threshold (CT) compares the amount of rainfall in the last 3 days (72 h) to the rainfall in the previous 15 days. The 3-/15-day cumulative rainfall threshold is based on an analysis of historical precipitation data associated with wet-season landslides that occurred during the period 1933–1997 in Seattle (187 + 108 additional landslides occurred in 1950-1990; Tubbs, 1974; Laprade and others, 2000). To make a prediction of landslides induced by rainfall, a level of landslide activity was defined for which it is reasonable to assume that rainfall is causally involved. The level selected was three or more landslides in a 3-day (72-hour) period. To incorporate the two ideas of antecedent wetness and unusual recent rainfall, two variables were defined: P3 the 3-day precipitation immediately prior to the landslide event and P15, the antecedent precipitation that occurred prior to the 3 days of P3 (Chleborad, 2006). The rainfall threshold thus defined is interpreted as an approximate lower-bound threshold, below which the specified level of rainfall-induced landslide activity (3-day events with three or more landslides) does not occur or occurs only rarely and above which it may occur under certain conditions. In practice, the CT is an indicator of antecedent rainfall that is a precursor to landslide activity. Moreover an intensity–duration threshold (ID) and antecedent water index (AWI) were developed for forecasting major landslide events in the Seattle area (Godt 2004, Godt et al. 2006). The ID is defined as I=82.73D–1.13 (in inches: I=3.257D–1.13), in which I is the average rainfall intensity, in millimeters per hour, for the entire storm, and D is the duration, in hours (Fig. 33). On the basis of observed hourly rainfall, rainstorms were bounded by periods of no rainfall at least 3 hours in duration at individual rain gages.

Figure 33: Rainfall intensity and duration threshold (ID) in inches for Seattle, Washington (after Godt, 2004).

The observation that landslides occur primarily during the rainy season at times when the soil is relatively wet led to the definition of an AWI (Godt et al. 2006). The AWI has dimensions of length and represents the depth of water above or below the amount required to bring a 2-m-deep column of soil to “field capacity” (estimated to be 0.18 m for Seattle area soils; Godt et al. 2006). The estimated field capacity is the basis for the seasonal antecedent rainfall amount threshold (180 mm for Seattle).

In Eqs. 1 and 2, kd is an empirical drainage constant (0.01 for Seattle; Godt et al. 2006), Δt is the time increment (1h), Ii is the current rainfall intensity minus the evapotranspiration rate (obtained from published measurements, where available), and the subscripts t and t−1 refer to the present and previous time steps. At the end of the summer dry season, the initial value of the AWI is set to −0.18 m to represent dry soil, and the rainfall increments (minus evapotranspiration) are added to the AWI until it becomes positive (Eq. 1). The exponential drainage terms in Eq. 2 are applied only after the AWI reaches zero (Godt et al. 2006). The AWI was defined in such a way as to mimic instrumentally observed variations in soil wetness (Baum et al. 2005). However, the index does not account for the time lag that results from downward movement of rainwater through the soil and thus usually leads the actual soil moisture response by several hours. Soil is considered too dry to produce large numbers of landslides when AWI<−0.1; soil is considered wet enough to produce abundant landslides if rainfall also exceeds the intensity–duration threshold when AWI>0.02. To clarify the meaning of these thresholds during a rainfall event in figure 34 are plotted the CT, the AWI, the ID and the alerts in time.

Automated tracking of the AWI has been substituted for soil moisture and pore pressure monitoring since landslides destroyed the sensors in 2006 (Godt et al. 2009).

Figure 34: Rainfall threshold indices, antecedent water index, alert levels, and rainfall at the Seattle-Tacoma International Airport, December 2004–January 2005. The indices indicate how far rainfall conditions at any given time are above (positive values) or below (negative) the thresholds

2 The warning strategy

Four alert levels was defined: Null, Outlook, Watch, and Warning (Chleborad et al. 2008). Exceedance of the CT by observed or predicted rainfall (or exceedance of the ID by predicted rainfall) constitutes an Outlook. An Outlook activates more intense monitoring of weather conditions, soil moisture, and pore pressure (if available, otherwise the AWI). Wet soil conditions (AWI> 0.02 or degree of saturation >60–80%) combined with rainfall exceeding the ID constitutes the highest level, Warning (Fig. 35).

In practice, the Outlook and Watch levels may be more useful than Warning because they allow government agencies adequate time for emergency preparedness planning and response. The NWS notifies government officials and the public when the Watch level has been reached through the use of a special weather statement.

Figure 35: Decisional algorithm

References

Baum RL, Godt JW, Harp EL, McKenna JW, McMullen SR (2005) Early warning of landslides for rail traffic between Seattle and Everett, Washington, USA. In: Hungr O, Fell R, Couture R, Eberhardt E (eds) Landslide risk management. Proceedings of the International Conference on Landslide Risk Management, Vancouver, Canada, May 30–June 3, 2005. Balkema, New York, pp 731–740

Baum RL, Godt JW (2010) Early warning of rainfall-induced shallow landslides and debris flows in the USA. Landslides 7(3):259–272.

Chleborad AF (2000) A method for anticipating the occurrence of precipitationinduced landslides in Seattle, Washington. U.S. Geological Survey Open-File Report 00-469

Chleborad AF (2003) Preliminary evaluation of a precipitation threshold for anticipating the occurrence of landslides in the Seattle, Washington, area. U.S. Geological Survey Open-File Report 03-463

Chleborad AF, Baum RL, Godt JW (2006) Rainfall thresholds for forecasting landslides in the Seattle, Washington, area—exceedance and probability. U.S. Geological Survey Open-File Report 2006-1064

Chleborad AF, Baum RL, Godt JW (2008) A prototype system for forecasting landslides in the Seattle, Washington, Area. In: Baum RL, Godt JW, Highland LM (eds) Engineering geology and landslides of the Seattle, Washington, area: Geological Society of America reviews in engineering geology, v. XX. Geological Society of America, Boulder, pp 103–120. doi:10.1130/2008.4020(06)

Godt JW, Baum RL, Chleborad AF (2006) Rainfall characteristics for shallow landsliding in Seattle, Washington, USA. Earth Surf Processes Landf 31:97–110

Godt JW, Baum RL, Lu N (2009) Landsliding in partially saturated materials. Geophys Res Lett 36:L02403 5pp. doi:10.1029/2008GL035996

Laprade WT, Kirkland TE, Nashem WD, Robertson CA (2000) Seattle landslide study. Shannon & Wilson, Inc. Internal Report W-7992–01. http://www.seattle.gov/dpd/ landslide/study/. Accessed 8 Apr 2009

Tubbs DW (1974) Landslides in Seattle. Washington Division of Geology and Earth Resources Information Circular 52