[Next Section] [Previous Section] [Contents]

DISCUSSION

The estimated amounts of coseismic subsidence are based on evidence from plant macrofossils, abundance of peat (visual and loss on ignition), and diatom assemblages. As previously noted, these paleotidal indicators address only zero, one, and two meters of paleosubsidence. This strategy does not account for settings above the forest edge or below the colonizing marsh. The identification of subtidal settings from diatoms in Willapa Bay yielded large ranges of 0.8 to 3.0 m (2.6 to 9.7 ft.) of estimated subsidence (Hemphill-Haley, 1995a).

Estimates of subsidence in this study rely on the most recent buried wetland deposit. However, the most recent buried wetland deposit may not be representative of other Cascadia earthquakes. For example, the record of burial for the most recent buried wetland deposit in Neawanna Creek, Oregon reveals a fairly weakly subsided muddy peat (high marsh setting) overlain by a rooted mud or a peaty mud (colonizing or low marsh setting). In contrast, the older 1100 yr. buried deposit at Neawanna Creek shows a much stronger difference in lithologies and is more recognizable in cores. For the 1100 yr. buried deposit the transition at the buried contact is from a rooted or barren mud to a muddy peat. Using established criteria this would appear to be stronger evidence for 1 m or more of paleosubsidence. However, using older deposits increases the spatial and temporal differences between the modern wetland setting that serves as an analogue for estimating subsidence. Another problem with older records is that the preservation of microfossil and macrofossil evidence is compromised through taphonomic processes. For these reasons the most recently buried wetland event is justified for preliminary estimates of subsidence related flooding hazards.

The coastal subsidence estimates from southwest Washington and northwest Oregon demonstrate generally decreasing subsidence with decreasing UTM Northing as the sites move west of the maximum trough of subsidence. The data confirm preliminary evidence of this trend presented by Peterson and Darienzo (1995). The data are also consistent with northward projection of the zero-isobase proposed by Briggs (1994). A further test of the regional elastic response model is provided by a plot of the amount of subsidence versus distance (east-west) to the Cascadia trench (Figure 21). This plot shows values of increasing subsidence and assumes a linear relation (first order approximation) from the first zero isobase to the maximum trough of subsidence. Regression statistics for this plot are located in Appendix F. The R² value is 0.62, and the Probability value is 6.89 x 10^-19 which is less than a (0.05) indicating that a linear trend exists. The relation between these two variables implies that the maximum trough of subsidence might lie landward of the coast in north-central Oregon. However, a lack of inland tidal basins in central Oregon precludes a direct test of this hypothesis in that area. Other data are required to substantiate the elastic response model used here. However, the existing data show that for the central Cascadia margin the distance from the coastal site to the trench is positively correlated with the amount of coseismic subsidence, for the most recent buried wetland event.

The estimates of paleosubsidence are based on tidal level indicators calibrated from modern Cascadia margin wetlands. Subsided wetlands in other subduction

zones might be used to compare qualitative changes in organic content, macrofossils, and microfossils across buried wetland contacts (Jennings and others, 1995). However, climatic and tidal variations that affect plant and diatom assemblages might preclude the use of historically subsided wetlands from other subduction zone settings as quantitative analogs to the Cascadia margin. For example, analyses of high marsh foraminiferal assemblages of Valdivia Estuary, Chile are dominated by Trochamminita salsa , which does not dominate high marsh assemblages from mid-latitude coasts of North America and eastern South America (Jennings and others, 1995). As another example, the diurnal tidal range from Mary Island Anchorage, Alaska that is 4.7 m (15.4 ft.) whereas the diurnal tidal range at the 12^Th Avenue Bridge in Seaside, Oregon is 1.8 m (5.9 ft.) (National Oceanic and Atmospheric Administration, 1996). These values differ by a factor of 2.5. Tidal ranges are important in controlling the vertical separation between plant communities. Therefore, small amounts of subsidence at Seaside are possibly magnified by the small tidal ranges there.

Problems also exist with using modern wetland analogs from Cascadia bays. Modern undisturbed wetlands are difficult to find in many Oregon bays. Often the best modern wetland sites are not adjacent to cutbanks or core sites documenting the most recent buried wetland event. Furthermore, the modern wetland sites might experience slightly different tidal ranges, sedimentation rates, and exposure to wind waves than the pre-subsided wetland sites. For example, consider the hydrodynamic funneling in Grays Harbor, WA, specifically near Aberdeen, where the geometry of the Harbor is such that it experiences a current increase of 0.37 m (1.2 ft.) at high tide when compared to the mouth of the Harbor. If the mouth of the Harbor were to experience subsidence induced erosion, then tidal forcing could increase, thereby increasing the tidal range most notably in Aberdeen. These differences could yield inaccuracies in estimating amounts of paleosubsidence from the modern wetland calibrations of paleotidal level. For these reasons broad intervals of estimated subsidence are used e.g., zero, one and two meters with broad error intervals of ± 0.5 m. The use of these broad categories of estimated subsidence yield two additional problems. First, subsidence greater than 2.5 m is not accounted for, and second small amounts of subsidence (0-0.5 m) are not recognized. Nevertheless, the criteria used for estimating paleosubsidence from new and existing core and cutbank data do provide a framework for comparing paleosubsidence throughout bays of the central Cascadia margin.

Flood insurance studies include flood elevation estimates for events of several frequencies. It is appropriate to discuss the impacts of applied subsidence estimates for the different elevations. For example, in Aberdeen, Washington the difference between the 10 yr. flood elevation 2.7 m (8.8 ft.) and the 100 yr. flood elevation 3.1 (10.0 ft.) is only 0.4 m (1.2 ft) (Army Corps of Engineers, 1971). If the coseismic subsidence estimate from Elliot Slough is applied, the resultant elevations are 4.3 m (14.1 ft) and 4.7 m (15.3 ft), respectively. Both the 10 yr. and 100 yr. post-subsidence flooding elevations are greater than the current elevation for the 500 yr. frequency flood at 3.2 m (10.5 ft). The susceptibility of Aberdeen to damaging post-subsidence floods at the 10 yr. frequency should be of concern to city planners and engineers.

The difference between the estimated 100 yr. flood elevation of 3.9 m (12.5 ft.) and the 10 yr. flood elevation 3.5 m (11.5 ft.) near Broadway Drive along the Neawanna Creek, Seaside, Oregon is 0.3 m (1.0 ft). Of interest, the 10 yr. flood following a coseismic subsidence event could have a magnitude greater than a 500 yr flood elevation of 4.0 m (13.0 ft.) at present pre-subsidence conditions.

Other potential submergence events along the central Cascadia margin include global sea level rise. Sea level rise from global warming over the next century is estimated at 0.1 to 0.2 m (0.3 to 0.6 ft.) on the west coast of North America (Barnett, 1984). This global submergence is about 20% to 50% of the abrupt submergence in northern Oregon and about 5% to 20% of that expected for southern Washington that occurred during the most recent regional subsidence event.

Two limitations in this study that are important, but were not addressed above include ecological interpretations of diatoms, morphological modifications to basins resulting from earthquakes, and the precision of flood insurance studies.

A problem in constraining diatom ecology is knowledge of allochthonous vs. autothnous diatoms (Vos and de Wolf 1993) Many studies of modern wetlands do not specify whether live diatoms or simply deceased diatom valves are used for ecological interpretations. This presents a problem in recreating sedimentary environments because deceased diatoms can be transported from the setting that they grew in.

Although this study did not include observations of sampled live diatoms, many of the species that were used for salinity tolerances interpretations were taken from the results of sampled live diatoms analyzed in modern settings from Willapa Bay, Washington (Hemphill-Haley, 1993). Nevertheless, diatoms for this regional study are used to show general water conditions (marine, brackish, and fresh) rather than for identifying key species that might be diagnostic of a certain wetland environment. The use of diatom assemblages should reduce interpretive errors associated with single species associations. Lastly, the top few cm of the modern marsh were used for diatom sampling in this study whereas, for live diatom studies only the top few mm are scraped (Hemphill-Haley, 1993). Modern samples used in this study represent an accumulation of diatoms on the order of a few years or decades rather than a single season or year.

Affects to basin morphology during an elastic response include a shift of the base level of the river landward, along with a positive change in the energy gradient as the seaward portion of the basin experiences subsidence and the landward portion undergoes uplift (Figure 2). This change in the energy gradient could produce greater peak flows, although to what extent is unknown. If the maximum trough of subsidence occurred within an estuary (2 m), then a gradient change could be as great as 3 m (10.0) from 50 km (30 mi.) landward if subsidence and uplift are combined. However, a great earthquake could have an effect upon these river sections in that the river gradient will be diminished by subsidence in the event that the maximum trough of subsidence lies east of the estuary. This could result in a river with the same volume discharging at a slower rate than under pre-subsidence conditions. Additional hydrodynamic modeling is needed to combine the effects of high river discharge, storm surge and post-seismic tidal levels in bay-riverine systems. For this study, it is assumed that peak river discharge potential is at least as great as pre-earthquake conditions. Other conditions that may affect river discharge include landslides temporarily damming rivers followed by critically high peak flow (Weischet,1963) as observed in Chile in 1960.

Flood insurance studies for the bays in this study have neither stream or tide gauge data that totals 100 years of record. Therefore these studies rely on frequency distributions. These historic distributions often lack the necessary field data to confirm extreme events(Dunne and Leopold, 1978). For example, discharge estimates in Tillamook Bay (Federal Emergency Management Agency, 1990) for ungaged areas are based on data from upstream gaging stations and the following equation (Harris and others, 1979).

Q_u = Q_g (Au/Ag)^a (5)

The above equation is useful for the Wilson River area, a gaged site, however, this approach is less useful might for the Kilchis, Miami, Tillamook, and Trask rivers that are ungaged (Hubbard and others, 1994). Furthermore these equations do not address high riverine discharge combined with storm surge events. It is more difficult to combine the effects of coseismic subsidence with the other flood data. For example, how might submergence change tidal level, storm surge, river propagation and discharge hydrodynamics in the bays?

The flooding recently experienced in coastal Oregon (February, 1996) represents a dominant riverine driven component, because tides and storm surges were moderate. Areas particularly impacted by this event were along the Siletz River east of Kernville within Siletz Bay, and the southern flood plain of Tillamook Bay composed of the Wilson, Trask, and Tillamook Rivers (pers. observation, February 1996). By contrast, no flooding was observed in Grays Harbor, and only moderate high water levels below flood stage were experienced in the Necanicum Estuary (Curt Peterson, pers. communication).

[Next Section] [Previous Section] [Contents]