Info

117 W

116 W

120 W

119 W

118 W

117 W

116 W

39 N

120 W

119 W

118 W

117 W

116 W

120 W

119 W

118 W

117 W

116 W

40 N

39 N

38 N

37 N

120 W

119 W

118 W

117 W

116 W

40 N

39 N

38 N

37 N

120 W

119 W

118 W

117 W

116 W

120 W

119 W

118 W

117 W

116 W

Figure 8.20 Shaded relief map of the Sierra Nevada (SN) and central Great Basin (CGB) showing (a) seismicity and (b) GPS velocities in a fixed North America reference frame (images provided by J. Oldow and modified from Oldow, 2003, with permission from the Geological Society of America). Seismicity data include 1967-2000 events for M < 6 and 1850-2000 events for M > 6 from the United States Geologic Survey (USGS) National Earthquake Information Center (NEIC) and Rogers et al. (1991). CNSB, Central Nevada Seismic Belt; WL, Walker Lane; ECSZ, Eastern California Shear Zone. Ellipses in (b) represent 95% uncertainty limits. Tectonic domains (dashed lines) in (b) are: I, extension; II, strike-slip-dominated transtension; III, normal fault-dominated transtension.

imation of a uniform strain rate across the Coast Ranges yielded a slip rate of ~39 mm a-1, which is consistent with the average slip rates assigned to the main faults using offset geologic and cultural features over time periods ranging from hundreds to several tens of thousands of years. However, in other areas, such as southern and eastern California (Fig. 8.19a) where the fault geometry is very complex, there are large mismatches between the geodetic and geologic slip rates. These mismatches have prompted investigators to use alternatives to the continuum model approach to describe the surface deformation (McCaffrey, 2005; Meade & Hager, 2005; Bos & Spakman, 2005). One of the most useful of these alternative approaches employs block rotations.

Block models of continental deformation provide a framework for incorporating aspects of the long-term, discontinuous deformation caused by faulting into estimates of the velocity field. In these models, calculated fault slip rates take into account the effects of both the rotation of fault-bounded blocks about vertical or inclined axes and the steady-state elastic accumulation of strain (i.e. creep) on or near faults. The blocks are defined as any number of closed polygons on the Earth's surface that cover the modeled region (Fig. 8.19c). In most applications, the block boundaries coincide with major faults; however, in some cases the choice is less clear. Each point inside the blocks is assumed to rotate with the same angular velocity (McCaffrey, 2005). The description of the motion is mathematically similar to methods of estimating the rotations of large tectonic plates (Section 5.3). However, a potential problem is that the use of short-term geodetic data results in elastic strain rates inside the blocks as well as along their boundaries, causing the surface velocities to deviate from the "rigid plate" requirement of plate tectonics.

In the case of southern California, the incorporation of block rotations and the small-scale displacement discontinuities associated with creep on and near major faults has provided a relatively good fit to the available geodetic data (Becker et al., 2005; McCaffrey, 2005; Meade & Hager, 2005). A common way of evaluating the fit of the models involves the calculation of residual velocities, which represent the difference between the modeled and observed values. An example of one of these comparisons is shown in Fig. 8.19d. In this application, crustal blocks were chosen to minimize the residuals, while still conforming to known boundary conditions, such as the orientation of fault traces and the sense of slip on them. The comparison shows that despite the improvement over some continuous models there are still areas of mismatch. In the Eastern California Shear Zone, for example, Meade & Hager (2005) found that slip rates estimated using geodetic data and the results of block models are almost twice as fast as the 2 mm a-1 geologic estimates (Beanland & Clark, 1994) for the past 10,000 years. A similar discrepancy occurs on the San Jacinto Fault. In addition, the modeled slip rates on the San Bernadino segment of the San Andreas Fault are much slower than geologically determined rates for the past 14,000 years. Finding ways to explain and minimize these mismatches remains an important area of research.

One possible explanation of why geodetic and geologic rates commonly mismatch lies with the mechanical behavior of large faults and the vertical extent of brittle faulting within the lithosphere. Because slip on a fault plane near the surface is controlled by its frictional properties (Section 2.1.5), there is a tendency for faults to become stuck or locked for certain periods of time (Section 8.5.2). This locking may result in elastic strain rates that are evident in short-term geodetic data but not in the long-term record of permanent displacements (McCaffrey, 2005). To address this problem, investigators utilize the concept of the elastic locking depth (Savage & Burford, 1973). This depth is defined as the level below which there is a transition from localized elastic strain accumulations on a fault plane to distributed aseismic flow. The value of the parameter is related directly to the mechanical strength of the fault and the geometry of deformation at the surface. Strong faults and wide zones of surface deformation correspond to deeper locking depths.

Published estimates of locking depths for the San Andreas Fault typically range from 0 to 25 km. However, locking depths are not known a priori and, therefore, must be inferred on the basis of seismicity, long-term geologic slip rates, deformation patterns at the surface, or inferences about the rheology of the lithosphere. Locking depths that fall significantly below the predicted depth of the brittle-ductile transition (8-15 km) for a typical geotherm, or below the seismogenic layer, usually require some sort of explanation. In some cases, slow slip rates on the faults have been used to infer relatively deep locking depths for some segments of the San Andreas Fault (Meade & Hager, 2005; Titus et al., 2005). These and other studies illustrate how the choice of locking depth is directly related to inferences about slip rates on or near major faults.

Other reasons why geodetic and geologic slip rates commonly differ may include inherent biases during sampling or changes in the behavior of faults over time. This latter possibility is especially important when the effects of long-term, permanent strains are considered (Jackson, 2004). Meade & Hager (2005) concluded that the differences between their calculated slip rates and geologic slip rates on faults might be explained by the time-dependent behavior of the fault system. In this interpretation, the San Bernadino segment of the San Andreas Fault is less active now than it has been in the past. By contrast, the San Jacinto Fault and faults in the Eastern California Shear Zone are relatively more active now compared with geologic estimates, possibly due to the effects of earthquake clusters. This possibility highlights the importance of combining geologic, geodetic, and seismologic information to better understand the relationship between the short- and long-term (permanent) behaviors of faults.

By incorporating elements of permanent deformation into block rotation models, McCaffrey (2005) found that the largest blocks in the southwestern US, including the Sierra Nevada-Great Valley and the eastern Basin and Range Province, show approximately rigid behavior after all nonpermanent (elastic) strain has been removed from the data. Most of the blocks rotate about vertical axes at approximately the same rate as the Pacific plate (relative to North America), suggesting that, locally, rotation rates are communicated from block to block. This and several other properties of the model support a plate tectonic-style description of deformation in the western USA, where the rotating blocks behave like microplates. Nevertheless, the problem of determining the mechanisms of the defor mation is far from resolved. Many other models have been proposed for this same region (e.g. Flesch et al., 2000) that also fit the geodetic observations and most investigators agree that the deformation probably results from a combination of mechanisms rather than a single one.

Part of the problem of determining the specific mechanisms of continental deformation is that success in fitting geodetic observations neither proves any given model nor precludes other possibilities (McCaffrey, 2005). In addition, the results of mechanical modeling have shown that the steady-state motion of an elastic upper crust is insensitive to the properties of any flow field below it (Savage, 2000; Zatman, 2000; Hetland & Hager, 2004). This latter result means that short-term geodetic observations of deformation between large earthquakes (i.e. interseismic deformation) provide no diagnostic information about the long-term behavior of a viscous layer in the deep crust or mantle. One especially promising area of research suggests that transient deformation following large earthquakes offers the prospect of inferring the rheology of lower viscous layers (Hetland & Hager, 2004). However, presently, the specific mechanisms and the relative contribution of edge forces, basal tractions, and buoyancy forces to the deformation in most regions remain highly speculative, with the results of models depending strongly on the imposed boundary conditions.

8.6 STRAIN LOCALIZATION AND DELOCALIZATION MECHANISMS

0 0

Post a comment