Do We Still Need Site-Specific Seismic Hazard Analyses in the Era of Multi-Period Response Spectra?

Since the adoption of the Multi-Period Response Spectrum (MPRS) methodology in ASCE 7-22, I have occasionally heard the question: "Do we still need site-specific seismic hazard analyses?" It is a reasonable question.....

After all, one of the historical criticisms of the code-based design spectrum was that it reduced a complex seismic hazard into just two spectral values, SDS and SD1, and then reconstructed the entire spectrum using a generic shape. Important features of the hazard were often lost in the process, particularly on soft-soil sites where the spectral shape can differ significantly from the idealized code spectrum.

The MPRS methodology represents a major step forward.

Most engineers recognize the most visible change: instead of defining the hazard primarily through two spectral values and reconstructing the remainder of the spectrum using a generic shape, the USGS now provides spectral ordinates at multiple periods. This preserves much more of the true spectral shape and avoids losing important features of the hazard, particularly on soft-soil sites.

Equally important, however, is a less visible change in how site effects are incorporated into the hazard calculations. Historically, seismic hazard was computed at reference rock conditions and then adjusted using generic site amplification factors. While practical, this approach inevitably simplified the complex interaction between earthquake ground motions and local site conditions.

The MPRS framework takes a fundamentally different approach. Hazard is now computed directly at the representative VS30 values associated with the ASCE 7 site classes within the ground motion prediction equations themselves. Rather than calculating hazard on reference rock and applying amplification factors afterward, site effects are incorporated directly into the hazard calculations.

This change not only improves the physical representation of site effects, but also enables a finer discretization of site conditions through the introduction of intermediate site classes such as BC, CD, and DE. Historically, sites with meaningfully different characteristics were often grouped into the same broad site class. The additional classes reduce these artificial jumps and provide a more refined representation of site response.

Taken together, these improvements preserve more of the true spectral shape, provide a more realistic representation of site effects, and produce a significantly more sophisticated design spectrum than the two-period approach that engineers relied on for many years.

So does that mean site-specific seismic hazard analyses have become obsolete?

Not necessarily.

While the MPRS methodology has significantly improved the national hazard model, it remains a national hazard model. It is designed to provide a practical and consistent solution for millions of locations across the United States. As a result, certain aspects of the hazard must still be generalized, including the discretization of site conditions into site classes, the representation of epistemic uncertainty through national logic trees, the characterization of seismic sources, and the interpolation of hazard values between grid points.

Many site-specific studies are based on more detailed source characterizations, including fault geometry, segmentation, recurrence relationships, and path effects beyond those incorporated into the national model. In some regions, consulting geotechnical and seismic hazard firms maintain source models that are considerably more detailed than those used in the national maps, particularly for major infrastructure projects and regions with complex seismicity.

Site-specific studies can also evaluate hazard directly at the measured VS30 of a site rather than at the representative VS30 associated with an ASCE 7 site class. In addition, site response depends on more than VS30 alone. Soil layering, impedance contrasts, basin depth, and nonlinear soil behavior can all influence the resulting ground motions and spectral shape, even for sites with similar VS30 values.

Site-specific studies may also employ project-specific logic trees to represent epistemic uncertainty, allowing alternative interpretations of seismic sources, recurrence relationships, and ground-motion models to be evaluated explicitly. In some cases, they may incorporate regional ground-motion models (GMMs) or basin effects that are not fully represented within the national hazard model.

As a result, site-specific studies may produce not only different amplitudes, but also different spectral shapes that more accurately reflect the characteristics of the local hazard.

The national model is ultimately interpolated from a grid, while a site-specific PSHA is computed at the actual site coordinates.

Perhaps most importantly, a site-specific seismic hazard analysis evaluates hazard at the exact project location. The USGS maps, regardless of how sophisticated they become, are still derived from values computed at discrete grid points and then interpolated to the site. For most projects this interpolation is entirely adequate, but in areas where hazard varies rapidly over relatively short distances—particularly near major seismic sources—the distinction between a site-specific calculation and an interpolated value can become important.

This distinction does not imply that the USGS maps are inaccurate. For most projects they provide an excellent representation of the seismic hazard. Rather, it highlights the difference between a national model designed to serve millions of sites and a site-specific study focused on a single location.

Whether these differences are significant enough to justify a site-specific study depends on the project, the site conditions, and the performance objectives of the design.

The introduction of MPRS has unquestionably improved the code-based design spectrum. It preserves more of the spectral shape, represents site effects more realistically, and addresses many of the shortcomings of the historical two-period approach. However, these improvements do not eliminate the fundamental distinction between a national hazard model and a site-specific hazard analysis.

The question is therefore no longer whether MPRS is better than the old methodology—it clearly is. The more interesting question is when the additional complexity and cost of a site-specific study provide meaningful engineering value. That is the topic we'll explore in the next article.