Challenges in the Selection and Scaling of Records for Seismic Analyses: A Geotechnical Engineer’s Perspective

Earthquakes can cause damage or even collapse to structures and the consequence of damage or failure dictates the acceptable level of seismic hazard. Uniform Hazard Response Spectra (UHRS) with certain probability level of exceedance is usually used to define seismic hazard. Seismic design codes either explicitly define UHRS, or provide adequate guidelines to develop site specific UHRS for critical structures. However, UHRS alone is not sufficient to represent seismic hazard and assess or predict seismic performance of structures. Records representing design earthquakes are required to predict seismic displacements and assess seismic performance. With increased reliance on performance-based design for infrastructures, time history analyses are becoming routine in seismic design. However, there is inadequate and sometimes conflicting guidelines on how to select and scale records. UHRS, although a key intensity parameter, is one among many earthquake and intensity parameters that may control seismic displacements. An unexhaustive list of these parameters may include earthquake magnitude, distance, site conditions, mechanism, Arias intensity, significant duration, cumulative absolute velocity, peak ground acceleration, velocity and displacement. Typically, scenario earthquakes representing the sources of hazard are obtained from seismic de-aggregation analyses, and defined in terms of magnitudedistance pairs. A suite of time histories is then selected from past earthquakes similar to the scenario earthquakes, and either uniformly scaled to approximately match the target spectra in the period range of interest, or modified to match the target spectra. Key challenges faced in the selection and scaling of the earthquake records include: how many earthquake records are sufficient to capture the epistemic uncertainty? Is uniform scaling or matching the target spectra better? Should UHRS or conditional mean spectra be used as target spectra? Are large scaling factors acceptable? How to prioritize and define ranges for various earthquake and intensity parameters? What is the impact of these parameters on the computed displacements; How to select records if the contributing sources of hazard are significantly different (for example, local crustal and mega subduction for a site located close to subduction zone)?

This paper presents an overview of the methods used in current practice for the selection and scaling of earthquake records, describes various challenges faced by the practitioners and their impact on the computed seismic displacements or seismic performance through examples. It also makes recommendation for selection and scaling of records.

Thavaraj, T., A. Sy, J. Williams and R. Liu. 2018. “Challenges in the Selection and Scaling of Records for Seismic Analyses: A Geotechnical Engineer’s Perspective,” in Geohazards 7: Canadian Geohazards Conference, June 3-6, 2018. Canmore, AB, Canada. CGS (Canadian Geotechnical Society).

If you are interested in obtaining this publication, please email We may be able to provide you with a copy, depending on copyright restrictions. For more papers visit the technical publications page.