Intraevent Residuals Research Articles

Potential Benefits of Seismic Monitoring for Ground-Shaking Estimation and Loss Assessment

ABSTRACT Seismic networks are fundamental to compute ground shaking in regions affected by destructive events, which can then be used for damage and loss assessment. Ground-motion data recordings can be used to estimate the event bias (or interevent residual) for one or multiple ground-motion models, as well as to condition the ground-shaking estimation process to reduce the intraevent residuals at sites near seismic stations. We evaluate how the error in the estimation of the event bias and ground shaking for a set of target sites can be reduced by incorporating an increasing number of seismic stations, using recordings from the 1999 M 7.7 Chi-Chi, Taiwan, earthquake and a large set of ground-motion fields for Taiwan using stochastic simulations. Then, we evaluate how the incorporation of data from seismic stations in the estimation of several impact metrics (i.e., economic losses, collapsed buildings, and fatalities) can reduce both the bias and the uncertainty, using the district of Lisbon (Portugal) as a case study. The results from this study indicate that the error in the estimation of the impact metrics can be reduced by one order of magnitude if at least 10 stations are considered, especially if the configuration of the seismic network considers the location of vulnerable buildings or the spatial distribution of earthquake risk.

The Community Stress-Drop Validation Study—Part II: Uncertainties of the Source Parameters and Stress Drop Analysis

Abstract As part of the community stress-drop validation study, we evaluate the uncertainties of seismic moment M0 and corner frequency fc for earthquakes of the 2019 Ridgecrest sequence. Source spectra were obtained in the companion article by applying the spectral decomposition approach with alternative processing and model assumptions. The objective of the present study is twofold: first, to quantify the impact of different assumptions on the source parameters; and second, to use the distribution of values obtained with different assumptions to estimate an epistemic contribution to the uncertainties. Regarding the first objective, we find that the choice of the attenuation model has a strong impact on fc results: by introducing a depth-dependent attenuation model, fc estimates of events shallower than 6 km increase of about 10%. Also, the duration of the window used to compute the Fourier spectra show an impact on fc: the average ratio between the estimates for 20 s duration to those for 5 s decreases from 1.1 for Mw<3 to 0.66 for Mw>4.5. For the second objective, we use a mixed-effect regression to partition the intraevent variability into duration, propagation, and site contributions. The standard deviation ϕ of the intraevent residuals for log(fc) is 0.0635, corresponding to a corner frequency ratio 102ϕ=1.33. When the intraevent variability is compared to uncertainties on log(fc), we observe that 2ϕ is generally larger than the 95% confidence interval of log(fc), suggesting that the uncertainty of the source parameters provided by the fitting procedure might underestimate the model-related (epistemic) uncertainty. Finally, although we observe an increase of log(Δσ) with log(M0) regardless of the model assumptions, the increase of Δσ with depth depends on the assumptions, and no significant trends are detected when depth-dependent attenuation and velocity values are considered.

The potential of region-specific machine-learning-based ground motion models: Application to Turkey

Conventional ground motion models have extensively been established worldwide based on classical regression analysis of records. Alternatively, advanced nonparametric machine-learning (ML) algorithms may capture the complex nonlinear behaviour of earthquake motions. This paper investigates the efficiency of artificial neural network (ANN) and extreme gradient boosting (XGBoost) in predicting peak ground acceleration (PGA), peak ground velocity (PGV) and pseudo-spectral acceleration (PSA) (period, T = 0.03–2.0 s) for the Turkish dataset. The dataset involves 1166 records of 383 events with a moment magnitude (Mw) of 4.0–7.6, Joyner and Boore distance (RJB) of 0–200 km, focal depth (FD) less than 35 km, and site condition as the averaged shear wave velocity of the soil on the top 30 m (VS30) of 131–1380 m/s. The performance of the models is compared against empirical models in terms of root-mean-square error (RMSE), coefficient of determination (R2), Pearson correlation coefficient (r), and inter-event and intra-event residuals. To perform residual analysis, a likelihood function is developed. Findings reveal that the XGBoost approach gives an unbiased model with a higher correlation and lower residual than ANN. Finally, an online platform is provided for any interested users.

New ground-motion prediction equations for significant duration of shallow crustal and upper mantle earthquakes in Japan and intra-event spatial correlation

Several models have been proposed and developed to formulate prediction equations for the significant duration (DS) of shallow crustal and subduction earthquakes. However, no equation for DS and a corresponding spatial correlation model have been proposed for upper mantle earthquakes. To address records are selected from the Kyoshin network (K-net) and Kiban Kyoshin network (KiK-net) databases in this study for shallow crustal and upper mantle earthquake with moment magnitudes of 4 ≤ Mw ≤ 7.3, rupture distances Rrup ≤ 300 km and peak ground accelerations (PGAs) ≥ 0.01 g. A new prediction equation is established for DS of shallow crustal and upper mantle earthquakes in Japan by considering the influences of Mw, Rrup, the time-averaged shear-wave velocity to 30 m VS30, the difference between the measured and predicted values of the overburden thickness for a shear wave velocity of 1000 m/s δZ1, and the depth to the top of the rupture Ztor on the DS of ground motion. The rationality and reliability of the proposed prediction model is confirmed using residual analysis and by comparison with existing prediction models. Based on intra-event residuals and single-station intra-event residuals, a semivariogram function and exponential model are used to analyze the spatial correlation of DS, which provides a reference basis for the development of GMPEs based on the nonergodic hypothesis, and regional seismic risk analysis and loss assessment.

Investigating the spatial correlations in univariate random fields of peak ground velocity and peak ground displacement considering anisotropy

The results of seismic risk assessment of spatially distributed infrastructure systems are significantly influenced by spatial correlation of earthquake intensity measures (IM). The assumption of isotropy is a basis for most of the existing correlation models of earthquake IMs. In this study, the isotropy assumption of intra-event residuals of peak ground velocity (PGV) and peak ground displacement (PGD) is investigated by implementing a nonparametric statistical test. Using recorded IMs of 9 earthquakes, it is concluded that there is not sufficient evidence to support the assumption of isotropy in general, and the set of intra-event residuals of PGV and PGD should be considered as the realization of anisotropic random fields. Investigations show that the anisotropy properties of intra-event residuals of PGV and PGD are related to anisotropy properties of local soil characteristics indicated by average shear wave velocity of soil profile from the 30 m depth to the surface (Vs30). Finally, predictive models are proposed based on obtained results in order to simulate the correlated univariate random fields of PGV and PGD considering anisotropy.

Uncertainty Analysis and Spatial Correlation of Ground Motion in the Kanto Basin, Japan

ABSTRACT Spatial correlation of ground motion in the Kanto Basin in Japan is studied based on 5760 dense network seismic records from 61 earthquake events. The inter-event, intra-event and total correlations of zero separation distance ground motion parameters are obtained, and the results show that the inter-event residuals have stronger correlation than the intra-event residuals and total residuals. A prediction equation for the correlation model of ground motion parameters in the Kanto Basin is proposed. Compared with others model, the proposed model obtained better spatial correlation and correlation length of ground motion in the Kanto Basin.

A Region-Specific Ground-Motion Model for Inelastic Spectral Displacement in Northern Italy Considering Spatial Correlation Properties

Abstract The peak inelastic displacement of single-degree-of-freedom bilinear systems (Sdi) is an effective intensity measure linking ground-motion features to the inelastic response and subsequent structural and nonstructural damage of engineered systems. This study develops a region-specific ground-motion model for Sdi considering source, path, and site effects and explicitly accounting for the spatial correlation between intraevent residuals when the model parameters are estimated. The model is developed based on 2427 two-component horizontal ground-motion records from 85 events in northern Italy with magnitudes ranging from 4.0 to 6.4 and source-to-site distances less than 200 km. An exponential stationary and isotropic model is considered to represent the spatial correlation properties of Sdi (after scrutinizing the appropriateness of the underlying assumptions for such a model). Comparisons are performed with existing models in the literature in terms of Sdi estimates, as well as the (spatial correlation) effective range parameter. Two practical applications of the developed model are presented: one on estimating the spatial distribution of Sdi (as an essential ingredient for seismic loss assessments) and one on the engineering validation of region-specific ground-motion simulations. Challenges regarding such validations are also discussed.

Correlation properties of integral ground‐motion intensity measures from Italian strong‐motion records

SummaryThis study investigates the correlation properties of integral ground‐motion intensity measures (IMs) from Italian strong‐motion records. The considered integral IMs include 5–95% significant duration, Housner intensity, cumulative absolute velocity, and Arias intensity. Both IM spatial correlation and the correlation between different integral and amplitude‐based IMs (i.e., cross‐IM correlation) are addressed in this study. To this aim, a new Italian ground‐motion model (GMM) with spatial correlation for integral IMs is first introduced. Based on the newly developed GMM, the empirical correlation coefficients from interevent and intraevent residuals are investigated and various analytical correlation models between integral IMs and amplitude‐based IMs are proposed. The effective range parameter representing spatial correlation properties and the trend in the cross‐IM correlations are compared with existing models in the literature. The variability of the effective range parameters with respect to event‐specific features is also discussed. Modeling ground‐motion spatial and cross‐IM correlations is an important step in seismic hazard and risk assessment of spatially distributed systems. Investigating region‐specific correlation properties based on Italian strong‐motion records is of special interest as several correlation models have been developed based on global datasets, often lacking earthquakes in extensional regions such as Italy.

Non-linear Principal Component Analysis of Response Spectra

ABSTRACT The present work aims at exploring the application of nonlinear principal component analysis in dimensionality reduction and prediction of response spectra. The evaluation is performed based on log10 scaled response spectra at 91 spectral periods corresponding to 13552 records available in the NGA-West2 database. The non-linear principal component analysis performed on the data showed that 91 spectral periods can be addressed with just 3 principal components. Further, an artificial neural network (ANN) model is developed to predict these three principal components with magnitude, distance, shear wave velocity and focal mechanism as input. The inter- and intra-event residuals obtained for the response spectra predicted using the developed model are comparable with the existing ground motion prediction equations (GMPEs) from the same database. The developed model is also observed to capture all the prominent attenuation features of ground motions. Hence, the study indicates that the response spectra can be described with just three uncorrelated variables.

Investigation of anisotropic spatial correlations of intra-event residuals of multiple earthquake intensity measures using latent dimensions method

SUMMARYConsidering spatial correlation of multiple earthquake intensity measures (IMs) is of particular importance in loss assessment of spatially distributed assets. This subject has been investigated in previous studies under the assumption of isotropy. Considering the fact that the assumption of isotropy is not valid in general, the present study employs a non-separable covariance model based on latent dimensions method to investigate anisotropic properties of spatial correlations and cross-correlations of intra-event residuals of multiple earthquake IMs. This method leads to the generation of valid covariance matrix in order to model anisotropic spatially distributed multivariate random fields. Two sets of IMs are considered in this study; the first set consists of peak ground intensity values (acceleration, velocity, and displacement), and the second set consists of spectral accelerations at three different periods. Data of 10 earthquake events in California and Japan are utilized in this study to estimate parameters of marginal and cross-covariance models. Moreover, parameters of covariance model of regional site condition, which is considered as average shear wave velocity of top 30 m of soil profile (Vs30), are obtained in order to investigate the effect of local sited conditions on spatial correlations of IMs. It is shown that maximum range and anisotropy ratio of covariance models of intra-event residuals of IMs are correlated with those of Vs30 values. Also, it is observed that the anisotropy direction of residuals of IMs is consistent with anisotropy direction of Vs30 values. Finally, predictive models are proposed to obtain marginal and cross-covariance functions for different earthquake IMs considering anisotropy.

Ground‐motion intensity measure correlations observed in Italian strong‐motion records

SummaryGround‐motion models (GMMs) are widely used in probabilistic seismic hazard analysis (PSHA) to estimate the probability distributions of earthquake‐induced ground‐motion intensity measures (IMs) at a site, given an earthquake of a certain magnitude occurring at a nearby location. Accounting for spatial and cross‐IM correlations in earthquake‐induced ground motions has important implications on probabilistic seismic hazard and loss estimates. This study first develops a new Italian GMM with spatial correlation for 31 amplitude‐related IMs, including peak ground acceleration (PGA), peak ground velocity (PGV), and 5%‐damped elastic pseudo‐spectral accelerations (PSAs) at 29 periods ranging from 0.01 to 4 seconds. The model estimation is performed through a recently developed one‐stage nonlinear regression algorithm proposed by the authors, known as the Scoring estimation approach. In fact, current state‐of‐practice approaches estimate spatial correlation separately from the GMM estimation, resulting in inconsistent and statistically inefficient estimators of interevent and intraevent variances and parameters in the spatial correlation model. We test whether this affects the subsequent cross‐IM correlation analysis. To this aim, based on the newly developed GMM, the empirical correlation coefficients from interevent and intraevent residuals are investigated. Finally, a set of analytical correlation models between the selected IMs are proposed. This is of special interest as several correlation models between different IMs have been calibrated and validated based on advanced GMMs and global datasets, lacking earthquakes in extensional regions; however, modeling the correlation between different IM types has not been adequately addressed by current, state‐of‐the‐art GMMs and recent ground‐motion records for Italy.

Ground-motion models for predicting vertical components of PGA, PGV and 5%-damped spectral acceleration (0.01–10 s) in Iran

This paper presents ground motion models to predict vertical components of peak ground acceleration, peak ground velocity, and 5%-damped response spectral acceleration at periods ranging from 0.01 to 10 s. These models are derived from the comprehensive Iranian strong ground-motion database containing 1350 three-component time-histories recorded during 370 earthquakes with 4.5 ≤ $${\text{M}}_{\text{w}}$$ ≤ 7.4. Possible regional dependency of vertical ground-motions to characteristics of three seismo-tectonic regions in Iran was investigated through analysis of variance (ANOVA) technique. Consistent models based on four distance measures; two for point-source ( $${\text{R}}_{\text{epi}}$$ and $${\text{R}}_{\text{hyp}}$$ ) and two for finite-fault ( $${\text{R}}_{\text{JB}}$$ and $${\text{R}}_{\text{rup}}$$ ), were developed. The distribution of inter and intra-event residuals presents satisfactory agreement between actual data and our proposed vertical ground motion prediction equation (GMPE). Furthermore, a detailed comparison between our predictions and four local and global GMPEs is presented, well as horizontal model of Darzi et al. (Bull Seismol Soc Am. https://doi.org/10.1785/0120180196 , 2018). In addition, this study provides period-dependent correlation coefficients between epsilons of different GMPEs corresponds to either horizontal, vertical, or vertical-to-horizontal spectral ratio intensity measures as a function of single or two-periods of interest.

Probabilistic seismic hazard analysis of Iran using reliability methods

Structural reliability methods are employed in this paper for nationwide probabilistic seismic hazard analysis of Iran and in-depth seismic hazard and hazard sensitivity analysis of Downtown Tehran. The companion paper proposes two analysis approaches based on reliability methods and presents the technical framework for such analyses. The first approach was based on Monte Carlo sampling reliability analysis, which is employed here to conduct hazard analysis at a grid comprising 3695 sites throughout Iran. The results are used to produce nationwide hazard maps for the peak ground acceleration, spectral acceleration, and cumulative absolute velocity. The other analysis approach employs the first- and second-order reliability methods and importance sampling and is applied to the seismic hazard analysis of Downtown Tehran. This analysis identifies the seismic sources that are most influential on the seismicity of Tehran. In addition, reliability sensitivity and importance measures are employed to identify the most influential sources of uncertainty on the seismic hazard at Downtown Tehran. For instance, they identify the geometry of, and the rupture location on, adjacent faults, the shear wave velocity, and the intra-event residuals of ground motion prediction models as the most important sources of uncertainty.

Determination of GMPE functional form for an active region with limited strong motion data: application to the Himalayan region

Advancement in the seismic networks results in formulation of different functional forms for developing any new ground motion prediction equation (GMPE) for a region. Till date, various guidelines and tools are available for selecting a suitable GMPE for any seismic study area. However, these methods are efficient in quantifying the GMPE but not for determining a proper functional form and capturing the epistemic uncertainty associated with selection of GMPE. In this study, the compatibility of the recent available functional forms for the active region is tested for distance and magnitude scaling. Analysis is carried out by determining the residuals using the recorded and the predicted spectral acceleration values at different periods. Mixed effect regressions are performed on the calculated residuals for determining the intra- and interevent residuals. Additionally, spatial correlation is used in mixed effect regression by changing its likelihood function. Distance scaling and magnitude scaling are respectively examined by studying the trends of intraevent residuals with distance and the trend of the event term with magnitude. Further, these trends are statistically studied for a respective functional form of a ground motion. Additionally, genetic algorithm and Monte Carlo method are used respectively for calculating the hinge point and standard error for magnitude and distance scaling for a newly determined functional form. The whole procedure is applied and tested for the available strong motion data for the Himalayan region. The functional form used for testing are five Himalayan GMPEs, five GMPEs developed under NGA-West 2 project, two from Pan-European, and one from Japan region. It is observed that bilinear functional form with magnitude and distance hinged at 6.5 M w and 300 km respectively is suitable for the Himalayan region. Finally, a new regression coefficient for peak ground acceleration for a suitable functional form that governs the attenuation characteristic of the Himalayan region is derived.

Comparison of Strong Ground Motion Recordings of the Lushan, China, Earthquake of 20 April 2013 with the Next Generation Attenuation (NGA)‐West2 Ground‐Motion Models

Abstract A total of 39 free‐field accelerograms in the 2013 M w 6.7 Lushan earthquake, with the closest site‐to‐rupture distances ( R rup ) being less than 200 km, are compared with the median ground motions predicted by the recent Next Generation Attenuation (NGA)‐West2 ground‐motion prediction equations (GMPEs). Overall, for most Lushan station recordings over the whole range of R rup , the spectral acceleration (SA) of the peak ground acceleration (PGA) and of T =0.06 s are within one standard deviation of the predictions from the five NGA‐West2 GMPEs. Most intraevent residuals also fall between plus and minus one standard deviation from the predictions. Moreover, the standard deviation of intraevent residuals are smaller for SA of T =0.06 s than for PGA, which suggests that the application of NGA‐West2 in predicting high‐frequency ground motion due to the Lushan earthquake is acceptable. The spatial contours of intraevent residuals for PGA predicted by the five NGA‐West2 models are almost the same, all showing four regions of negative intraevent residuals. Three of these regions are mainly distributed with the source‐to‐site azimuth approximating to 90° or −90° and the other one approximating to −180°, which are consistent with the extrusion and rupture directions of the Lushan earthquake fault. In general, the five NGA‐West2 GMPEs generate relatively large intraevent residuals at the long period ( T =1.0 s) for R rup exceeding 100 km, although the intraevent residuals are not as large. The short‐period hanging‐wall effects are more consistent with the NGA‐West2 hanging‐wall scaling than those of long period, and the short‐period ground motions on the footwall are more consistent with the NGA‐West2 models than those on the hanging wall.

Applicability of ground-motion prediction equations to a Greek within-slab earthquake dataset

In the present work I investigate the compatibility of the Abrahamson et al. (Earthq Spectra 32:23–44, 2015) and Zhao et al. (Bull Seismol Soc Am 96:898–913, 2006a) ground-motion prediction equations (GMPEs), developed using non-Greek data from subduction earthquakes, with the magnitude, distance and site scaling suggested by a Greek dataset from subduction, within-slab, earthquakes. The dataset is comprised of recordings from force-balance accelerometers as well as broadband seismometers. The motivation for this study comes from (a) the relatively small number of recordings from large subduction, within-slab, events which to a certain extent limits the applicability and accuracy of the regional subduction GMPEs and (b) the necessity of capturing the epistemic uncertainty in the probabilistic seismic hazard studies for the broader Aegean area, which is currently misrepresented due to the small number of available regional or properly adjusted worldwide ground-motion prediction models. For the Abrahamson et al. (2015) model, the analysis shows that the magnitude scaling implied by the dataset requires moderate adjustments to the magnitude terms. The study of intra-event residuals with distance revealed strong positive trends for the back- and negative trends for the fore-arc data, especially at short periods. The original analysis and the evaluation of the adjusted Zhao et al. (2006a) model revealed that the events in the deeper parts of the subduction (D > 140 km) introduce a significant bias to the model, which results in unrealistically high spectral values for events that occur in that shallower parts of the subduction. Consequently, the applicability of the proposed adjusted Zhao et al. (2006a) model, should be limited to events with D < 125 km. The adjusted Abrahamson et al. (2015) and to a certain extend Zhao et al. (2006a) along with the Skarlatoudis et al. (Bull Seismol Soc Am 103:1952–1968, 2013) GMPEs, form a suite of models that accurately describe the ground motions for the Hellenic subduction. The use of these models in future probabilistic seismic hazard studies in the Hellenic subduction zone will adequately represent the current knowledge of subduction GMPEs and result in more realistic estimates of the seismic hazard in the area.

An Improved Approach for Nonstationary Strong Ground Motion Simulation

A new stochastic ground motion model for generating a suite of ground motion time history with both temporal and frequency nonstationarities for specified earthquake and site characteristics is proposed based on the wavelet method. This new model is defined in terms of 6 key parameters that characterize the duration, evolving intensity, predominant frequency, bandwidth and frequency variation of the ground acceleration process. All parameters, except for peak ground acceleration (PGA), are identified manually from a database of 2444 recorded horizontal accelerations. The two-stage regression analysis method is used to investigate the inter- and intra-event residuals. For any given earthquake and site characteristics in terms of the fault mechanism, moment magnitude, Joyner and Boore distance and site shear-wave velocity, sets of the model parameters are generated and used, in turn, by the stochastic model to generate strong ground motion accelerograms, which can capture and properly embody the primary features of real strong ground motions, including the duration, evolving intensity, spectral content, frequency variation and peak values. In addition, it is shown that the characteristics of the simulated and observed response spectra are similar, and the amplitude of the simulated response spectra are in line with the predicted values from the published seismic ground motion prediction equations (SGMPE) after a systematic comparison. The proposed method can be used to estimate the strong ground motions as inputs for structural seismic dynamic analysis in engineering practice in conjunction with or instead of recorded ground motions.

Macroseismic intensity prediction equations for Vrancea intermediate-depth seismic source

This paper focuses on the development of a macroseismic intensity prediction equation (MIPE), for Vrancea intermediate-depth seismic source, that is able to differentiate between the sites located in the fore-arc and back-arc regions of the Carpathians Mountains. The database consists of more than 9000 observed data points (pairs of sites and macroseismic intensities) for six strong earthquakes originating from Vrancea intermediate-depth seismic source. Throughout this paper, the macroseismic intensities are expressed on MSK scale. The coefficients of the MIPE and the standard deviations of the model are obtained through the maximum likelihood method. The MIPE proposed for Vrancea intermediate-depth seismic source is valid for a magnitude range from M w 6.0 to M w 7.7 and for epicentral distances up to 1000 km in the fore-arc region and up to 800 km in the back-arc region. The inter- and intra-event residuals of the model are investigated, and magnitude and distance scaling are discussed. The predictive results obtained with the proposed MIPE are compared with other results from the literature. In the end, the correlation between the observed values of ground motion amplitudes and macroseismic intensities is discussed. In addition, an empirical relation between ground motion amplitudes and macroseismic intensities is obtained.

On the Calculation of Peak Ground Velocity for Seismic Performance Assessment

Ratios of pseudo-spectral velocity (averaged over the period range of 0.5–2 seconds and at a period of 1 second) to peak ground velocity (PGV) were computed for hundreds of ground motion records selected from the PEER NGA Strong Motion Database. Two-stage regression analyses were performed to develop models for the ratios. The inter-event terms of the ratios trend to moderate decrease as moment magnitude increases; the intra-event residuals do not depend on distance. The results of this study provide the technical basis for a procedure presented in the American Technology Council's ATC-58 project, Guidelines for Seismic Performance Assessment of Buildings, for estimating PGV by scaling values of pseudo-spectral velocity.

A Study of Site Effects in Ilan, Taiwan, Based on Attenuation Relationships of Spectral Acceleration

Abstract The purpose of this paper is to study the site effects for Ilan in northeastern Taiwan based on attenuation relationships of spectral acceleration (SA). Over 2660 accelerograms recorded from 92 shallow earthquakes with M w magnitude ranging from 4.0 to 7.7 are used to develop these regional attenuation relationships. Site effects are investigated using the site residuals obtained from the SA attenuation relationships. The results show the standard deviations of residuals decrease by 1.7%–16.4% in horizontal SA after incorporating the site‐effect term V S 30 in the attenuation relationship of periods greater than 0.3 s. Clearly, the attenuation relationships are improved by including the site‐effect term, especially for the longer periods. In addition, from the contour maps of intraevent site residuals relative to an attenuation equation of our model 2 phase 1 for horizontal SA, we find high amplification factors of short period (0.1–0.2 s), reaching 2.7 in Nanao with hard schist rocks. High amplification factors of intermediate periods (0.3–1.0 s) reach 1.9 and 2.0, respectively, in the city of Ilan and Zhuangwei. High amplification factors of long period (1.0–3.0 s) reach 2.1 in Zhuangwei, where there is the greatest sediment thickness. The results call for special attention to the seismic design of structures in these locations. Furthermore, the intraevent residuals of the four site classes converging the 0.15–0.2 s period range are in agreement with the findings by Aki (1988). Accordingly, site classes D and E have the largest ground‐motion amplification factor in the 0.15–0.5 s and >0.5 s period ranges, respectively. Finally, the predominant‐period contour maps deduced from V S 30 , V S 75 , and sediment depth in the Ilan Plain fall in the ranges of 0.4–1.0, 1.0–1.5, and 1.5–8.0 s, respectively. This suggests that construction of buildings taller than nine stories in the Zhuangwei area should pay special attention to avoid resonance of seismic excitation. Online Material: Contour maps of relative corrected site amplification factors at 10 periods and tables of site parameters, list of earthquakes, intraevent site residuals, and corrected site amplification factors.