Monica D. Kohler

This study investigates the vertical dynamic response of a 52‐story high‐rise building in downtown Los Angeles using small‐magnitude earthquake data recorded by the Community Seismic Network. While seismic design traditionally emphasizes horizontal ground motion, vertical accelerations can have a significant impact on tall structures, leading to amplification effects and complex wave propagation. Using data‐driven techniques, such as subspace system identification, spectral analysis, and transfer function (TF) estimation, we identify a global vertical mode at approximately 1.86 Hz, along with a potential secondary mode and beating phenomena. Results indicate that vertical‐component motions can influence structural response and design considerations, particularly in buildings with dual lateral systems. Our findings contribute to a deeper understanding of vertical seismic effects on high‐rise buildings, emphasizing the need for a more sophisticated treatment of these effects in earthquake engineering.

The Community Seismic Network (CSN) has partnered with the NASA Jet Propulsion Laboratory (JPL) to initiate a campus-wide structural monitoring program of all buildings on the premises. The JPL campus serves as a proxy for a densely instrumented urban city with localized vibration measurements collected throughout the free-field and built environment. Instrumenting the entire campus provides dense measurements in a horizontal geospatial sense for soil response; in addition five buildings have been instrumented on every floor of the structure. Each building has a unique structural system as well as varied amounts of structural information via structural drawings, making several levels of assessment and evaluation possible. Computational studies with focus on damage detection applied to the campus structural network are demonstrated for a collection of buildings. For campus-wide real-time and post-event evaluation, ground and building response products using CSN data are illustrating the usefulness of higher spatial resolution compared to what was previously typical with sparser instrumentation.

Recent results are presented to illustrate how predictions of tsunami wave impact and tsunami warning mes-sages can be improved by including information about multiple, large-amplitude wave arrivals over longer time durations and at refined spatial resolution. A deployment of ocean bottom seismometers off the coast of southern California recorded the March 2011 Tohoku tsunami on 22 differential pressure gauges. The pressure gauge tsu-nami records across the entire array show multiple large-amplitude, coherent phases arriving one hour to more than 36 hours after the initial tsunami phase. Analysis of the pressure gauge recordings reveals possible locations of the geographical sources that contributed to secondary tsunami arrivals in southern California. A beamform-ing technique is applied to the pressure gauge data to determine the azimuths and arrival times of scattered wave energy. In addition, a backward ray-tracing procedure is applied to a wide range of back azimuth starting values from the pressure gauge array to map possible scattering source locations. The results show several possible candidates of secondary tsunami source structures. These include: (1) southeastern Alaska producing a tsunami arrival 1–2 hours after the first arrival; and elongated bathymetry structures near: (2) the northern Hawaiian Is-land chain producing an arrival 1–2 hours, (3) Papua New Guinea producing an arrival 8–9 hours, and (4) French Polynesia producing an arrival 10–11 hours, all after the first arrival. These results are then incorporated into tsunami warning messages to improve clarity of the hazard threat and protective action guidance, and the specificity of impact location over time. Revised tsunami messages have been tested through online experiments with the public in order to determine how changes in message clarity and specificity affects message receiver understanding, believing, and personalizing, all of which are pre-decisional sense-making activities. The geo-physical results are mapped into modified tsunami warning messages to show how a time-varying hazard could be communicated with more effective message format and content. The results are demonstrating the effects of including clearly described locations, time of impact, and hazard impact consequences on message perception among the public.

Rapid response after earthquakes is vital to mitigate the effects of catastrophic structural failures and to save lives. In particular, critical infrastructure relies on accurate yet low-latency damage detection to facilitate timely responses. Although traditional techniques rely on hand-crafted feature extraction and data interpretation to identify the presence of damage within a structure, deep learning has resulted in newer methods that combine feature extraction and data interpretation in one. However, these models often require exorbitant amounts of data to train, are often computationally expensive, and are difficult to cross-validate with established theory due to their black box nature. In this paper, we present our current workflow that addresses the aforementioned challenges: (i) constructing representative finite-element models of a 15-story steel frame building with simulated damage pattern scenarios in the form of weld fractures, (ii) generating a synthetic waveform dataset for earthquake-induced dynamic response using a well-known dynamic analysis simulator, OpenSees, for the finite-element models and (iii) implementing a deep learning model that performs damage identification in addition to damage localization, i.e., locating the damage. To construct the dataset, we apply recorded earthquake ground motions to the 15-story building model modified to include small-scale damage scenarios imposed at beam-column connections. We then compute accelerations at each of the four corners within the structure for each individual modified model and earthquake input. Our deep learning model consists of a neural operator in conjunction with a machine learning mechanism called self-attention. To our knowledge, our work is the first work exploring the use of neural operators for earthquake-specific vibration-based damage identification. In addition, we present an analysis of our model performance and model fit, and compare our model to a subset of machine learning techniques.

This study presents a spectral scaling method that utilizes transfer functions to generate ground motions for larger magnitude earthquakes using data from smaller events or swarms occurring in the same seismic source region. The method is based on Aki’s theory of universal similarity of earthquake radiation in which the Fourier amplitude spectra (FAS) of far-field radiated body waves can be approximated as a truncated power law with frequency, and far-field body-wave displacements scale as the moment-rate function together with constants that account for radiation pattern and geometric spreading. Assuming self-similarity in earthquake source properties, an FAS of a smaller magnitude earthquake can be scaled through a transfer function to predict the FAS of a larger magnitude earthquake. We assessed the performance of our spectral scaling approach by analyzing large datasets from the 2019 Ridgecrest and 2010 El Mayor-Cucapah earthquake sequences, comparing it with the performance of a commonly used ground motion model (GMM). The results demonstrate the effectiveness of the spectral scaling method compared with the GMM in predicting ground motions, particularly for long-period response in basin areas.

Structural Health Monitoring (SHM) uses pervasive sensors to monitor the health and integrity of buildings and civil infrastructure, significantly reducing maintenance costs and improving safety. Despite its potential, widespread adoption of SHM is hindered by high deployment costs and privacy concerns from building tenants. This work introduces IceSpy as one solution to both problems: reducing the cost and improving the privacy of SHM applications. IceSpy uses a programmable multi-chip systolic array of small, low-power Commercial off-the Shelf (COTS) FPGAs to implement data filtering and differential privacy. Data filtering reduces cost by reducing power-hungry wireless data transmission, leading to smaller batteries and power harvesters. Meanwhile, local differential privacy removes privacy-sensitive information before transmitting it to an untrusted server. Even with additional differential privacy measures, IceSpy achieves a 3× reduction of power consumption and a consequent reduction in battery costs due to decreased wireless communication. This low-power privacy-preserving filtering technology reduces deployment costs and mitigates privacy concerns from potential deployment sites. With many obstacles removed, we are working on deploying IceSpy in dams and commercial buildings to obtain deeper insight into the structural health of buildings and infrastructure.

A source of ground‐motion recordings in urban Los Angeles that has seen limited prior application is the Community Seismic Network (CSN), which uses low‐cost, micro–electro–mechanical system (MEMS) sensors that are deployed at much higher densities than stations for other networks. We processed CSN data for the 29 earthquakes with M > 4 between July 2012 and January 2023 that produced CSN recordings, including selection of high‐ and low‐pass corner frequencies ( f cHP and f cLP , respectively). Each record was classified as follows: (1) Broadband Record (BBR)—relatively broad usable frequency range from f cHP < 0.5 to f cLP > 10 Hz; (2) Narrowband Record (NBR)—limited usable frequency range relative to those for BBR; and (3) Rejected Record (REJ)—noise‐dominated. We compare recordings from proximate (within 3 km) CSN and non‐CSN stations (screened to only include cases of similar surface geology and favorable CSN instrument housing). We find similar high‐ to medium‐frequency ground motions (i.e. peak ground acceleration (PGA) and Sa for T < 5 s) from CSN BBR and non‐CSN stations, whereas NBRs have lower amplitudes. We examine PGA distributions for BBR and REJ records and find them to be distinguished, on average across the network, at 0.005 g, whereas 0.0015 g was found to be the threshold between usable records (BBR and NBR) and pre‐event noise. Recordings with amplitudes near or below these thresholds are generally noise‐dominated, reflecting environmental and anthropogenic ground vibrations and instrument noise. We find nominally higher noise levels in areas of high‐population density and lower noise levels by a factor of about 1.5 in low‐population density areas. By applying the 0.0015 g threshold, limiting distances for the network‐average site condition, based on the expected fifth‐percentile ground‐motion levels, are 89, 210, 280, and 370 km for M 5, 6, 7, and 8 events, respectively.

Coda wave interferometry is applied to data from Community Seismic Network MEMS accelerometers permanently installed on nearly every floor of a 52‐story steel moment‐and‐brace frame building in downtown Los Angeles. Wavefield data from the 2019 M7.1 Ridgecrest, California earthquake sequence are used to obtain impulse response functions, and time‐varying damping ratios and shear‐wave velocities are computed from them. The coda waves are used because of their increased sensitivity to changes in the building’s properties, and the approach is generalized to show that a building’s nonlinear response can be monitored through time‐varying measurements of representative pseudo‐linear systems in the time domain. The building was not damaged, but temporary nonlinear behavior observed during the strong motions provides a unique opportunity to test this method’s ability to map time‐varying properties. Reference damping parameters and velocities are obtained from a month‐long period during which no significant seismic activity had occurred. Damping ratios measured over narrow frequency bands increase by up to a factor of 4 over short time durations spanning the main shock, as well as M > 4.5 aftershocks and a foreshock. The largest damping ratio increases occur for the highest frequencies, and the increase is attributed to friction associated with structural and non‐structural surface discontinuities which experience relative motion and impact during shaking, resulting in energy loss. Shear‐wave velocities in the building’s east–west and north–south directions are found by applying a waveform stretching method to the direct and coda waves. The broadband velocities are reduced by as much as 10% during building shaking, and their restoration to pre‐earthquake levels is found to be a function of shaking amplitudes. Until recently, these techniques had been limited by temporal and spatial sparsity of measurements, but in this study, variations of the impulse response functions are resolved over time scales of tens of seconds and on a floor‐by‐floor spatial scale.

Abstract Large music festivals and stadium concerts are known to produce unique vibration signals that resemble harmonic tremor, particularly at frequencies around 1–10 Hz. This study investigates the seismic signals of a Taylor Swift concert performed on 5 August 2023 (UTC) as part of a series at SoFi Stadium in Inglewood, California, with an audience of ∼70,000. Signals were recorded on regional seismic network stations located within ∼9 km of the stadium, as well as on strong-motion sensors placed near and inside the stadium prior to the concert series. We automatically identified the seismic signals from spectrograms using a Hough transform approach and characterized their start times, durations, frequency content, particle motions, radiated energy, and equivalent magnitudes. These characteristics allowed us to associate the signals with individual songs and explore the nature of the seismic source. The signal frequencies matched the song beat rates well, whereas the signal and song durations were less similar. Radiated energy was determined to be a more physically relevant measure of strength than magnitude, given the tremor-like nature of the signals. The structural response of the stadium showed nearly equal shaking intensities in the vertical and horizontal directions at frequencies that match the seismic signals recorded outside the stadium. In addition, we conducted a brief experiment to further evaluate whether the harmonic tremor signals could be generated by the speaker system and instruments, audience motions, or something else. All evidence considered, we interpret the signal source as primarily crowd motion in response to the music. The particle motions of the strongest harmonics are consistent with Rayleigh waves influenced by scattered body waves and likely reflect how the crowd is moving. Results from three other musical performances at SoFi in summer 2023 were similar, although differences in the signals may relate to the musical genre and variations in audience motions.

Large music festivals and stadium concerts are known to produce unique vibration signals that resemble harmonic tremor, particularly at frequencies around 1-10 Hz.This study investigates the seismic signals of a Taylor Swift concert performed on 5 August 2023 (UTC) as part of a series at SoFi Stadium in Inglewood, CA, with an audience of ~70,000.Signals were recorded on regional seismic network stations located within ~9 km of the stadium, as well as on strongmotion sensors placed near and inside the stadium prior to the concert series.We automatically identified the low-frequency signals from spectrograms using a Hough transform approach and characterized their start times, durations, frequency content, particle motions, radiated energy, and equivalent magnitudes.These characteristics allowed us to associate the signals with individual songs and explore the nature of the seismic source.The signal frequencies matched the song beat rates well, whereas the signal and song durations were less similar.Radiated energy was determined to be a more physically-relevant measure of strength than magnitude given the tremor-like nature of the signals.The structural response of the stadium showed nearly equal shaking intensities in the vertical and horizontal directions at frequencies that match the seismic signals recorded outside the stadium.Additionally, we conducted a brief experiment to further evaluate whether the low-frequency signals could be generated by the speaker system and instruments, audience motions, or something else.All evidence considered, we interpret the signal source as primarily crowd motion in response to the music.Particle motions of the strongest harmonics are consistent with Rayleigh waves influenced by scattered body waves and likely reflect how the crowd is moving.Results from three other musical performances at SoFi in summer 2023 were similar, though differences in the signals may relate to the musical genre and variations in audience motions.

In this study, various system identification approaches are utilized to estimate the dominant, vertical-component modes of a 52-story, steel, moment-and-braced frame building in downtown Los Angeles resulting from vertical seismic accelerations. Tall buildings exhibit complex threedimensional responses during an earthquake due to the varying material and geometric properties along the building’s height. For high-rise buildings, the dynamic response during shaking events is often sensitive to multiple vibration modes, and multiple-mode structural behavior under horizontal ground motion has been extensively studied. However, the vertical component of ground motion can also excite higher modes and vertical-polarity propagating seismic waves. Their effects are seldom studied due to the scarcity of data. Still, they are important because they can provide information on the axial loads on columns or stresses at floor slab connections. The 52-story high-rise, with its dense triaxial sensor array distributed vertically along the height of the building, provides a suitable basis for examining vertical responses. System identification is performed using state-space methods with low-amplitude earthquake data. Given the high spatial density of the building recordings, we show how we can detect modal characteristics and identify the type of deformation that can occur when considering the vertical component of seismic responses.