Microtremor Horizontal-to-Vertical Spectral Ratios (mHVSR) at North American Seismic Stations

This database provides microtremor horizontal-to-vertical spectral ratios (mHVSRs) at seismic stations located in North America, primary in the contiguous United States. Three-component continuously archived noise recordings for 5,219 permanent seismic stations in North America were downloaded from the IRIS-DMC and processed using hvsrpy (Vantassel, 2025). The database includes the mHVSR (computed using both the geometric mean of the as-recorded components and azimuthally-rotated components) in addition to the raw microtremor data. The raw microtremor data is provided to allow reproduction of the data and/or the use of different mHVSR-processing settings. Window rejection and resonance identification were performed using the machine-learning supported automated algorithm, AutoHVSR (Vantassel et al. 2023). We present up to four peaks extracted from the computed mHVSR curves, along with the SESAME quality score for the lowest-frequency (i.e., fundamental) peak. As the microtremor recordings were remotely accessed, installation conditions, noise wavefield and subsurface variability (i.e., 1D site conditions) are mostly unknown. However, mHVSR can only be interpreted in the context of such metadata. Therefore, we propose empirical parameters to identify sites that may have SSI effects, anthropogenic monochromatic noise, and significant subsurface variability. The empirical parameters provide site-specific context for interpreting the mHVSR resonances when additional site information is not available. The data is presented in four main folders: Miniseed_files containing the raw noise recordings in MiniSEED format, Station_XML containing station metadata following the FDSN StationXML schema, HVSR_curves containing the HVSR curves and Azimuthal_HVSR_curves containing azimuthally-rotated HVSR curves in the hvsrpy file format. A summary file containing the station metadata, mean HVSR curves, and identified peaks in the HVSR curves is provided in an Excel workbook. A detailed description of all the files in the dataset is provided in the Readme.txt.

The motivation for publishing this dataset is the widespread and expanding use of mHVSR for seismic site response. In particular, recent studies have shown mHVSR as a promising predictor of site effects in ground motion models. However, continued research in this area requires high-quality and consistently-processed mHVSR measurements at strong motion recording sites. While collecting such data at each seismic station using temporary broadband seismometers would be ideal, this is not yet feasible. As such, this dataset provides the community with site-specific and uniformly processed mHVSR curves at 5,219 stations across North America to enable future research. The authors hope that this dataset will serve as a valuable resource to the community for mitigating earthquake hazard in the United States.