Usability of records by China earthquake early warning network for ground motion modeling

Since 2020, the China Earthquake Early Warning Network (CEEWN) has acquired extensive strong motion data using dense arrays of microelectromechanical system (MEMS) sensors and force-balanced accelerographs (FBAs), offering new opportunities for near-fault motion studies and regional ground motion model (GMM) development. However, reliability concerns about low-cost MEMS sensors hinder their adoption. We evaluate CEEWN data quality through noise analysis, useable bandwidth assessment, and influencing factors. Analyzing 9007 three-component records from 11 M > 4 earthquakes (7 MEMS models, 6 FBA models), we compared instrument performance, installation methods, and distance effects. Both sensor types showed PGA-correlated bandwidth expansion, with FBAs achieving wider ranges (10 −4 –0.1 cm/s 2 noise levels reflecting ambient conditions). MEMS records displayed clustered noise distributions dominated by instrumental noise, enabling classification into two groups: low-noise instruments (0.04–0.06 cm/s 2 ) and high-noise instruments showing comparable noise levels to California's Community Seismic Network. Wall-mounted configurations increased average noise by factors of 1–1.5 relative to ground installations, though vertical component noise showed marginal reduction. We developed a four-tier classification system for MEMS records: Very Broadband (VBBR), Broadband (BBR), Narrowband (NBR), and Rejected (REJ), with corresponding PGA thresholds for category differentiation. Magnitude-dependent distance limits were established for M5–M8 events across instrument noise categories. This work demonstrates CEEWN MEMS data's value for seismic research despite noise limitations, with proposed classification criteria enhancing its utility for GMM development and engineering applications. • Effects of instrument model, installation style, and signal intensity on MEMS data quality. • Develop PGA thresholds and limiting distances to distinguish signals from noise records. • Facilitate the use of MEMS data and data sharing between Chinese and international communities.