Mittal, HimanshuHimanshuMittalLin, Yueh-HoYueh-HoLinWu, Yih-MinYih-MinWu2026-02-262026-02-262026-01-20https://www.scopus.com/pages/publications/105028112487https://scholars.lib.ntu.edu.tw/handle/123456789/736036Background: The reliable earthquake magnitude estimation is a critical component of earthquake early warning (EEW) systems. Conventional P-wave–based amplitude parameters, such as peak vertical displacement (Pd), are widely used but often suffer from saturation and instability, particularly for larger earthquakes and when low-cost MEMS sensors are employed. The cumulative absolute absement (CAA), a time-integrated displacement parameter, has recently emerged as a promising alternative for improving early magnitude estimation. Methods: This study analyzes strong-motion records from the dense P-Alert low-cost MEMS sensor network in Taiwan to evaluate the performance of CAA for earthquake magnitude estimation. CAA values were computed using P-wave windows ranging from 1 to 5 s after P-wave arrival, using stations within a hypocentral distance of 70 km as well as the nearest six stations. Empirical regression relations were developed to estimate magnitude from CAA and Pd, and the resulting magnitudes were compared with the catalog moment magnitude (Mw). A generalized moment magnitude (Mwg) was additionally used to assess magnitude-scale consistency and bias in small to moderate earthquakes. Results: The standard deviation between CAA-derived magnitude (Mcaa) and Mw decreases systematically with increasing window length, from ±0.383 for a 3 s window to ±0.333 for a 5 s window when using all stations within 70 km. In contrast, Pd-derived magnitudes (Mpd) show larger deviations, reducing from ±0.504 (3 s) to ±0.398 (5 s). Reliable magnitude estimates are also achieved using only the nearest six stations, with standard deviations of ±0.341 (CAA) and ±0.460 (Pd) for the 5 s window. CAA exhibits a stable scaling with earthquake magnitude, while Pd tends to stagnate and underestimate events approaching Mw 7.0. Magnitude scale consistency tests using Mwg confirm the robustness of the proposed CAA relations after correcting for Mw bias. Conclusions: The results demonstrate that CAA provides a more stable and reliable early magnitude estimator than Pd, particularly for low-cost MEMS sensor networks and limited station availability. The reduced dispersion, lower saturation tendency, and robustness across different window lengths highlight the strong potential of CAA for operational on-site EEW systems using dense, cost-effective seismic networks.CAAEarly warningEarthquakeMagnitudeP wavesjournal article10.1186/s40677-025-00358-0