Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Computational Technology Reviews
ISSN 2044-8430
Computational Technology Reviews
Volume 8, 2013
The State of the Art in Processing Ground Motion Timeseries
N.A. Alexander1 and A.A. Chanerley2

1University of Bristol, United Kingdom
2University of East London, United Kingdom

Full Bibliographic Reference for this paper
N.A. Alexander, A.A. Chanerley, "The State of the Art in Processing Ground Motion Timeseries", Computational Technology Reviews, vol. 8, pp. 93-124, 2013. doi:10.4203/ctr.8.4
Keywords: accelerograph, correction, filtering, deconvolution, wavelet denoising, displacements.

Summary
In this paper, we review the methods employed in the scientific literature for the correction and processing of timeseries data obtained from strong motion accelerographs. Noise/error reduction is the key. For the legacy analogue, the sources of error found in strong motion instruments and their recordings were, (i) instrument noise (caused by low dynamic range, saturation, etc.), (ii) nonlinear instrument system responses (including poor performance at deconvolution DC), (iii) triggering post event start, (iv) analogue data storage and digitization. For digital accelerographs, many of these problems have been ameliorated by improved design. However, the presence of noise caused by instrument tilting/rotating has not been corrected as modern digital accelerographs are still only 3 (translational) axis instruments. The signal processing techniques conventionally applied were baseline correction, band pass filtering, and instrument deconvolution. The least controversial of these techniques is the high-cut filter which acts as an anti-alias filter. Instrument deconvolution does require knowledge of its characteristic frequency response function, which may be unknown. In this case, the authors applied inverse system identification using various adaptive least squares approaches. Low-cut filtering is more problematic, as it assumes that information within some frequency stop-bands is all noise. Noise reduction (at low frequencies) is critical in obtaining ground displacements from ground acceleration timeseries. Boore suggests a piecewise linear detrend; but a better alternative is a de-noising approach. Others use the stationary wavelet transformer to separate the acceleration timeseries into low frequency sub-band (LFS) and high frequency subbands (HFS). The LFS is de-noised and resulting displacements are a better match to GPS results. This highlights one of the main weaknesses of modern instruments, namely that they are still only 3-axis and not 6-axis instruments.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to Computational Technology Reviews
purchase this volume (price £65 +P&P)