Premium
Statistical analysis of non‐stationary voltage recordings in geoelectrical prospecting 1
Author(s) -
Lapenna V.,
Macchiato M.,
Patella D.,
Satriano C.,
Serio C.,
Tramutoli V.
Publication year - 1994
Publication title -
geophysical prospecting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.735
H-Index - 79
eISSN - 1365-2478
pISSN - 0016-8025
DOI - 10.1111/j.1365-2478.1994.tb00248.x
Subject(s) - autoregressive model , noise (video) , signal (programming language) , white noise , gaussian noise , autoregressive integrated moving average , signal processing , computer science , spectral density , gaussian , stationary process , algorithm , time series , statistics , mathematics , artificial intelligence , physics , digital signal processing , quantum mechanics , computer hardware , image (mathematics) , programming language
We study the geoelectrical problem of picking out the useful signal from voltage time series, monitored under conditions of a low signal‐to‐noise ratio and non‐stationary noise. Statistical tests performed at different sites show that geoelectrical noise often belongs to the class of non‐stationary phenomena with non‐Gaussian probability distributions. In such cases, the application of conventional methods of geoelectrical useful signal extraction, based on the stationary white‐noise assumption, gives biased estimates. For the on‐line processing of geoelectrical recordings, we recommend the use of the periodogram technique combined with the Kolmogorov–Smirnov test, a suitable algorithm of which is described in detail. The suggested procedure allows data acquisition to stop as soon as the useful signal power is estimated with a relative error smaller than a predetermined value. Finally, we compare the suggested procedure with the autoregressive approach. The previously used and simpler periodogram method, applied to the solution of problems of this kind, appears to give better performances than the autoregressive analysis.