Arrival-time picking error of microseismic events

Abstract

The accuracy of arrival time picking yields valuable information for applied geophysics and seismology. Arrival time picking uncertainties are used to weigh the data in inversion algorithms and to define the resolution of reconstructed velocity models. Arrival time uncertainties intrinsically define the accuracy of all traveltime-based measurements, e.g., the precision of acoustic well-logging measurements. Although a wide range of methods for arrival time uncertainty estimation have been proposed in the literature, the physically most prominent ones are based on the probabilistic formulation. In this paper we review two probabilistic approaches for assessment of the lower boundary of picking error which are popular in the signal processing community - the Cramér-Rao Bound (CRB) and the Ziv-Zakai Bound (ZZB). CRB determines the minimum-achievable picking error under the assumption of high signal-to-noise ratio (SNR). It underestimates the error for small SNR. ZZB utilizes a priori information and, hence, provides a more sound result for noisy data. Both CRB and ZZB require a knowledge of the spectral variance of the signal that is often hard to determine in seismic experiments. To overcome this problem, we propose an alternative easy-touse expression that is solely based on the signal-to-noise ratio and the period corresponding to the dominant frequency in the data. The new expression yields comparable results with that of CRB. The numerical example demonstrates that the new expression and CRB provide a reasonable assessment for data with SNR>10 dB. When the data are seriously contaminated by noise (SNR<10 dB), the picking error can be estimated by ZZB. Furthermore, we estimated the picking errors for downhole microseismic monitoring data. The reconstructed errors significantly vary from station to station and are in average three times less than the ones assigned manually by the analyst. Finally, we demonstrate that weighing the data based on the picking uncertainties can improve the resolution of microseismic source localization.

Presentation Date: Tuesday, September 17, 2019

Session Start Time: 9:20 AM

Presentation Start Time: 9:45 AM

Location: Poster Station 3

Presentation Type: Poster