Refinement of arrival-time picks using a cross-correlation based workflow Jubran Akram ⁎, David W. Eaton University of Calgary, Canada abstract article info Article history: Received 18 November 2015 Received in revised form 21 July 2016 Accepted 23 September 2016 Available online 25 September 2016 We propose a new iterative workflow based on cross-correlation for improved arrival-time picking on microseis- mic data. In this workflow, signal-to-noise ratio (S/N) and polarity weighted stacking are used to minimize the effect of S/N and polarity fluctuations on the pilot waveform computation. We use an exhaustive search tech- nique for polarity estimation through stack power maximization. We use pseudo-synthetic and real microseismic data from western Canada in order to demonstrate the effectiveness of proposed workflow relative to Akaike in- formation criterion (AIC) and a previously published cross-correlation based method. The pseudo-synthetic mi- croseismic waveforms are obtained by introducing Gaussian noise and polarity fluctuations into waveforms from a high S/N microseismic event. We find that the cross-correlation based approaches yield more accurate arrival- time picks as compared to AIC for low S/N waveforms. AIC is not affected by waveform polarities as it works on individual receiver levels whereas the accuracy of existing cross-correlation method decreases in spite of using envelope correlation. We show that our proposed workflow yields better and consistent arrival-time picks re- gardless of waveform amplitude and polarity variations within the receiver array. After refinement, the initial arrival-time picks are located closer to the best estimated manual picks. © 2016 Elsevier B.V. All rights reserved. Keywords: Arrival-time picking Signal processing Cross-correlation 1. Introduction Accurate location of microseismic events relies on the quality of ar- rival picks, which are typically obtained using automatic algorithms on the recorded waveforms for time efficiency. A multitude of algo- rithms exist for arrival-time picking that operate in either time or fre- quency domain and use either single or multi-component data. Some examples of commonly used algorithms include short and long time av- erage ratio (STA/LTA; Allen, 1978), modified energy-ratio (MER; Han et al., 2009), modified Coppens' method (MCM; Sabbione and Velis, 2010), Akaike information criterion (AIC; Zhang et al., 2003) and algo- rithms based on fractals (Jiao and Moon, 2000), cross-correlation (Raymer et al., 2008; De Meersman et al., 2009), neural networks (Gentili and Michelini, 2006), digital image segmentation (Mousa et al., 2011) and higher-order statistics such as skewness and kurtosis (Saragiotis et al., 2002, 2004). Akram and Eaton (2016) provide a comprehensive review on differ- ent single-level, hybrid and multi-level algorithms for arrival-time picking on microseismic data. The single-level algorithms operate on waveforms from individual receiver levels. Unlike multi-level algorithms, these do not take advantage of waveforms from other receiver levels within the array to improve the arrival-time picks. Cross-correlation is an example of a multi-level algorithm that is widely used for time-delay estimation in electrical engineering (Tamim and Ghani, 2009), in the es- timation of static corrections for surface seismic data (Bagaini, 2005) and in microseismic and earthquake data processing for event identifica- tion and phase arrival picking (Raymer et al., 2008; De Meersman et al., 2009; Al-Shuhail, 2015). In this paper, we propose a new iterative workflow based on cross- correlation for improved arrival-time picking on microseismic data. The existing cross-correlation approach (De Meersman et al., 2009, hereafter denoted as DKV's method) is modified to yield better results in the presence of S/N and polarity fluctuations as is typically the case with microseismic data. The effects of varying S/N and polarity flips on the pilot waveform computation are mitigated by using S/N and polarity weighted stacking. An exhaustive search technique is used for polarity estimation through stack power maximization. We use pseudo- synthetic and real microseismic data from western Canada in order to demonstrate the effectiveness of proposed workflow relative to Akaike information criterion (AIC) and DKV's method. The pseudo-synthetic microseismic waveforms are obtained by adding 100 realizations of Gaussian noise into waveforms from a high S/N microseismic event. Furthermore, the waveform polarities for specific receiver levels are flipped to evaluate the performance of proposed workflow. Journal of Applied Geophysics 135 (2016) 55–66 ⁎ Corresponding author. E-mail addresses: akramj@ucalgary.ca (J. Akram), eatond@ucalgary.ca (D.W. Eaton). http://dx.doi.org/10.1016/j.jappgeo.2016.09.024 0926-9851/© 2016 Elsevier B.V. All rights reserved. Contents lists available at ScienceDirect Journal of Applied Geophysics journal homepage: www.elsevier.com/locate/jappgeo