Identifying P-phase arrivals with noise: An improved Kurtosis method based on DWT and STA/LTA

Abstract

A discrete wavelet transform (DWT) and short time average to long time average (STA/LTA)-based Kurtosis algorithm (W-S/L-K method) is proposed to pick the arrival time of the P-phase; this method advantageously combines the outstanding ability of retrieving the P-phase arrival information from wavelet coefficients at high resolutions with inherent reliability in obtaining the P-phase arrival time using the STA/LTA picking method. The W-S/L-K method uses local maximum amplitudes and local kurtosis onsets from the wavelet detail components to determine the P-phase arrival times reliably and accurately. It was tested and verified using microseismic data collected from the Yongshaba mine. The results show that the W-S/L-K method's rates of picking errors smaller than 5ms, 10ms, and 15ms were 58%, 86%, and 97.5%, respectively, and the W-S/L-K method was able to pick higher quality P-phase arrival times than those determined using the Kurtosis, Skewness, STA/LTA, Kurtosis+STA/LTA, and Skewness+STA/LTA methods. The proposed method provides a reliable technique for accurately picking P-phase arrival times, especially for signals with low signal to noise ratios (SNRs), heavy tails, and spikes. Moreover, it is able to detect pure noise.