Edge-detection-driven first-arrival picking method for borehole radial velocity imaging

Abstract

Accurately determining the radial velocity structure of formations near the borehole is essential for evaluating borehole stability, detecting mud invasion, and optimizing reservoir production. Currently, the most widely used and reliable approach involves calculating the radial velocity of near-borehole formations using first-arrivals from monopole acoustic logging. However, the accuracy of this method is constrained by errors in first-arrival picking, which limits the precision of near-borehole formation velocity imaging. To address this limitation, this study introduces a first-arrival picking method based on image edge detection, aiming to enhance the accuracy of radial velocity imaging near the borehole. Our method improves the accuracy of first-arrival picking through three steps: (1) wavelet transformation, which extracts the signal's time-frequency characteristics; (2) mathematical morphology, which removes noise and enhances image edges; and (3) edge detection techniques, which accurately pick the first-arrivals of seismic signals. Numerical experiments validate the accuracy of the proposed first-arrival picking algorithm under varying signal-to-noise ratio (SNR) conditions for synthetic waveforms, significantly outperforming the conventional short-term average/long-term average (STA/LTA) algorithm. At an SNR of 5 dB, the algorithm reduces the average picking error from 0.43 to 0.07 and the relative error of near-borehole velocity inversion results from 3.843 to 0.131. Field data validation further demonstrates the algorithm's reliability, with imaging results aligning closely with gamma-ray lithology analysis. These findings provide strong technical support for hydraulic fracturing optimization and borehole completion engineering.