Seismic Multicomponent Image Alignment Using Dynamic Time Warping and Its Application in Areas of Gas Bearing Sediments

Abstract

Abstract Proper alignment of the PP- and PS-wave images is an integral step in multicomponent seismic data processing and may be critical in specialized studies such as elastic seismic inversion. However, carrying out this process on a large 3D project and over many horizons represents a significant challenge. To address this issue, a robust method for time-variant seismic trace alignment is essential and we therefore evaluate a novel solution for this problem – Dynamic Time Warping. Dynamic time warping has recently been introduced as a solution to the seismic trace alignment problem. Here, we share our experience in implementing this method for registration of PP and PS images from a survey acquired offshore Peninsular Malaysia. This area is characterized by a shallow gas bearing overburden which masks the objective zone and results in poor conventional PP reflection images. This imaging challenge significantly complicates interpretation and subsequently the PP-PS registration process fails in the gas affected interval. Keeping this issue in mind, we have implemented two new methods that allow for supervised and unsupervised alignment of PP and PS images. Through this paper we will demonstrate how an initial manual registration of major horizons significantly improves the image alignment and subsequently facilitates the application of automatic high definition alignment methods. In conclusion, we will show results of PP-PS image alignment using dynamic time warping and examples of auxiliary outputs – time shift and Vp/Vs ratio – that can be generated from this process. Introduction Time- and space-variant PP and PS seismic image alignment is an essential step for multicomponent seismic data processing that may provide two important insights (cf. Gaiser, 1996; Fomel et al., 2003; Stewart et al., 2003). First, a combined interpretation of PP and PS images improves our understanding of the subsurface, in particular for areas where either PP or PS reflections are masked. Second, computing the time shift between PP and PS reflections can give us the Vp/Vs ratio. However, detailed registration of large 3D seismic PP and PS volumes over many horizons represents a significant challenge for the interpreter. Also, estimating the Vp/Vs ratio requires differentiation of the time shift function. The quality of this differentiation over time depends on the observability of common reflectors in both images. If the PP and PS reflectivity vary greatly in any part of the data automatic alignment techniques need to be applied with great care. To address these problems, a robust automatic method for seismic image alignment is required. One such method, dynamic time warping (DTW), has recently been introduced as a novel solution to the seismic trace alignment problem (Hale, 2013) and subsequently been applied to the registration of PP and PS images (Hale & Compton, 2013). Here, we present our experience with the implementation of DTW and its application to multicomponent seismic data from offshore Malaysia in an area where gas bearing sands obscure the conventional PP reflection image. For this particular imaging challenge, we have implemented two novel processing sequences that allow the alignment of PP and PS images with or without guidance by an interpreter. We also demonstrate that data preconditioning, through the manual registration of major horizons, greatly improves the image alignment and facilitates the subsequent application of automatic alignment methods. Lastly, we show that in areas of gas bearing sediments, the improved PP and PS image alignment and Vp/Vs ratio estimation using DTW adds significant value to the seismic data.