Automatic traveltime picking using the instantaneous traveltime

Introduction

Event picking is an important step in seismic processing. It is used in static corrections, velocity analysis, seismic tomography, amplitude versus offset (AVO) analysis, and geological interpretation. By event, here, we mean a lineup on a number of traces of a seismic record that indicates the arrival of new seismic energy. The arrival of such new seismic energy is denoted by a discontinuity in the values of one or more attributes that characterize the various layers of the subsurface.

The recognition of seismic events is usually based on some measure of coherency or similarity from trace to trace (Marfurt et al., 1998; Bienati and Spagnolini, 1998). The underlying assumption is that when a wave reaches the receiver line, it will produce approximately the same effect on all neighboring receivers. If the wave is strong enough to override other energy arriving at the same time, the traces will look similar during the interval in which this wave arrives. Therefore, coherence is considered a necessary condition for the recognition of any event.

Image processing techniques are also very popular for event picking (Bondár, 1992). Seismic sections are considered as images and seismic events can be thought of as edges on the seismic image. Edge detection or edge linking algorithms are therefore used to detect events usually after edge-preserving filtering (Bondár, 1992) or multiresolution analysis (Maroni et al., 2001) of the image. Other features used to recognize seismic events are amplitude standout, which refers to an increase in amplitude due to the arrival of coherent energy, wave signature, dip moveout and normal moveout (Garotta, 1971).

In this paper, we propose a new methodology for automatic event picking based on a new attribute of seismic signals, which we call instantaneous traveltime. Our approach is based on the assumption that a recorded trace consists of the superposition of sufficiently concentrated signals that arrive at different times. A time-frequency decomposition allows the separation of those signals and the calculation of the traveltimes of different events within a trace. The computed traveltimes are initially time and frequency dependent but the dependency on the frequency is dropped by an appropriate operator. Testing this approach on simple synthetic and field data examples reveals its properties and shows promising results for automated seismic event picking.

Automatic traveltime picking using the instantaneous traveltime