Drillability Profiles Derived From 3D Seismic Surveys

Abstract

Abstract A computer algorithm is outlined by which velocities can be derived from three dimensional multi-fold data, based on three dimensional migration before stack. The operation of the algorithm is illustrated with synthetic data for a layered model. It is shown that the velocities derived are practically the same whether the input traces are digitized with 16 bits or 1 bit. The practical significance is that velocity analysis can be extended by this method to areas where the data quality is poor by using 1 bit data. Sign bit digitization is, in effect, an extreme form of trace equalization. The reduction in amplitude information by using sign bits is compensated for by the vastly increased number of measurements spread over an area around the well. A practical application for the method is the construction of drillability profiles based on velocity measurements, with emphasis on the detection of formations with abnormally high fluid pressures. Introduction Three dimensional, multi-fold, seismic surveys generate a notable amount of data for the analysis of subsurface problems. A typical marine survey, for example, may have source points 100 ft apart along parallel lines that are 200 ft apart. Thus, if the data are recorded with a 48 channel streamer, each square mile surveyed contains over 250,000 seismic traces. Three dimensional processing combines all these data to form an image of each point of the subsurface. Similar quantities of data can be acquired on land with the use' of 1024 channel sign bit recording systems. The amplitude recovered on correlating a Vibroseis sign bit recording with a sign bit representation of the sweep signal is remarkably similar to that obtained using 16 bit digitization for both the recording and the sweep signal. This property suggests that in some data processing routines the input data can be made sign bit with little loss in accuracy but with a great gain in data compression and reduction in computation time. An example of this was given by Cochran for CDP stacking velocity analysis. In this paper, we explore a velocity analysis based on three dimensional migration before stack. The input data are migrated for a wide range of velocities using the Kirchhoff method of migration. No approximations are made. The computation is simplified by transforming time to a logarithmic scale, as explained below. The display, instead of being the usual plot of semblance as a function of velocity and time, is a plot of power as a function of depth Z and time T. Maxima in this plot are picked and velocity obtained from the relation V = 2Z/T. The method of computation is illustrated with 16-bit synthetic data to which noise has been added. These data are also reduced to sign bits and the analysis repeated. The velocities inferred in the two cases are shown to be virtually identical.