The past three decades have seen the development of an approach, known as the “pattern recognition approach,” to problems of retrieval of information from data. This approach has been successfully used in a number of applications (ICPR-V, 1980), such as in the classification of agricultural crops from remotely sensed satellite data; in the machine recognition of speech and handwritten text; in the computer-vision-based inspection of industrial objects; in radar, sonar, and other defense-oriented applications; in the classification of electrocardiograms and electroencephalograms; and in medical diagnosis. The purpose of this chapter is to introduce the main concepts underlying pattern recognition theory to those explorationists not familiar with the subject. For a survey of the basic techniques and their actual and potential applications to oil and gas exploration problems the reader is referred to de Figueiredo (1982). [A selected bibliography follows this chapter.]
By a “partern” we mean some form or structure present in a data set. In oil and gas exploration, examples of patterns of interest are the waveshape of a seismic trace in the neighborhood of a lithological interface; the pattern exhibited by an entire set of traces, in a variable area display of a common-depth stack, in the region of a fault; the configuration of equipotential contours on a map pertaining to an aeromagnetic survey of a basin; and the shapes of signatures of one or more well logs as a function of depth. These are only a few of the many types of patterns which occur
Figures & Tables
Exploration for oil and gas has witnessed dramatic changes in its nearly 120 year history. Initially, prospects were located by surface shows or seeps; random drilling was predominant. In the early twentieth century, the anticlinal theory became a dominant element in locating traps. Direct mapping of structures by using magnetic, gravity, and seismic data began in the mid 1920s.
In the last 10 years a true revolution has occurred in the use of seismic data in exploration. To a minor degree, the first use of seismic data to locate reefs and carbonate buildups took place in about 1950, but the main era of more quantitative stratigraphic trap detection began in the late 1960s when direct hydrocarbon detection by the so-called “bright spot” concept was first used in the Gulf Coast Cenozoic offshore. In a very short time since then, an increasingly sophisticated seismic mapping approach has swept through the exploration industry. It is now possible to map seismically subsurface stratigraphy, model stratigraphic analogs, and make comparisons of the recorded data with known analogs in order to “read the subsurface.”
We are now on the verge of an era of synergism in revolution where exploration techniques are being integrated with reservoir delineation and production engineering methods. Synergism, according to Webster, is “the joint action of agents…, which when taken together increase each other's effectiveness.” Recognizing this need as a key to success, several companies have begun to integrate the know-how from geology, petrophysics, and reservoir engineering in developing plays such as the Ozona-Sonora gas play in West Texas.
Predicting accurately all the variables in the subsurface requires the solution of an extremely complex equation. This is because there are so many parameters which cannot be scientifically measured adequately ahead of (or, for that matter, after) the drilling bit. Thus, our exploration efforts are designed to reduce the risk of being wrong in our solution of the “subsurface equation.”
Risk reduction in exploration can be greatly facilitated by bringing all the necessary technical expertise to bear on the problem. The exploration hexagon in Figure 1 illustrates the interrelation among six broad technologies that can be used to minimize the risk of drilling a dry hole. Frequently, there are insufficient data available to make use of all disciplines and often only one or two create a play.
A play is an exploration activity involving a geographically designated and geologically definable volume of rock in which one or more targets for hydrocarbons can be described (see Figure 2).