A model for earth science data structures designed to promote generalization of earth science data processing systems

During the past 15 years earth science survey data compilation has gradually changed from a manual process to one almost entirely computer automated. Computer methods were at first an optional means for improvement of the speed and efficiency of certain phases of the work. They eventually became mandatory as in-flight digital acquisition devices were introduced and rates of data acquisition increased to the point where manual methods became totally inadequate. As a consequence of the phased replacement of manual systems with digital ones, and the pressing need to avoid interruptions to the work, the resulting software is highly specialized. Software systems are incompatible between different organizations compiling the same kind of data and even between systems for different kinds of data within one organization. The result is wastage of manpower in maintaining, learning to use, and improving the great variety of existing systems, and wastage of man and computer time in converting data to a form compatible with the peculiarities of each particular system. Comparative analysis of highly specialized compilation systems form several different earth science disciplines reveals that processes do exist at the fundamental level which are generally applicable to a range of disciplines. Hence a generalized compilation system capable of eradicating the problems inherent in multiple specialized systems is at least conceptually feasible. The most serious obstacle to its realization however, is the diversity of data content and structure among the various disciplines. To surmount this obstacle, a system with a great degree of physical and logical independence of data from software will be necessary. Existing methods of achieving such independence within data base management systems are found to be largely inapplicable to earth science survey compilation, the principle reasons being the very large quantities of data involved and significant differences between the data retrieval requirements of compilation systems and data base management systems. It is found that, although the data contents and structures vary greatly between the various compilation systems examined, a general abstract model can be constructed which adequately represents all of them and which incorporates the means for achievement of data independence. The logical data structures of current data base management systems are based variously on relational calculus, set theory, and graph theory. The compilation data structure model is based on simple algebra with the addition of some components of vector algebra. It is further found that simple algebraic manipulation of the model expressions faith-fully simulates the actual data manipulation processes which are applied by the various compilation systems. Furthermore, unlike the logical models of the data base management systems, the logical structure of the compilation data model also represents the actual physical structure exhibited by the data in its most common, sequential form.