Application of probability theory to integrated geological mapping from remotely sensed data of the Precambrian Shield (Manitoba, Canada)

• Main Author: Rao, Govindaraju Suresh Kumar
• Format: Doctoral or Postdoctoral Thesis
• Language: English
• Published: 1995
• Subjects: Canada; Churchill
• Online Access: http://hdl.handle.net/1993/7358

Investigation of earth resource potential requires the integration of various information obtained from different survey techniques. Such information is commonly represented as two-dimensional digital map layers. The ability to selectively combine, by spatial data integration processes, diverse data types, is increasingly becoming a mainstay of geological exploration programs. Such automated mapping of a terrain usually requires some previous knowledge of the terrain. This knowledge is used to constrain the mapping algorithm or to formulate a set of rules that govern the integration process. However, in cases where a completely unknown terrain is being investigated, the lack of prior knowledge can be a serious obstacle. The present research addresses the fundamental problem of integrating remotely sensed satellite and geophysical data, in the absence of an initial data base. Three types of integration techniques, based on probability theory, are presented: algebraic probability, spatial index, and Bayesian probability. The target proposition is the mapping of the boundary zone between the Proterozoic Churchill Province and Archean Superior Province. In an hitherto unmapped study area, the algebraic probability method demonstrates the assignment of probabilities to input data sets, based on the target proposition and visual interpretation. The input data are integrated by an algebraic additive process. The probability assignment and the algebraic additive process are validated by application to another area of known geology. Once validated, the results of the algebraic probability method are treated as an a priori indicator for the next two methods. The spatial index method is developed to quantify the spatial correlation in the input data sets with the a priori information. This correlation is then converted into a probability measure, and the integration process is carried out. The results show that the probability method is a useful technique for data integration. Based on the results of this study, a revised ...