Evidential reasoning for geological mapping with multisource spatial data

Journal Title
Journal ISSN
Volume Title
Spatial data acquired from different sources may have different measurement scales, different resolutions, different reliabilities, partial coverages, etc. Evidential reasoning provides a heuristic scheme for handling multisource spatial data. Under the evidential framework, different data sources are treated as different pieces of evidence. The evidential support of a data source over a set of hypotheses can be evaluated based on the measurements from the data source. Then the combined evidential support from different sources can be calculated using Dempster's Rule. With Dempster's Rule one can also solve the problem of incompatibility of multisource data. The major difficulty in applying evidential reasoning is the evaluation of evidential support or mass function from different pieces of evidence. In this research, some methods on the evaluation of evidential support from different data sources in the application of multisource spatial data analysis are proposed. Some practical considerations of implementing evidential reasoning are also discussed. The methods presented here have been evaluated in application of geological unit classification. The study site is located in Hall Lake area of Melville Peninsula, Northwest Territories, Canada. The data set used in this research contains: Landsat TM imagery, ERS SAR, radiometric, gravity, aeromagnetic data. The highest overall geological classification accuracy obtained by using the evidential reasoning algorithm is 95.13%.
Bibliography: p. 81-85.
Shi, G. (1994). Evidential reasoning for geological mapping with multisource spatial data (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/19658