BME Analysis of Neural Network Residual Data from the Chernobyl Fallout: Bayesian and Non-Bayesian Approaches

  • G. Christakos
  • M. Serre
  • V. Demyanov
  • V. Timonin
  • M. Kanevski
  • E. Savelieva
  • S. Chernov
Conference paper
Part of the Quantitative Geology and Geostatistics book series (QGAG, volume 11)

Abstract

Radioactively contaminated territories after the Chernobyl accident are characterized by non-stationary trends and soft (uncertain) information about the average concentration of radionuclide in soil. Large-scale decision-oriented mapping in this situation involves using the Neural Network method to determine the general non-linear trends of the data, and the BME method to analyze the hard and soft residual information generated after the mean trend is removed from the data. In this work we explore different approaches to map the residual soft data, which include a Bayesian and a non-Bayesian framework. These approaches are illustrated by means of the Cs137 mapping case study in Briansk, Russia

Keywords

Covariance Radionuclide 

Copyright information

© Springer Science+Business Media Dordrecht 2001

Authors and Affiliations

  • G. Christakos
    • 1
  • M. Serre
    • 1
  • V. Demyanov
    • 2
  • V. Timonin
    • 2
  • M. Kanevski
    • 2
    • 3
  • E. Savelieva
    • 2
  • S. Chernov
    • 2
  1. 1.Center for the Advanced Study of the Environment -CASEUniversity of North CarolinaUSA
  2. 2.Institute of Nuclear Safety (IBRAE)Moscow
  3. 3.IDIAP Dalle Molle Institute of Perceptual Artificial In IntelligenceMartihnySwitzerland

Personalised recommendations