Artificial Neural Networks And Geostatistics For Environmental Mapping
Free (open access)
Kanevsky M., Arutyunyan R., Bolshov L., Demyanov V. , Savelieva E., Haas T. & Maignan M.
The report presents several approaches and models that were used for spatial estimations of radioactively contaminated territories after Chernobyl nuclear accident followed by large scale environmental consequences: well known traditional deterministic models (inverse distance weighted algorithms, multiquadric equations), recently developed geostatistical models (Moving Window Regression Residual Kriging/CoKriging), and artificial neural networks (ANN). The models used differ in amount of preliminary analysis, levels of complexity and assumptions about the data behaviour.