Nafta, Vol. 62 No. 11-12, 2011.
Izvorni znanstveni članak
Identification of palaeoenvironments using clustering methods and indicator kriging, case study from Late Miocene sandstones, the Sava depression
Janina Horváth
; Sveučilište u Segedu, Zavod za geologiju i paleontologiju, Sveučilište u Segedu, Seged, Mađarska
Kristina Novak Zelenika
; INA-Industrija nafte d.d., Sektor za geologiju i upravljanje ležištima, Zagreb, Hrvatska
Sažetak
The clustering and Indicator Kriging methods are standard exploration tools in some kind of geological problems and here is presented a case study for the identification of depositional environments using two kinds of clustering processes and the Indicator Kriging. The clustering is based on Kohonen's neural network and hierarchical agglomerative clustering. Through the resulting clusters, the geometry of depositional sub-environment can be visualised by spatial mapping processes and it can be interpreted by the probabilities of clusters through the Indicator Kriging. In this manner the clustering tools can be defined as a spatial pattern recognition method. The research area in the demonstrated study is located in the Sava Depression, Croatia. The field sediments belong to Neogene and Quaternary, and the analysed are Upper Pannonian when sedimentation was accompanied by influence of pre-Neogene basement palaeorelief and the accumulation by turbidity system.
Ključne riječi
clustering; palaeoenvironment; Kohonen's neural network; hierarchical agglomerative clustering; Indicator Kriging; Sava Depression; Croatia
Hrčak ID:
75937
URI
Datum izdavanja:
30.12.2011.
Posjeta: 2.064 *