Napredna pretraga

Pregled bibliografske jedinice broj: 1001973

Lithology prediction in the subsurface using artificial neural networks on well and seismic data – a stochastic approach


Cvetković, Marko; Kamenski, Ana; Kolenković Močilac, Iva; Rukavina, David; Saftić, Bruno
Lithology prediction in the subsurface using artificial neural networks on well and seismic data – a stochastic approach // Abstracts book of the GEOMATES 2019 / Gabor Hatvani, Istvan ; Tanos, Peter ; Fedor, Ferenc (ur.).
Pecs, Hungary: Pécs Regional Committee of the Hungarian Academy of Sciences, 2019. str. 28-28 (predavanje, međunarodna recenzija, sažetak, znanstveni)


Naslov
Lithology prediction in the subsurface using artificial neural networks on well and seismic data – a stochastic approach

Autori
Cvetković, Marko ; Kamenski, Ana ; Kolenković Močilac, Iva ; Rukavina, David ; Saftić, Bruno

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
Abstracts book of the GEOMATES 2019 / Gabor Hatvani, Istvan ; Tanos, Peter ; Fedor, Ferenc - Pecs, Hungary : Pécs Regional Committee of the Hungarian Academy of Sciences, 2019, 28-28

ISBN
978-963-7068-11-9

Skup
GEOMATES 2019

Mjesto i datum
Pečuh, Mađarska, 16-18.05.2019

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Artificial neural networks, stochastics, lithology, subsurface

Sažetak
Analysis of lithology and lithology related variables in the subsurface is a key component in exploration of subsurface. The conventional way is to use different mapping algorithms to determine the properties in the inter well area based solely on well data or using seismic explorations (attribute analysis ; Radovich & Oliveros, 1998) in order to reduce uncertainty. Artificial Neural networks are also used for this purpose but more as a deterministic approach than a stochastic one (Brcković et al., 2017). For this purpose, a small volume of subsurface in the SW part of Pannonian Basin, representing an old small oil field which is covered by 3D seismic and several wells was selected. The artificial neural networks were first trained on a seismic attribute set belonging to the well traces and afterword the prediction was performed in the inter-well volume using 100 trained networks. The final result was obtained by P90 values of the categorical values that represent different lithologies. By this way the uncertainty of the lithology prediction in the inter well area has been significantly reduced, especially in this case where to few well point data were available to provide a variogram model for conventional deterministic and stochastic approaches.

Izvorni jezik
Engleski

Znanstvena područja
Matematika, Geologija



POVEZANOST RADA


Ustanove
Rudarsko-geološko-naftni fakultet, Zagreb