THE UTILIZATION OF MACHINE LEARNING IN THE CHARACTERIZATION OF CLINOFORMS

The application of machine learning has proven to be an effective and efficient tool in seismic interpretation and analysis. Seismic interpreters have capitalized on machine learning methods to improve studies that examine sequence stratigraphy, characterize reservoirs, and distinguish varying geological features. Few studies have applied machine learning techniques to explore the genetic make-up of clinoform structures; and in particular, small-scale details typically hidden in the seismic data which can provide additional insight into the depositional professes. This study involves the development of an adaptable workflow utilizing seismic attributes, geological knowledge, and machine learning to better understand the mud-prone clinoforms in the Northern Taranaki Basin offshore New Zealand. A machine learning algorithm called Principal Component Analysis (PCA) was used to help provide accuracy and reliability on the selection of seismic attributes. The PCA’s output provided a selection of useful attributes to be further analyzed in a multi-dimensional space using a technique known as self-organizing maps. These results enhanced details in the seismic data, which in turn allowed for additional characterization of the clinoforms to be more easily identified by the interpreter.