Fiorini, Sandro R.; Abel, Mara; Scherer, Claiton M.
The petroleum exploration workflow usually involves a multitude of data intensive tasks. For instance, the reconstruction of the stratigraphic history of a given area of interest might require description and interpretation core, logs and seismic data; a resource-consuming task. A huge research effort has been invested in search of intelligent software system that could to automate or semi-automate some these tasks. In answer to that, we developed the Intelistrata system, a complete software framework capable of recognizing stacking pattern and stratigraphic surfaces in gamma ray well logs. It takes a digitalized a gamma ray log as input and recognizes the presence of sequences and parasequences along the log. A sequence is a series of genetically related layers whose limits are defined as a response to a relative sea level fall. Parasequences, on the other hand, indicate small cycles of sea level variation within sequences. From a technical standpoint, the Intelistrata interpretation engine innovates by integrating signal processing and semantic reasoning techniques to produce a coherent interpretation of the data. The signal processing module employs wavelet and movable means algorithms to segment the gamma ray signal, based on an expected feature length informed by the geologist. The extracted features are then fed to a semantic reasoning engine able to recognize and filter which of the features match the conceptual definitions of sequence and parasequence. The conceptual definitions consist of visual knowledge models elicited from an expert in sequence stratigraphy; i.e., a domain ontology. The visual knowledge model represents the stratigraphic features that will be recognized in the gamma ray logs; it tries to incorporate the knowledge the geologist applies when individualizing the geological limits. The generated result is a semantically rich description of the well log in terms of stratigraphic features. The Intelistrata system has been validated with a total of 3.2 km of gamma ray logs from the USA (Book Cliffs, Utah) and Brazil (Caçapava do Sul, RS), comparing generated interpretation with the expert interpretation. In brief, the system demonstrated a mean hit ration to 75% per well for sequences, while generating only about 10% of false positives per well. Intelistrata is now being integrated to the Strataledge® system for sedimentary facies description of cores and outcrops.
AAPG Search and Discovery Article #90163©2013AAPG 2013 Annual Convention and Exhibition, Pittsburgh, Pennsylvania, May 19-22, 2013