XLOGS: An Expert System for Environmental Interpretation of Wireline Logs

Abstract

Wireline logs provide information on vertical sequences that can be useful for preliminary interpretation of depositional environments. Although the specific causes of any log response are related to many factors, including lithology, density, and pore fluid, geologists often can gain a general impression of vertical heterogeneity, cyclicity, and gradation without performing a rigorous petrophysical analysis of log responses. The authors have attempted to model these heuristics of well logs in a computer program - XLOGS - that emphasizes pattern recognition and fuzzy logic. XLOGS begins with segmentation of a given digitized log in interactive graphical mode. Next, the log shape is characterized by comparison with a set of type motifs, such as straight, serrated, funnel, and bell shapes. Shapes are described in terms of numerical attributes such as amplitude and slope. By using fuzzy values to express membership, any shape can be described. Related subenvironments can be recognized from characteristic vertical sequences, such as the upward-coarsening prograding delta lobe that produces a funnel-shaped mouth-bar segment above a straight to serrated prodelta segment. Variability in the depth scale is accommodated by iterative segmentation; trials with coarser or finer segmentation can be evaluated as alternative interpretations for distinguishing sequence and parasequence boundaries.