Automated multi‐well stratigraphic correlation and model building using relative geologic time

Stratigraphic correlation of geophysical well logs is one the most important—and time-consuming—tasks that applied geoscientists perform on a daily basis. Using dynamic time warping (DTW) algorithm, automated two wells fairly simple task; DTW can also be used to correlate large number along single path. However, errors accumulate path and loops cannot closed. To create three-dimensionally consistent framework, we use Python implementation (SEG Technical Program Expanded Abstracts, 2014, pp. 618–622) approach, which based idea stretching squeezing all into chronostratigraphic diagram has relative geologic (RGT) its y-axis. The depth shifts needed for RGT transformation are computed by translating outputs pairwise correlations least-squares optimization problem solved through conjugate gradient method. resulting provides an overview overall stratigraphy variability; framework significantly different from what could get correlating lithologic similarity between logs. geologically intuitive well-log cross sections, multi-scale blocking method relies continuous wavelet transform identify stratigraphic units certain scale in then propagate these boundaries other wells. We demonstrate usefulness this approach data set with close 700 Permian Basin, West Texas. Linear channel bodies deepwater Spraberry Formation easily detected clearly highlighted maps sections. methodology robust enough mapping subtle details, previously considered feasible only manual interpretation. More importantly, it quickly build three-dimensional models segments sedimentary basins where log available.