A drill hole query algorithm for extracting lithostratigraphic contacts in support of 3D geologic modelling in crystalline basement

Abstract

The identification and extraction of lithostratigraphic contacts in crystalline basement for constraining 3D geologic models is commonly hampered by the sparseness of diagnostic lithostratigraphic features and the limited availability of geophysical well log data. This paper presents a query algorithm that, instead of using geophysical well log measurements, extracts lithostratigraphic contacts by exploiting diagnostic patterns of lithology-encoded intervals, recurrent in adjacent drill holes. The query algorithm allows defining gaps in the pattern to search across unconformable, intrusive and tectonic contacts and allows combining multiple search patterns in a single query to account for lateral lithofacies variations. The performance of the query algorithm has been tested in the Precambrian Flin Flon greenstone belt (Canada) by evaluating the agreement between queried and logged lithostratigraphic contacts in 52 lithostratigraphic reference drill holes. Results show that the automated extraction of the unconformable and partly tectonized contact between metavolcanic rocks and its metasedimentary cover was relatively unambiguous and matched all the contacts previously established by visual inspection of drill core. The 100% match was nevertheless paired with 23% false positives due to mafic and felsic sills emplaced in sandstone and conglomerate, which overlap in composition and thickness with extrusive volcanic rocks. The automated extraction of the contact between a mine horizon, defined by laterally complex volcanic and volcaniclastic lithofacies variations and overlying basalt flows, matched the visually logged contacts for 83% with 27% false positives. The query algorithm supplements geological interpretation when patterns in drilled lithostratigraphic successions, suspected to be diagnostic for lithostratigraphic contacts, need to be extracted from large drill hole datasets in a systematic and time-efficient manner. The application of the query algorithm is particularly relevant for 3D modelling projects in Precambrian shield areas where unique lithostratigraphic markers are difficult to establish or are completely lacking.