Seal and reservoir risk evaluation using hierarchical clustering analysis with seismic attributes in Northwestern Australia

Assessing the presence and quality of reservoir rocks and their sealing capacity is crucial for various applications, including hydrocarbon, geothermal, and CO₂ sequestration projects. Typically, exploration geoscientists rely on seismic attributes and borehole logs into interpretation to integrate diverse data for estimating reservoirs and seals. However, for all seismic interpreters, the process is time-consuming.

In this study, we explore the application of Hierarchical Clustering Analysis (HCA), an unsupervised machine learning technique, to streamline the integration of multidisciplinary information. While HCA and similar techniques may occasionally misclassify critical data, we demonstrate how to enhance their accuracy by carefully selecting the number of clusters and their calibration with borehole data.

The novelty of our work is the innovative transformation of HCA clusters into a 3D lithology model, which can significantly facilitate the estimation of reservoir rock and seal-rock juxtaposition risk. Using the HCA clustering hierarchy, five clusters effectively discern the presence and quality of seal and reservoir rock in two different datasets. The classification, in combination with the fault probability, addresses the seal risk offshore the Northern Carnarvon Basin.