Iterative Non-hierarchical Cluster Analysis (INCA): a novel approach to analysing downhole logging and petrophysical data

Iterative Non-Hierarchical Cluster Analysis (INCA) is a multivariate statistical approach used to analyse data, suited to large datasets where a defined number of clusters can be hypothesised. Although the fundamentals of this methododology are well-known, its use for the analysis of downhole logging and petrophysical measurements is a novel approach for quantitatively interpreting these datasets. Cluster analysis works by starting with k random clusters then grouping the set of data such that values within a cluster are more similar to each other than to those in the other clusters i.e. to minimise variability within a cluster and maximise variability between clusters. INCA provides a valuable tool for understanding downhole logging and petrophysical datasets by interpreting the weightings attached to each cluster in terms of petrophysical characteristics. The INCA program has been successfully used on datasets from varied geological environments to characterise formations based on their petrophsyical properties alone. Here, we discuss details of the specific algorithms chosen, size constraints and departures from standard commercially available packages. We present examples of results that use a variety of different petrophysical parameters; from spectral gamma ray logs used to independently assess lithology, to resisitivity and sonic logs used to identify the position of major boundaries. Examples of the methodology applied to subduction zone accretionary prism sediments (IODP Expedition 314), oceanic crustal rocks (e.g. ODP Hole 735B) and from continental shelf siliciclastic sediments (IODP Expedition 313) are illustrated.