Automatic lithology modelling of coal beds using the joint interpretation of principal component analysis (PCA) and continuous wavelet transform (CWT).

Identification of thin interbedded non-coal bands and coal seams with varying carbon contents within a coal seam is of paramount interest in coal exploration due to its banded nature. The manual interpretation and conventional modelling based on Fourier/Walsh transform techniques fail to derive such information accurately from geophysical well log data due to its non-stationary nature. The present study proposes a combined principal component analysis (PCA) and continuous wavelet transform (CWT) algorithm for automatic lithological modelling of geophysical well log data. In the first step of well log lithology modelling, a median filter is applied on well log data to preserve the thinner beds and other valuable geological signatures of coal seams. In the second step, the filtered log data is subjected to PCA, and the variance level of PC scores is determined to study the physical relationship of input parameters. The third step is to apply CWT on the selected PC scores and determine lithological discontinuities from the modulus maxima lines drawn on the wavelet scalogram. For filling the lithology skeleton with the proper interpretation, a database is also created by correlating the selected PC score values with input parameters. We have applied the proposed algorithm to gamma ray, density, and resistivity logs of two boreholes located in the Bisrampur and Jharia coalfields of eastern India. The results of the proposed PCA-CWT based modelling match well with core data and manual interpretations of the boreholes. At a few depth ranges, the proposed algorithm also reveals some additional lithological discontinuities that were not mapped in the core data. The study further conveys that PCA-CWT-based lithological modelling of geophysical logs is helpful to pace up the exploration work in coal blocks with poor core recovery. [ABSTRACT FROM AUTHOR]