The Data Mosaic Project: Multi-scale spatial domaining of drill hole geochemistry data

[摘要] Multiscale edge detection methods using the continuous wavelet transform (cwt) can be used to detect local changes in geochemical data from drill hole samples. However, the products of the cwt can be difficult for the non-expert to interpret. The tessellation method was a first attempt to improve on the cwt product by providing a rectangularised transformation of the cwt scale-space plot. In this report a new method is proposed, called “Data Mosaic”, which improves on the tessellation method by providing a rectangular transform of the cwt that is faster to compute and is free from artefacts found in the tessellation. The new method also differs significantly from the tessellation in that the x-axis of the plot is the absolute maximum wavelet coefficient of the boundary, rather than the wavelet scale. This x-axis parameter is considered more geologically useful as it represents the amplitude of the signal changes across scales (i.e. the strength of the domain boundaries); while wavelet scale represents the signal wavelength across scales (i.e. the width of the domain on the depth axis). An additional plotting technique is provided which allows the user to visualise compositional variability within each domain using gradient fill.

[发布日期] 2017-03-21 [发布机构] CSIRO

[学科分类] 地球科学(综合)