BSDWormer; an Open Source Implementation of a Poisson Wavelet Multiscale Analysis for Potential Fields

Thursday, 18 December 2014
Franklin G Horowitz, Cornell University, Ithaca, NY, United States and Oliver Gaede, Queensland University of Technology, Brisbane, QLD, Australia
Wavelet multiscale edge analysis of potential fields (a.k.a. “worms”) has been known since Moreau et al. (1997) and was independently derived by Hornby et al. (1999). The technique is useful for producing a scale-explicit overview of the structures beneath a gravity or magnetic survey, including establishing the location and estimating the attitude of surface features, as well as incorporating information about the geometric class (point, line, surface, volume, fractal) of the underlying sources — in a fashion much like traditional structural indices from Euler solutions albeit with better areal coverage. Hornby et al. (2002) show that worms form the locally highest concentration of horizontal edges of a given strike — which in conjunction with the results from Mallat and Zhong (1992) induces a (non-unique!) inversion where the worms are physically interpretable as lateral boundaries in a source distribution that produces a close approximation of the observed potential field. The technique has enjoyed widespread adoption and success in the Australian mineral exploration community — including “ground truth” via successfully drilling structures indicated by the worms.

Unfortunately, to our knowledge, all implementations of the code to calculate the worms/multiscale edges (including Horowitz’ original research code) are either part of commercial software packages, or have copyright restrictions that impede the use of the technique by the wider community.

The technique is completely described mathematically in Hornby et al. (1999) along with some later publications. This enables us to re-implement from scratch the code required to calculate and visualize the worms. We are freely releasing the results under an (open source) BSD two-clause software license. A git repository is available at <https://bitbucket.org/fghorow/BSDWormer.git>.

We will give an overview of the technique, show code snippets using the codebase, and present visualization results for example datasets (including the Surat basin of Australia, and the Lake Ontario region of North America).

We invite you to join us in creating and using the best worming software for potential fields in existence — as both gratis and libre software!