Specifications

University of Pretoria etd – Combrinck, M (2006)

Chapter 4

4 AUTOMATED INTEGRATED ANALYSIS OF TDEM DATA

“Nature is not a competition. It doesn’t really matter, when you go out, if you don’t identify anything. What

matters is the feeling in the heart.” (Richard Adams, B. 1920, British Writer)

"Exploring the unknown defines the essence of humanity." (The author)

4.1 An integrated analysis algorithm

A software supported approach was developed for the:

• More effective visualization of the field data for rapid anomaly detection

• Automatic calculation and visualization of decay curve characteristics

• The use of the S-layer transform for the creation of conductivity-thickness cross

section for the outline of finite conductors.

An automation of some strategies as outlined in chapter 3 was implemented in a C++

algorithm. Data analysis is performed on a time basis (i.e. decay curves) and not on a spatial

basis (i.e. profile plots). Furthermore, the S-layer transform part of the algorithm assumes

centre of the loop (sounding) data only. The algorithm does an automatic analysis,

discussed below, of each measured decay curve. The only user input required is to specify

system parameters and regression analysis limits, although default values for these are

given.

4.2 Visualization of the field data for rapid anomaly detection

Even though the ultimate gaol of TEM interpretation is to end up with a conductivity-

depth section or conductor location and properties, it is always good to have the original

data readily available as well. This often helps to clarify ambiguities in interpretations or

models, e.g. the presence of a small, three-dimensional feature visible on raw data profiles

would imply possible failure of layered interpretations in that region. Noise is also

identified with more confidence on original data than after processing or transforms. TEM

data can, however, be quite challenging to present in a complete manner without losing

detail. Two factors contributing to this are the logarithmic span of data (with the small,