QuickDepth Module

An optional extension to ModelVision, QuickDepth applies AI principles to the estimation of depth, magnetic properties and geological style from magnetic data.  The AI speeds up the interpretation process but leaves the interpreter in control of the geological interpretation.  It uses both the line data for the highest possible depth precision and the associated grids to gather intelligence on the shape characteristics of each anomaly. 


QuickDepth is a new approach to calculating the depth to a magnetic source for isolated magnetic anomalies using a variety of interpretation techniques that do not require inversion.  The method operates by dragging the mouse across the anomaly you want to interpret to produce immediate feedback on the depth to the source of the anomaly. 

The input data is a high quality total magnetic intensity grid and the magnetic line data including the levelled magnetic field, sensor elevation and ground surface elevation or radar altimeter channels. 

The magnetic tensor and a range of other magnetic field measures are computed from the total magnetic field grid.  This data provides key geological information that is used to assist in the depth interpretation and improve the quality of the estimates compared with conventional automated processes.  Importantly, the depth is estimated from the data along the flight line to preserve the highest quality gradient information to produce the best possible depth estimates.

The tensor of the total magnetic field is used to derive geological characteristics such as strike direction, body type, centre of magnetization and depth to the top of the magnetic unit. This information is used to constrain and improve the precision of the other geophysical methods which include:
  • Euler 2D Method
  • Peters' Length Method
  • Tilt Depth Method
  • Euler 3D Method
  • Werner Deconvolution Method
We use the peak of the normalised source strength (NSS) from Clark (2014) to define the horizontal (X, Y) location of the centre of magnetization which simplifies the calculations of depth for the Euler 2D, Werner and Tilt Depth methods. The strike direction of the anomaly (Pederson & Rasmussen, 1990) is used to correct the depth estimates for acute angle flight lines for the Tensor, Peters' Length, Werner Deconvolution and Tilt Depth methods.

Experimental Survey ModelVision provides tools that allow you to create experimental surveys over any geological model. This example shows a geological model composed of all the body shapes that can be sup-ported by QuickDepth with lines that are at 45 degrees and elongate bodies that are perpen-dicular to the line direction. In this example, the true depth is within 5 to 15% of the true depth when the correct body type is selected. In cases where there is no appropriate model from the original AI training such as the thick pipe, you can expect an overestimation of the depth. The strike direction is estimated within two degrees of the true direction.
The tensor analysis also provides a dimensionality index which automatically differentiates between pipe-link magnetic sources and linear magnetic formations or dykes.  This allows for different depth correction techniques to be applied according to the geology. Some methods such as Euler 2D analysis are very sensitive to an incorrect choice of the geological magnetic source type.  The tensor also provides some information about the width of the magnetic source and classifies it as thin, intermediate or thick.

Importantly, the user is in control of the geology. While the underlying code does its best to determine the characteristics of the geological target, the automated selection can be overridden when appropriate.

The solutions can be exported as a point dataset to a Geosoft GDB database or CSV file for use in other applications. The data can be imported into ModelVision using the Convert Point to Body option for display of bodies in cross-section and map views.

For further details on QuickDepth refer to the QuickDepth Technical Sheet on our website.

For further reading on the principles of QuickDepth refer to the AEGC 2019 paper by David Pratt et al, titled, “An AI approach to using magnetic gradient tensor analysis for quick depth and property estimation” .
Comments