Question 10 of 10: Image Processing for Gravity and Magnetic Data — Colour Stretch and Enhancement

Question 10: Image Processing for Gravity and Magnetic Data — Colour Stretch and Enhancement (Choose 6 of 10 – equal value)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Setting the colour stretch (look-up table)

A gridded gravity or magnetic dataset is first converted to a colour image by mapping each grid value, through a LOOK-UP TABLE (LUT), to a colour on a chosen palette. The simplest choice, a LINEAR stretch across the data's full min–max range, is often a poor choice in practice because potential-field data are frequently dominated by a small number of high-amplitude anomalies, which compresses the vast majority of (geologically interesting, lower-amplitude) values into a narrow, visually indistinguishable band of colour. A histogram-equalized stretch instead assigns colour bins so that an EQUAL NUMBER of data points fall in each colour interval, which allocates more of the available colour range to the densely-populated, low-to-moderate-amplitude values where most of the subtle geological detail actually lies, at the cost of a non-linear (harder to read absolute values from) colour-to-value relationship. A percentile-clipped linear stretch (e.g. clipping the top/bottom 1–2% of values before applying a linear stretch to what remains) is a common compromise, keeping a readable linear scale while preventing a handful of extreme outlier values from washing out the rest of the map. Palette CHOICE also matters: a perceptually uniform, non-rainbow palette (e.g. a linear grey-to-colour ramp) avoids introducing FALSE visual edges at colour-band boundaries that a naive rainbow LUT can create, which an interpreter could otherwise mistake for a real geological contact.

Other enhancement methods

Sun-shaded (sun-angle/hill-shade) relief images illuminate the gridded surface, treated as a pseudo-topography with amplitude as "elevation," from a chosen low sun-angle azimuth and inclination, casting shading that dramatically enhances subtle, low-amplitude linear trends (faults, dyke margins, fabric) that would be invisible on a flat colour map, though a chosen illumination azimuth will emphasize features trending perpendicular to it while suppressing features PARALLEL to it, so multiple sun-angles (or a composite of several) are typically examined. Horizontal and vertical derivative grids, the analytic signal amplitude (which peaks directly over source edges regardless of magnetic inclination/declination), the tilt-angle transform of Question 7, and band-pass/wavelength FILTERING to isolate a target's expected anomaly wavelength range are all standard complementary displays, each emphasizing a different aspect (edges, depth, amplitude-independence, or a specific source-depth range) of the same underlying data.

Practical workflow

In practice an interpreter examines a dataset through SEVERAL of these displays side by side — a well-chosen colour stretch of the raw (or regional-residual-separated) anomaly, one or more sun-shaded views at different azimuths, and one or more derivative/analytic-signal/tilt products — because each enhancement can reveal features the others suppress, and a feature that appears consistently across multiple, independently-derived displays is far more likely to be a genuine geological signal than an artifact of any single processing choice.