Short answer

Geophysical data — airborne magnetics, gamma-ray radiometrics, gravity, electromagnetics, seismic — enters GIS mostly as gridded raster layers and is used to map structure and lithology beneath cover where outcrop is missing. The technical work is threefold: get the grids into GIS with the right CRS and data type, apply or verify the standard geophysical transforms (reduction to the pole, derivatives, analytic signal), and display them with stretches and shading that make subtle gradients visible without distorting the data.

Done well, a magnetic or radiometric layer becomes a structural interpretation tool that extends mapped faults and contacts hundreds of metres or kilometres into covered terrain. Done carelessly — wrong projection, byte-scaled values, a rainbow stretch over an unreduced grid — it produces confident-looking nonsense.

The data you will actually receive

Magnetics

Total Magnetic Intensity (TMI) measures the strength of the Earth's field plus the contribution of magnetic minerals (mostly magnetite) in the rock. Surveys are flown on lines (e.g. 200 m line spacing, 50 m terrain clearance) and gridded to a cell size roughly a quarter to a fifth of line spacing. The raw deliverable is usually a Geosoft .grd, an ER Mapper .ers, NetCDF, or GeoTIFF. Magnetic units are nanotesla (nT).

Radiometrics (gamma-ray spectrometry)

Measures natural gamma emission from Potassium (% K), Uranium (ppm eU) and Thorium (ppm eTh) in roughly the top 30–45 cm of ground. The classic product is a ternary (RGB) image assigning K to red, Th to green, U to blue, which maps near-surface lithology and weathering.

Gravity

Bouguer and free-air anomalies in milligal (mGal), sensitive to density contrasts — useful for basin architecture, intrusions, and deep structure.

Seismic

Reflection sections and interpreted horizons; in GIS these usually appear as line geometries (shot lines), gridded surfaces, and 2D section images rather than as the volume itself.

Getting grids into GIS correctly

The first discipline is data type. Geophysical grids are continuous floating-point values. If you load them as 8-bit and let the software scale TMI from −300 to +1200 nT into 0–255, you destroy the dynamic range you came for. Keep them as Float32.

The second is CRS. Survey grids are often delivered in a projected datum specific to the survey (a local UTM zone, e.g. EPSG:32750 for UTM Zone 50S). Confirm it before overlaying anything:

gdalinfo magnetics_tmi.ers
gdal_translate -of GTiff -ot Float32 magnetics_tmi.ers tmi.tif
gdalwarp -t_srs EPSG:32750 tmi.tif tmi_utm.tif

Use gdalwarp only to reproject if a true CRS mismatch exists; needless resampling of a geophysical grid blurs anomaly edges. Set NoData explicitly (-a_nodata) so survey-margin padding does not contaminate stretches and statistics.

Reduction to the pole — the transform you must not skip

Magnetic anomalies are dipolar and their shape depends on the inclination and declination of the Earth's field at the survey location. At mid and low latitudes a simple body produces an asymmetric anomaly that is offset from the source — a fault that sits at X appears shifted from X. Reduction to the pole (RTP) mathematically transforms the grid as if it were measured at the magnetic pole (inclination 90°), so anomalies sit symmetrically over their sources and line up with the geology.

RTP is typically applied by the geophysical contractor in the frequency domain. Your job in GIS is to verify which product you have. If a grid is labelled "TMI" with no reduction, anomaly positions are not reliable for structural mapping at low latitudes; ask for the RTP grid. At very low magnetic latitudes RTP becomes unstable and reduction to the equator (RTE) or a pseudo-gravity transform may be used instead — note this in metadata.

Derivative products that reveal structure

From the (reduced) magnetic grid, several derived layers sharpen edges and faults:

  • First Vertical Derivative (1VD) enhances shallow, high-frequency features and sharpens contacts.
  • Tilt Derivative (TDR) normalises amplitude, bringing weak and strong anomalies onto a comparable scale; its zero contour approximates source edges.
  • Analytic Signal (AS) produces peaks centred over source edges and is independent of magnetisation direction, so it is robust where RTP is uncertain.

These usually arrive as separate grids. Load them stacked, toggle them, and trace structural lineaments across them — a fault confirmed on 1VD, TDR, and AS is far more credible than one seen on a single product.

Display: stretches and shading

A geophysical grid with a linear stretch over its full range looks flat because a few extreme values dominate. Use a percentile stretch (commonly 2–98%) or histogram equalisation so the bulk of the data uses the bulk of the colour range. In QGIS, set the raster's Min/Max Value Settings to Cumulative count cut at 2–98%.

Pair colour with sun shading. Generate a shaded relief of the grid itself (treating nT or mGal as "elevation") to expose linear fabric:

gdaldem hillshade tmi_rtp.tif tmi_shade.tif -az 45 -alt 45 -z 2

Then drape a semi-transparent colour ramp over the shade. Vary azimuth (e.g. 45° and 135°) to avoid biasing toward structures perpendicular to a single light direction — structures parallel to the illumination disappear, which is a classic mapping trap. For radiometrics, build the K/Th/U ternary as a 3-band RGB composite rather than three separate single-band layers.

Worked integration: extending a fault under cover

You have mapped a fault from outcrop that disappears beneath alluvium. Load the RTP magnetic grid, its tilt derivative, and the radiometric ternary. The fault's strike continues as a linear truncation in the TDR and as a lithological boundary in the K channel of the radiometrics. Digitise the lineament as an interpreted line feature, attribute it with confidence (e.g. high/moderate/inferred) and evidence (e.g. "TDR truncation + K boundary"), and keep it in a layer distinct from the field-observed structures. This preserves the difference between what was seen on the ground and what was inferred from geophysics — the single most important honesty rule in this work.

Common pitfalls and why they happen

  • Byte-scaling a float grid. The dynamic range collapses and subtle anomalies vanish, because 256 levels cannot represent the full nT range. Keep Float32.
  • Interpreting unreduced TMI for structure at low latitude. Anomalies are offset from sources, so traced faults land in the wrong place. Use RTP/RTE products.
  • Single-azimuth hillshade. Structures parallel to the light vanish, biasing the structural map. Use multiple azimuths.
  • Rainbow ramps. A spectral rainbow creates false edges at colour-band boundaries and is not perceptually uniform. Prefer a perceptual ramp.
  • Mixing survey grids of different line spacing without noting resolution. A 400 m-line regional grid cannot resolve features a 100 m-line detailed grid shows; combining them without flagging the resolution misleads.

QA and validation

Confirm the grid's CRS and cell size against the survey report; check that NoData masks the survey margin so statistics and stretches are not skewed; verify you are using reduced (RTP/RTE) magnetics for positional interpretation; and cross-check at least one strong anomaly against known geology (a mapped magnetite-bearing intrusion should be a magnetic high). Record in metadata which transforms were applied and by whom.

Bathyl perspective

We treat geophysics layers as interpretation aids with explicit provenance, not as basemaps. Every derived grid carries a note of its transform chain, and every lineament we digitise from geophysics is tagged as inferred and separated from field observations. The goal is a structural picture a reviewer can audit line by line.

Related reading

Sources