Short answer

ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer, on NASA's Terra satellite) is useful for mineral mapping because it samples the shortwave infrared (SWIR) with six bands between ~1.6 and 2.43 µm and the thermal infrared (TIR) with five bands — exactly the wavelengths where hydroxyl-, carbonate-, sulfate-, and silica-bearing minerals show diagnostic absorptions. Broadband sensors like Landsat have only one or two coarse SWIR bands, so they cannot separate, say, kaolinite from muscovite. With ASTER you compute band ratios and indices (or spectral-matching classifiers) to highlight clay, iron-oxide, carbonate, and silica alteration that often haloes hydrothermal mineral systems.

Two hard constraints frame everything below: ASTER's SWIR detectors failed in April 2008, so SWIR-based mineral work needs pre-2008 archive scenes; and ratio products map mineral groups, not confirmed geology — they are evidence layers that must be ground-truthed.

Why ASTER's bands matter spectrally

Alteration minerals are identified by the position and shape of absorption features:

  • Al-OH (clays/micas — kaolinite, muscovite/sericite, alunite, illite) absorb near 2.20 µm → ASTER band 6 (2.185–2.225 µm).
  • Mg-OH / carbonate (chlorite, epidote, calcite, dolomite) absorb near 2.30–2.35 µm → ASTER bands 8 and 9.
  • Ferric iron (Fe³⁺) has a charge-transfer feature in the visible/NIR → ASTER bands 1–3.
  • Silica (quartz) has a Reststrahlen emissivity feature in the TIR near 8.6–9.3 µm → ASTER bands 10–14.

ASTER's three subsystems: VNIR bands 1–3 at 15 m, SWIR bands 4–9 at 30 m, TIR bands 10–14 at 90 m. The 30 m SWIR pixel is the practical limit on detail for alteration mapping.

Diagnostic band ratios

Ratios suppress topographic shading (illumination cancels in the numerator/denominator) and emphasise the absorption contrast. Widely cited ASTER ratios (formulas differ slightly between authors — treat as starting points, not gospel):

  • Al-OH / phyllic-argillic clay: (B4 + B6) / B5 or simply B5 / B6 — highlights sericite/kaolinite/alunite.
  • Mg-OH / carbonate (propylitic): (B7 + B9) / B8 — chlorite, epidote, carbonate.
  • Iron oxide (gossan): B2 / B1 — ferric iron, useful for gossans and limonite.
  • Ferrous iron: B5 / B3 + B1 / B2.
  • Ferric oxide vs hydroxyl (Crósta-style): principal component analysis on selected VNIR/SWIR bands, an alternative to fixed ratios.
  • TIR silica index: combinations of bands 11–14 (e.g. B13 / B14 or (B11 + B13)/(B12)) to flag quartz/silicification.

The Sabins (1999) and Rowan/Mars USGS work are standard references for these ratios; pick a published set, cite it, and keep the exact formula in your metadata so the product is reproducible.

A processing workflow

  1. Acquire a pre-2008 SWIR scene. The on-demand AST_07XT product gives surface reflectance with the crosstalk correction already applied; AST_05 gives TIR emissivity. Otherwise start from L1T radiance.
  2. Atmospheric / crosstalk correction. ASTER SWIR suffers a known crosstalk artefact (band 4 light leaking into 5–9); use crosstalk-corrected products or apply the JSS correction. Then convert to surface reflectance (FLAASH, ATCOR, or the AST_07XT product).
  3. Geometric and CRS hygiene. Confirm the scene's UTM zone, and if you mosaic or stack with a DEM, reproject consistently: gdalwarp -t_srs EPSG:32643 -r bilinear. Resample SWIR/TIR to a common grid only when necessary, with awareness of the native resolutions.
  4. Masking. Remove cloud, cloud shadow, water, and vegetation before ratios — vegetation has its own Al-OH-like SWIR features and will produce false clay anomalies. An NDVI threshold (e.g. NDVI > 0.3 masked out) and a shadow/illumination mask are minimum.
  5. Compute ratios with the raster calculator: B5 / B6 in QGIS Raster Calculator or gdal_calc.py -A b5.tif -B b6.tif --calc="A/B" --outfile=clay.tif.
  6. Spectral matching (optional, stronger). Resample USGS spectral library signatures to ASTER bands and run Spectral Angle Mapper (SAM) or matched filtering for per-mineral classification rather than group ratios.
  7. Stretch and threshold the ratio to highlight anomalies, then overlay on terrain/geology.

Worked example: argillic alteration over a porphyry target

Over an arid, sparsely vegetated belt: acquire an AST_07XT scene, mask NDVI > 0.25 and deep-shadow pixels, compute the Al-OH clay ratio (B4 + B6)/B5 and the iron-oxide ratio B2/B1. Display clay in cyan and iron oxide in red over a hillshade. A concentric pattern — central silicic/argillic core (high clay ratio) ringed by a propylitic (B7+B9)/B8 Mg-OH halo — is the classic remote signature of a porphyry/epithermal system. Treat it as a drill-target prioritisation layer, then confirm with field spectrometry (e.g. a portable ASD/TerraSpec) and assays before any geological claim.

Limitations and pitfalls (with reasons)

  • No SWIR after April 2008. The SWIR detector cooler failed; modern scenes are VNIR/TIR only, so clay/carbonate ratios need archive data. This is the single biggest planning constraint.
  • 30 m mixed pixels. A 30 m SWIR pixel averages multiple surfaces, diluting weak alteration. ASTER finds districts and haloes, not outcrop-scale detail.
  • Vegetation false positives. Green and dry vegetation mimic Al-OH absorptions; unmasked vegetation creates phantom clay anomalies. Mask aggressively.
  • Crosstalk left uncorrected. Band-4 leakage biases SWIR ratios; using uncorrected L1 data produces systematic errors. Use AST_07XT or apply the correction.
  • Atmosphere and regolith. Iron-oxide ratios respond to surface weathering/regolith that may be unrelated to bedrock alteration; cross-date comparisons need matched sun angle, season, and moisture.
  • Reading ratios as lithology. A high clay ratio means an Al-OH mineral group is present, not a confirmed rock type or ore. Always validate.

QA / validation

Validate the spectral product against ground truth: field-spectrometer measurements, XRD/assay on samples, and known mapped alteration. Check at least a handful of anomaly pixels against field sites and compute simple agreement (true/false positives). Confirm the mask removed vegetation and shadow by overlaying the ratio on the masked NDVI. Keep the exact ratio formula, the source product (AST_07XT vs L1T), the crosstalk-correction status, the scene date, and the CRS in the layer metadata so the result is reproducible and auditable.

Bathyl perspective

We treat ASTER ratios as prioritisation evidence, not conclusions: they widen the search, focus field time, and must be labelled with their formula, scene date, and confidence. The strongest deliverable pairs the spectral anomaly layer with the ground-truth that confirms or rejects it, so a decision-maker sees both the signal and how far to trust it.

Related reading

Sources