Is ASTER or Landsat better for mapping hydrothermal alteration?

ASTER, because it splits the SWIR into five narrow bands (2.145–2.430 µm) that resolve Al-OH, Mg-OH and carbonate absorption features. Landsat collapses that region into two broad bands, so it can flag iron oxides and general clay zones but cannot separate kaolinite from muscovite from chlorite.

Which band ratios detect clay and iron-oxide minerals on each sensor?

On Landsat 8/9 OLI, ferric iron is band 4/band 2 and clay/hydroxyl is band 6/band 7. On ASTER, the classic indicators are band 2/band 1 for ferric iron, band 4/band 6 for Al-OH clays, and band 7/band 6 (or band 5+7 over 6) for argillic alteration.

What spatial and spectral resolution do ASTER and Landsat offer?

ASTER has 15 m VNIR, 30 m SWIR (now failed) and 90 m TIR across 14 bands. Landsat 8/9 OLI has 30 m multispectral, 15 m panchromatic and 100 m thermal across 11 bands. ASTER wins on SWIR detail; Landsat wins on revisit and continuity.

Can I still acquire new ASTER SWIR data?

No. The ASTER SWIR detector failed in April 2008 due to a cryocooler problem, so SWIR bands 4–9 are unusable in scenes acquired after that date. Mineral work now relies on the pre-2008 archive or on Sentinel-2 and hyperspectral missions like EnMAP and EMIT.

ASTER vs Landsat for Mineral Indicators

Short answer

For mineral indicators the choice comes down to the shortwave infrared (SWIR). ASTER splits the 2.0–2.43 µm region into five narrow bands and can resolve the diagnostic Al-OH, Mg-OH and carbonate absorption features that distinguish kaolinite, muscovite/illite, chlorite and calcite. Landsat 8/9 collapses that same region into two broad bands, so it is good for first-pass iron-oxide and general "clay versus no clay" screening but cannot tell alteration minerals apart. Use Landsat (or Sentinel-2) for synoptic, repeatable coverage and time depth; reach for the pre-2008 ASTER archive when you need mineralogical specificity over arid, well-exposed terrain.

The physics: why SWIR band geometry decides the outcome

Most mineral mapping from multispectral data is not classification of the mineral itself — it is detection of an absorption feature caused by molecular bonds. The features that matter for hydrothermal alteration sit in the SWIR:

Al-OH (kaolinite, alunite, muscovite/illite, pyrophyllite): absorption near 2.165–2.215 µm.
Mg-OH and carbonate (chlorite, epidote, calcite, dolomite): absorption near 2.32–2.35 µm.
Ferric iron (hematite, goethite, jarosite): a charge-transfer drop into the blue and a reflectance peak near 0.75–0.9 µm in the VNIR.

A sensor can only "see" an absorption if it has a band sitting inside it and reference bands on the shoulders. This is where the two missions diverge sharply.

ASTER carries 14 bands: three VNIR at 15 m, six SWIR at 30 m, and five TIR at 90 m. The SWIR bands are deliberately placed: band 4 (1.600–1.700 µm) as a high-reflectance reference, band 5 (2.145–2.185), band 6 (2.185–2.225), band 7 (2.235–2.285), band 8 (2.295–2.365) and band 9 (2.360–2.430). Bands 5–8 straddle the Al-OH and Mg-OH features, which is exactly why ASTER became the workhorse of regional alteration mapping in the 2000s.

Landsat 8/9 OLI has 11 bands but only two in the SWIR: band 6 (1.566–1.651 µm) and band 7 (2.107–2.294 µm). Band 7 spans nearly 190 nm — wider than the entire diagnostic Al-OH window — so any clay absorption is averaged into a single number. You lose the shape of the feature, and shape is what separates minerals.

Band ratios you can actually run

Ratios suppress topographic shading (illumination divides out) and isolate the spectral contrast across an absorption. The established indicators:

Landsat 8/9 OLI

Ferric iron / iron oxide: B4 / B2 (red over blue).
Ferrous iron: B6 / B5.
Clay / hydroxyl (argillic): B6 / B7.
Gossan / iron alteration false-color composite: R=B6/B7, G=B4/B2, B=B6/B5.

ASTER (pre-2008 SWIR)

Ferric iron: B2 / B1.
Al-OH clays (argillic, "sericite"): B4 / B6 or the normalized (B4+B6) / B5.
Phyllic / muscovite: B5 / B6.
Mg-OH, chlorite/epidote/carbonate (propylitic): B7 / B8 or (B6+B9) / (B7+B8).
Crósta-style alteration composite: R=B4/B6, G=B4/B7, B=B5/B7.

A worked GDAL example for the Landsat clay ratio, assuming surface-reflectance bands stored as separate GeoTIFFs scaled 0–10000:

gdal_calc.py -A LC09_B6.tif -B LC09_B7.tif \
  --calc="A.astype(float)/(B+0.0001)" \
  --outfile=clay_ratio.tif --type=Float32 --NoDataValue=0

Then stretch to a 2nd–98th percentile range in QGIS (Layer Properties → Symbology → Singleband pseudocolor) so the high-clay tail is visible rather than buried in the histogram. Always run ratios on surface reflectance, never on raw DN or top-of-atmosphere radiance, because path radiance is additive and corrupts a ratio in a scene-dependent way.

Resolution, revisit and continuity

Property	ASTER	Landsat 8/9 OLI
VNIR resolution	15 m	30 m (15 m pan)
SWIR resolution	30 m (detector failed 2008)	30 m
TIR resolution	90 m, 5 bands	100 m, 2 bands
SWIR bands	6	2
Revisit	on-demand, ~16 day	16 day (8 day with two satellites)
Archive	2000–present (SWIR usable to early 2008)	continuous since 1972 across the program

Two operational facts shape real projects. First, ASTER's SWIR detector failed in April 2008 when the cryocooler degraded; bands 4–9 are saturated/unusable after that date. Mineral work therefore lives off the pre-2008 archive. Second, ASTER was a pointable, request-driven instrument, so coverage is patchy — some prospective belts have dozens of clean scenes, others almost none. Landsat images everything on a fixed path/row grid, every pass, which is why it dominates change context and multi-temporal compositing.

ASTER's five TIR bands (8–12 µm) are an underused advantage: TIR responds to Si-O stretching, so band ratios there map quartz and silica content (silicification) and broad lithology — something neither Landsat's two coarse thermal bands nor Sentinel-2 (no thermal) can do.

Worked example: a porphyry-style target in arid terrain

Suppose you are screening a semi-arid belt for porphyry-related alteration zoning (a potassic/phyllic core grading out to argillic and propylitic halos). A defensible workflow:

Pull a cloud-free, dry-season ASTER L2 surface-reflectance scene from before 2008 (NASA Earthdata / LP DAAC). Confirm SWIR is intact.
Reproject everything to the project CRS, e.g. gdalwarp -t_srs EPSG:32612 -r bilinear input.tif utm.tif for UTM zone 12N.
Compute the phyllic (B5/B6), argillic (B4/B6) and propylitic (B7/B8) ratios; load as RGB to reveal zoning.
Bring in a recent Landsat 9 composite for the ferric-iron (B4/B2) gossan signal and for current land cover (mine workings, roads, vegetation flush) that postdates 2008.
Overlay structural lineaments from a DEM hillshade and intersect alteration anomalies with faults/intersections — alteration without a structural control is a weaker target.
Rank anomalies, then plan ground checks. Treat every pixel as a candidate, confirmed only by field spectrometry, XRD or assay.

Common pitfalls and why they happen

Calling a clay ratio "kaolinite." Landsat's broad band 7 cannot distinguish kaolinite from muscovite or smectite; even ASTER needs the full band shape, not one ratio, to be confident. The ratio flags hydroxyl-bearing material, which includes agricultural soils and dry vegetation.
Running ratios on TOA or DN data. Atmospheric path radiance is additive and varies by wavelength, so it biases ratios non-linearly. Mixing scenes from different dates without correction guarantees artefacts.
Ignoring vegetation and shade. Green and especially dry vegetation has Al-OH-like SWIR behavior and will light up "clay" ratios. Mask with an NDVI threshold (often NDVI < 0.2–0.25) before interpreting.
Using post-2008 ASTER SWIR. Those bands are dead. If band 4–9 statistics look flat or saturated, check the acquisition date.
Forgetting ASTER's pointing gaps. A "no anomaly" result may simply mean no usable scene exists over your block.

Quality and validation

Verify the SWIR detector was alive: acquisition date before ~2008-04 for ASTER mineral ratios.
Confirm inputs are surface reflectance (L2), and check CRS, NoData and scaling are explicit before any gdal_calc.py step.
Mask cloud, shadow, water and vegetation; inspect ratio histograms for saturation.
Validate against an independent layer: published geology, a known occurrence, or field/lab spectra (ASD, XRD). A ratio anomaly with no ground tie is a hypothesis, not a result.
Stretch and review the edges of the scene, where cross-track illumination and atmospheric effects are strongest.

Bathyl perspective

We treat ASTER and Landsat as complementary, not competing: ASTER's pre-2008 SWIR/TIR archive for mineralogical specificity, Landsat (and Sentinel-2) for resolution, revisit and an up-to-date base. Every alteration layer we deliver carries its sensor, acquisition date, ratio definitions and the explicit caveat that pixels are exploration vectors to be confirmed on the ground.