Mapping software

Minerals

from spectra to maps

Tetracorder · USGS
Expert system that compares each spectrum with a library of laboratory spectra and assigns the closest mineral. Used on M³ data.
Modified Gaussian Model · MGM
Splits a spectrum into its individual absorption bands, to tell apart different pyroxenes or olivine, for example.
Radiative transfer models · Hapke
Estimate the percentage of each mineral in a mixture. The global maps of olivine, pyroxene and plagioclase from Kaguya data were made this way.
Lucey algorithms
Formulas that derive iron and titanium from ratios between colour bands. Written for Clementine and still in use.
ENVI · PyHAT
Environments for hyperspectral analysis that support M³ data. ENVI is commercial; PyHAT is a Python tool from USGS.

Research code · many groups

Machine learning

Neural networks and “unmixing” methods applied to M³ cubes to classify minerals automatically. Mostly research code tied to single papers, not established products.

One spectrum per pixel · the method splits it into mineral shares · the shares become the map · schematic

Caves and pits

from images, radar and gravity

LROC team · Arizona State University

PitScan

Algorithm that searched LRO’s high-resolution images automatically for anomalous shadows. It produced most of the catalogue of over 200 pits, published as the Lunar Pit Atlas. Source

Sun under 50° from vertical: craters lose their shadows · a shadow lit on its up-Sun side is a rock · the rest go to a person · schematic

University of Kent

PITS · Pit Topography from Shadows

Calculates the depth of a pit from the shape of its shadow. Source

Shadow is outlined automatically · its width L and the Sun angle i give the depth · schematic, 100 m pit

University of Kent

ESSA · Entrances to Sub-Surface Areas

Deep-learning model trained to recognise cavity entrances in images of the Moon and Mars. It flagged some new candidates. Source

Learns from labelled pits · scores every part of a new image · high scores are checked by a person · schematic

University of Trento

Radar simulations

3D electromagnetic models that predict the radar echo of a pit with or without a conduit, to compare with the real data. The method behind the 2024 result at Mare Tranquillitatis. Source

Pulses that enter a conduit return late · late echoes plot as a bright tail · only model B matches the observation · schematic

Purdue University

GRAIL gravity analysis

Gradiometry and correlation techniques that look for the linear mass deficits typical of buried lava tubes. Source

Less mass below, weaker pull · the gap between the twins changes · the dip is matched to the signal of a tube · schematic

The animations are schematics. Shapes, angles and numbers are illustrative, not data.

Sources

Caves
Wagner & Robinson 2018 · PitScan · LROC · Lunar Pit Atlas · Le Corre 2023 · PITS · Le Corre 2025 · ESSA · Carrer 2024 · Nature Astronomy · Chappaz 2017 · GRAIL
Minerals, no link
Entries rest on the published record: Clark 2003 (Tetracorder), Sunshine 1990 (MGM), Hapke 1981, Lucey 2000, Lemelin 2015 (Kaguya mineral maps).