The Chandrayaan-2 spacecraft, launched on July 22, 2019, carries the Dual-Frequency Synthetic Aperture Radar, or DFSAR, an instrument that operates in both L-band and S-band frequencies. What makes the L-band particularly valuable is its penetration capability: longer radio wavelengths can probe several meters beneath the dusty lunar surface, retrieving subsurface parameters that earlier S-band missions such as Chandrayaan-1’s Mini-SAR could not reach. In fully polarimetric mode, the radar transmits and receives signals in multiple polarization combinations, allowing researchers to distinguish how electromagnetic waves bounce off surfaces, scatter within volumes of rubble, or reflect in double-bounce geometries from rocks and angled structures.

The team focused on four craters: two polar craters, Hermite-A and Peary, which sit within permanently shadowed regions where temperatures remain below 120 Kelvin, and two non-polar comparison craters, Gardner and Korolev X. Permanently shadowed regions are natural cold traps. Because the Moon’s axial tilt is only about 1.5 degrees, the floors of high-latitude polar craters have not seen sunlight for billions of years, making them ideal repositories for water molecules delivered by comets, asteroids, and solar wind interactions over geological timescales.

The analytical workflow was rigorous. Level-1A single-look-complex DFSAR data from the ISRO Science Data Archive were processed using the MIDAS software developed at ISRO’s Space Applications Centre. The researchers generated a 2×2 scattering matrix capturing the HH, HV, VH, and VV polarization channels, applied a multilooking factor of 38 to improve signal-to-noise ratio, and then used a 5×5 Refined Lee filter to suppress speckle noise while preserving fine surface details. From the resulting coherency matrix, they deployed two complementary decomposition techniques: the eigenvalue-eigenvector-based H/A/alpha decomposition and the model-based Yamaguchi four-component decomposition, which separates scattering into surface, double-bounce, volume, and helix contributions.

The results were striking. Within the Hermite-A crater, volume scattering dominated the interior, contributing approximately 74 percent of the total scattering power, while surface scattering accounted for 21 percent and double-bounce just 5 percent. Outside the crater, the picture reversed entirely: surface scattering dominated at roughly 85 percent. The H/A/alpha analysis reinforced this contrast, showing high entropy values concentrated along crater interior walls, indicative of multiple or random scattering mechanisms, while the smooth exterior terrain exhibited low entropy consistent with simple surface reflection.

Complementing the decompositions, the team computed the Circular Polarization Ratio, or CPR, a diagnostic metric defined as the ratio of same-sense to opposite-sense circularly polarized backscatter. Inside Hermite-A, CPR values ranged from 0.250 to 1.950, with a mean of 1.221, suggesting rough textures and a high probability of water ice deposits. The crater exterior showed a lower mean of 0.594. Similar patterns emerged at Peary crater, where interior CPR values spanned 0.246 to 1.930 compared with 0.064 to 0.962 outside. These elevated CPR values arise from the coherent backscatter opposition effect, a phenomenon in which radio waves traversing low-loss dielectric media such as water ice undergo constructive interference along reciprocal scattering paths, producing anomalously bright, highly polarized returns.

Crucially, the non-polar control craters behaved differently. Gardner crater’s interior was dominated by surface scattering, indicating a relatively smooth floor, while Korolev X showed predominantly surface scattering at its base with only minor volume-scattering regions. Although localized CPR values exceeding unity appeared at both non-polar sites, attributable to rough rocky terrain rather than ice, the pattern of dominant volume scattering within crater walls remained unique to the permanently shadowed polar craters. This distinction supports the idea that the polar anomalies reflect subsurface volatiles rather than mere topographic roughness, aligning with earlier Mini-RF studies that identified so-called anomalous craters whose radar properties differ from typical lunar craters.

To validate the radar interpretation against ground truth, the researchers turned to an unprecedented natural experiment: the Chandrayaan-3 landing site. Using high-resolution 75-megahertz L-band compact polarimetric DFSAR imagery captured both before and after the Vikram lander’s touchdown, they applied the improved S-Omega decomposition. Before landing, the site’s smooth fine-grained regolith produced 61.8 percent surface scattering, 11.0 percent double-bounce, and 27.1 percent volume scattering. After landing, the radar response shifted dramatically toward double-bounce and volume scattering, reflecting the lander’s structure and the ejecta-blanketed, disturbed terrain. CPR jumped from 0.5 pre-landing to approximately 2 post-landing, with a maximum of 1.99 plus-or-minus 0.50, demonstrating exactly how rough, blocky debris alters polarimetric signatures.

The authors are careful to emphasize that high CPR values and volume scattering alone cannot uniquely confirm water ice; rough surfaces, blocky ejecta, and fresh impacts can mimic the same signatures. However, when integrated with permanently shadowed conditions, persistently frigid temperatures, and prior neutron spectrometer evidence of hydrogen-rich regions, the combined polarimetric framework provides a substantially more reliable means of flagging candidate ice deposits. This matters enormously for exploration strategy: water ice at the lunar poles represents a potential resource for drinking water, oxygen, and rocket propellant, making accurate identification of ice-bearing craters essential for planning future crewed and robotic missions.

Looking ahead, the team proposes integrating additional radar-derived parameters, including degree of polarization, dielectric constant, and surface roughness, with complementary datasets such as Diviner thermal measurements, LOLA topography, and neutron spectrometer observations. Such multi-sensor fusion should further reduce ambiguities inherent to radar-only detection. For now, the study stands as a demonstration that Chandrayaan-2’s fully polarimetric L-band radar, combined with disciplined polarimetric decomposition and ground-truth validation from Chandrayaan-3, has given planetary scientists a powerful new toolkit for reading the Moon’s frozen archives, one shadowed crater at a time.

Subject of Research: Detection of potential lunar water ice using polarimetric decomposition of Chandrayaan-2 L-band DFSAR radar data in permanently shadowed polar craters

Article Title: Polarimetric decomposition-based analysis of Chandrayaan-2 L-band DFSAR data for potential water ice detection in lunar polar regions

Article References: Polarimetric decomposition-based analysis of Chandrayaan-2 L-band DFSAR data for potential water ice detection in lunar polar regions. (n.d.). https://doi.org/10.1007/s10686-026-10077-5

Image Credits: AI Generated

DOI: 10.1007/s10686-026-10077-5

Keywords: Chandrayaan-2, DFSAR, water ice, lunar polar regions, permanently shadowed regions, polarimetric SAR, circular polarization ratio, Hermite-A crater, Peary crater, Yamaguchi decomposition, Chandrayaan-3, lunar regolith

Water is the most valuable resource candidate on the Moon. It can be consumed by astronauts, split into oxygen for breathing, used as radiation shielding, and converted into liquid hydrogen and oxygen propellant, the lifeblood of any space transportation economy. Evidence for lunar water comes from orbital measurements and from NASA’s LCROSS mission, which slammed a kinetic impactor into the permanently shadowed floor of Cabeus crater near the lunar south pole in 2009 and detected water vapour in the ejecta plume. Yet significant uncertainty remains about how much ice exists, where it sits, and whether it appears as thin films, porous frost or consolidated layers, all of which complicate the design of extraction hardware before a single watt of power is budgeted.

The DLR team systematically searched Scopus, Web of Science and Google Scholar for studies that detailed experimental or simulated lunar water extraction with quantitative data. They recorded design, objectives, environmental parameters, masses and power figures for every qualifying study, with at least two reviewers independently verifying each entry. The resulting survey spans an enormous range of scales. Some laboratory experiments processed only a few grams of icy simulant, while the largest test used nearly twenty-nine kilograms. Simulations ranged from millimetre-scale setups to concepts larger than one metre, drawing from a few watts to several solar constants’ worth of concentrated energy.

From this survey, the authors built the first unified classification of extraction methods, grouping designs by their underlying physical mechanism. Thermal extraction dominates the field: heat from solar concentrators, electrical heaters, microwaves or future nuclear sources is applied to icy regolith, causing the ice to sublimate directly into water vapour in the Moon’s near-vacuum. Thermal methods split further by how the soil is handled. In situ approaches heat regolith where it lies, using capture domes, heated rods or corer drills, while excavated approaches truck the soil to a processing unit, either continuously through augers and conveyors or in discrete batches inside crucibles and rotating heated drums.

Beyond heat, three more exotic families appear in the classification. Mechanical extraction, exemplified by the Aqua Factorem concept from the University of Central Florida, treats ice as a mineral ore, grinding icy regolith and separating ice grains pneumatically, magnetically, vibrationally and electrostatically without melting anything. Convection-assisted extraction is represented by the Rocket Mining System developed by Masten, now part of Astrobotic, with Honeybee Robotics and Lunar Outpost, which fires rocket plumes into the soil to fluidise icy regolith and vacuum up the ejected particles. Ionisation extraction, in the form of ablative arc mining, uses an electric arc to ablate and ionise the regolith itself, steering ionised particles into collection reservoirs sorted by mass with electric and magnetic fields, potentially harvesting multiple elements at once.

The review’s most pointed criticism concerns reporting. Ice content was sometimes given as weight percent and sometimes as volume percent. Only twelve of fourteen experimental studies reported electrical power, only nine reported total simulant mass, and a mere five explicitly stated the mass of ice they started with. Simulations fared little better: just six of thirteen listed numerical starting pressures, and only two explicitly mentioned captured water rather than sublimated mass. The median experiment used roughly half a kilogram of simulant, and only one test exceeded five kilograms, leaving the crucial five-to-thirty-kilogram bracket, the bridge to pilot-scale operations, nearly empty. Most experiments ran for minutes to a few hours, and endurance beyond twenty-four hours of steady operation remains essentially untested, leaving questions of hardware degradation unanswered.

To fix this, the authors propose a standardised reporting framework for the field. They call for every study to report total simulant and ice mass, electrical and thermal power with duty cycles, total energy consumed, run times, and captured and recovered water masses, enabling direct calculation of extraction, capture and recovery efficiencies. They also stress that energy accounting must cover the whole chain, not just sublimation. Capturing water vapour, a step handled by cold traps or condensers, introduces additional power draw and losses, so the highest-level metric, total energy per kilogram of recovered water, must include every input. Comparisons, they caution, should only be made between studies at similar environmental fidelity, since a test in a dusty thermal vacuum chamber cannot fairly compete with one run in open laboratory air.

The review also shines a light on a quieter source of inconsistency: the simulants themselves. Icy lunar regolith is prepared in strikingly different ways, from the common gradual freezing or mud-pie method, which produces strong, cemented, permafrost-like material, to vapour deposition and dry-mix sintering, which avoid bulk liquid water and better resemble genuine lunar conditions. Microstructural studies show wet-mixed samples have smaller pores, lower porosity and higher cohesion than sintered ones, directly altering thermal conductivity and vapour transport. Simulant choice matters too: JSC-1, LHS-1, LMS-1, NU-LHT-2M and others differ in particle size distribution and thermal properties, which control permeability and diffusion under low pressure, meaning that two experiments using different simulants may not be comparable at all. And no current simulant fully reproduces real lunar regolith, so terrestrial performance will inevitably deviate from what a lunar mission encounters.

The identified research gaps are as much about integration as invention. Almost no study couples extraction hardware with vapour capture in a single end-to-end test; the few pilot-scale efforts, such as a drilling-based thermal extraction unit that achieved water collection rates up to roughly 0.75 millilitres per minute with recovery efficiencies above eighty percent, and the European LUWEX project validating extraction, capture and purification with up to fifteen kilograms of simulant, remain rare exceptions. Lunar dust, abrasive and electrostatically charged, is barely addressed in extraction designs despite its known threat to mechanisms and thermal surfaces. Pre- and post-processing steps such as prospecting, excavation, beneficiation, purification and storage must be folded into any honest comparison, and scalable surface demonstrations on the Moon itself will ultimately be required to prove that laboratory success survives radiation, vacuum and extreme temperatures.

What emerges from this review is less a ranking of winners and losers than a roadmap for a maturing field. The authors frame their classification as a starting point, to be refined with input from the wider in-situ resource utilisation community, and their standardised metrics as an invitation to collective agreement on terms, definitions and transparent reporting. If researchers adopt them, the next generation of water extraction studies could finally be compared apples to apples, accelerating the selection of architectures capable of turning permanently shadowed craters into the solar system’s first off-world wells. The alternative, the authors imply, is a patchwork of incomparable results that slows the very missions counting on lunar water to fuel humanity’s return to the Moon and, eventually, the journey beyond it.

Subject of Research: Lunar water extraction technologies for in-situ resource utilisation

Article Title: A review of lunar water extraction technologies: comparison, classification, and research gaps

Article References: Kiewiet, L., Fälker, S., & Zabel, P. (2025). A review of lunar water extraction technologies: comparison, classification, and research gaps. Space and Planetary Resources, 1(1), Article 4. https://doi.org/10.1007/s44461-025-00005-4

Image Credits: AI Generated

DOI: 10.1007/s44461-025-00005-4

Keywords: lunar water extraction, in-situ resource utilisation, icy regolith, thermal mining, permanently shadowed regions, LCROSS, water vapour capture, cold traps, lunar regolith simulants, ISRU, Moon, space resources