Lunar Mining & Landing Sites

by Kevin M. Cannon

1. Ice Favorability Index Maps
2. Polar Terrain Type Maps
3. Cold Trap Accessibility
4. Metals, Oxygen, & Basalt Fiber Index Maps
5. Polar Topographic Maps
6. Ice Prospecting Guide

Lunar Mining & Landing Sites

This page hosts a growing list of tools related to lunar mining and landing sites. Contact me:; 401-340-6608.

Ice Favorability Index

The Ice Favorability Index (IFI) maps below are based on a geologic system model for ice deposition and evolution (Cannon and Britt 2020, Icarus 347, 113778). The index maps are predictive, not based on surface ice detections, and highlight regions likely to host the most favorable ice deposits for mining. Desired characteristics include older ages, ice stability closer to the surface, and higher areal fraction of cold traps. These maps are agnostic in terms of mining architecture, and some areas may only be accessible with nuclear power for example.

GeoTIFF for north pole (80-90°)

GeoTIFF for south pole (80-90°)

References (Please cite if you use this data):
Cannon, K. M., and D. T. Britt (2020), A geologic model for lunar ice deposits at mining scales. Icarus 347, 113778.

Click to view full size.

Polar Terrain Type Maps

In Cannon and Britt (2020) we divided the polar terrains into 9 different Terrain Types based on a simple adaptation of Matt Siegler's ice stability depth maps (Siegler et al. 2016). These have implications for mining strategies and methods, as well as landing site selection. The divisions are:

TT1: Ice stable at the upper surface (macro cold traps)
TT2: Ice stable at <1 m depth (micro cold traps at surface)
TT3: Ice stable at >1 m depth (micro cold traps at surface)

Based on the episode of true polar wander proposed by Siegler et al. (2016), there are 9 permutations for terrains both before and after this event, assuming a terrain is old enough that it pre-dates polar wander. These permutations are designated like TT2→3 for a terrain that changed from TT2 to TT3. The permutation maps are available below:

GeoTIFF for north pole (80-90°)

GeoTIFF for south pole (80-90°)

References (Please cite if you use this data):
Cannon, K. M., and D. T. Britt (2020), A geologic model for lunar ice deposits at mining scales. Icarus 347, 113778.
Siegler, M. A. et al. (2016), Lunar true polar wander inferred from polar hydrogen. Nature 531, 480-484.

Click to view full size.

Data values: 1 = TT3→3 (dark blue); 2 = TT3→2 (green); 3 = TT3→1 (dark gray); 4 = TT2→3 (teal); 5 = TT2→2 (blue); 6 = TT2→1 (light gray); 7 = TT1→3 (pink); 8 = TT1→2 (purple); 9 = TT1→1 (white)

Cold Trap Accessibility

In Cannon and Britt (submitted), we calculated the accessibility of large permanent cold traps at the lunar poles for wheeled vehicles. Accessibility metrics included minimum energy paths, minimum distance paths, and lowest maximum slope paths for entry, egress and round trips between a target location within the cold trap, and illuminated, low-slope staging areas outside the cold trap. 55 of the 59 cold traps studied were found to be accessible with <25° slopes. Smaller cold traps are generally more accessible, but several of the huge south pole cold traps have very low-energy entry paths.

References (Please cite if you use this data):
Cannon, K. M., and D. T. Britt, Accessibility dataset for large permanent cold traps at the lunar poles, manuscript submitted to Earth and Space Science.

Click to view full size.

ArcGIS shapefiles are available for the cold traps in the study, with attribute fields for all the accessibility metrics:

Download shapefiles (.zip; 60 KB)

Accessibility metrics for north polar cold traps. Distance and slope metrics are given for round trip routes. Italicized entries have no paths with a maximum slope <25°.Click cold trap ID for summary images

Cold trap ID Min. energy down (joules) Min. energy up Min. energy round-trip Dist. (km) Max. slope (°) Best dist. Best max slope Score
NP_UN04 3.40E+05 1.20E+06 1.60E+06 15 22 15 17 397
NP_UN05 7.70E+05 1.10E+06 1.80E+06 16.8 13 16.8 13 392
NP_UN01 4.80E+05 1.10E+06 1.60E+06 17 19 17 17 355
NP_UN02 4.60E+05 9.80E+05 1.40E+06 14.4 19 14.4 19 354
NP_UN03 6.40E+05 1.10E+06 1.80E+06 18.4 22 18 16 343
NP_HEVE 5.00E+05 1.60E+06 2.20E+06 21 22 21 15 332
NP_UN07 4.30E+05 1.40E+06 1.90E+06 18.2 24 18.2 19 299
NP_FIBI 5.80E+05 1.20E+06 1.90E+06 19.1 24 19.1 18 284
NP_UN06 5.10E+05 1.20E+06 1.80E+06 16.9 22 16.9 22 275
NP_UN08 9.30E+05 1.50E+06 2.60E+06 24.6 16 24.2 16 267
NP_ROZW 7.20E+05 2.00E+06 2.70E+06 26.8 20 26.8 13 266
NP_NANF 5.50E+05 1.70E+06 2.20E+06 21 20 21 20 249
NP_HERM 6.30E+05 2.00E+06 2.60E+06 27 19 27 16 236
NP_ROZH 4.80E+05 2.00E+06 2.40E+06 24 25 24 19 233
NP_LENA 7.30E+05 2.30E+06 3.20E+06 32.8 21 32.8 13 216
NP_NANA 7.30E+05 2.30E+06 3.00E+06 30 23 30 18 170
NP_ROZU 5.50E+05 2.20E+06 2.80E+06 27.3 25 27.3 20 159
NP_HERA 24.1 34 100
NP_LOVE 6.80E+05 3.10E+06 3.80E+06 36.4 25 36.4 23 86
NP_SYLV 1.00E+06 3.20E+06 4.20E+06 40.7 24 40.7 22 69

Accessibility metrics for south polar cold traps. Distance and slope metrics are given for round trip routes. Italicized entries have no paths with a maximum slope <25°.Click cold trap ID for summary images

Cold trap ID Min. energy down (joules) Min. energy up Min. energy round-trip Dist. (km) Max. slope (°) Best dist. Best max slope Score
SP_UN14 2.4E+05 8.6E+05 1.2E+06 11.2 18 11.2 12 496
SP_UN18 2.8E+05 7.7E+05 1.0E+06 9.9 18 9.9 14 486
SP_UN03 3.8E+05 1.1E+06 1.5E+06 14.3 16 14.3 13 457
SP_UN17 1.7E+05 8.4E+05 1.0E+06 9.6 18 9.6 17 453
SP_UN16 4.2E+05 1.0E+06 1.4E+06 14.6 21 14.6 13 446
SP_UN19 1.7E+05 8.6E+05 1.0E+06 9.6 23 9.6 18 440
SP_SLAT 2.0E+05 9.0E+05 1.1E+06 11.4 23 11.4 18 417
SP_UN20 4.2E+05 9.7E+05 1.8E+06 16.7 15 16.7 15 407
SP_UN10 3.5E+05 1.1E+06 1.5E+06 14.4 18 14.4 17 403
SP_UN04 4.4E+05 1.4E+06 1.8E+06 17.8 19 17.8 14 390
SP_UN07 6.1E+05 1.1E+06 1.7E+06 15.9 21 15.9 15 389
SP_UN15 2.8E+05 1.4E+06 1.6E+06 15.8 23 15.8 17 381
SP_WIEC 4.4E+05 1.6E+06 2.1E+06 19.3 23 19.3 12 380
SP_NEFE 3.7E+05 1.7E+06 2.0E+06 19.2 24 19.2 15 350
SP_UN09 4.6E+05 1.3E+06 1.8E+06 18.2 22 18.2 16 345
SP_UN02 6.7E+05 1.6E+06 2.3E+06 21.9 16 21.9 14 327
SP_WIJ2 4.1E+05 1.0E+06 1.4E+06 14.4 24 14.4 23 322
SP_UN01 4.4E+05 1.4E+06 2.3E+06 23.5 23 22.5 16 306
SP_UN21 3.2E+05 1.7E+06 2.1E+06 20.9 22 20.9 19 277
SP_NOBI 7.2E+05 2.0E+06 2.7E+06 26.2 21 26.2 16 249
SP_SCOT 4.5E+05 2.0E+06 2.4E+06 23.0 23 23.0 20 232
SP_UN11 4.6E+05 2.0E+06 2.5E+06 24.2 23 24.2 19 210
SP_FAUS 3.7E+05 2.1E+06 2.5E+06 25.2 23 25.2 22 204
SP_UN05 8.7E+05 2.2E+06 3.2E+06 30.4 25 30.4 16 198
SP_SHOE 3.3E+05 2.2E+06 2.5E+06 26.6 23 25.9 22 193
SP_CABB 4.3E+05 1.8E+06 2.7E+06 28.4 22 28.4 21 187
SP_WIJ1 3.7E+05 2.2E+06 2.5E+06 24.5 25 24.5 24 177
SP_SVER 6.8E+05 2.3E+06 3.1E+06 30.8 18 30.8 18 160
SP_UN06 4.5E+05 1.9E+06 3.6E+06 36.3 24 36.3 21 134
SP_UN12 1.3E+06 3.0E+06 4.3E+06 42.3 18 42.1 18 128
SP_DEGE 7.8E+05 2.2E+06 3.0E+06 28.8 25 28.8 22 118
SP_AMUN 1.2E+06 3.0E+06 4.2E+06 40.9 20 40.9 19 106
SP_CAB1 7.4E+05 3.1E+06 3.9E+06 38.2 20 38.2 20 104
SP_HAWO 7.3E+05 2.4E+06 3.2E+06 31.0 24 31.0 23 90
SP_IDLL 35.3 30 55
SP_CAB2 2.2E+06 3.8E+06 6.0E+06 61.4 22 61.4 22 50
SP_SHAC 40.9 35 41
SP_UN13 1.3E+06 4.3E+06 5.7E+06 55.0 25 55.0 25 27
SP_UN08 54.3 33 23

Metals, Oxygen, & Basalt Fiber Index Maps

These maps are based on existing datasets produced by the Clementine and LRO missions.

The Heavy Metals & Oxygen Index (HMOI) is based on measured concentrations of FeO and TiO2, and OMAT, the optical maturity of the regolith. High OMAT (low maturity) is favored because of the difficulty using magnetic separation on mature regolith. The index shows areas rich in iron and titanium, which are also favored for oxygen production using Molten Regolith Electrolysis (MRE), and Hydrogen Reduction, respectively. As expected, the index mostly captures Lunar mare regions.

GeoTIFF (equirectangular projection; 1.4 GB)

The Light Metals Index (LMI) is currently based on inferred concentrations of Al2O3 based on an anticorrelation with (FeO + TiO2). As expected, the index mostly captures highlands regions. Future updates will incorporate Al, Mg & Si based on modeled mineral abundances.

GeoTIFF (equirectangular projection; 2.3 GB)

The Basalt Fiber Index map is based on the fact that more acidic basalts are better suited for pulling basalt fibers, with about 46% SiO2 or higher required. The index map uses an inverse correlation between TiO2 and SiO2 in lunar soils.

GeoTIFF (equirectangular projection; 1.2 GB)

Polar Topographic Maps

The collection of maps below were created using publicly available datasets (as indicated) and are free to use or modify in any way. Additional mapping products and ArcGIS files will be added in the future. Customized maps are available upon request.

Blank maps: 75° to 90°

Click to view full size (~55 MB). Based on LRO_LOLA_DEM_NPole75_30m and LRO_LOLA_DEM_SPole75_30m datasets.

Annotated maps with PSRs: 75° to 90°

Click to view full size (~55 MB). Based on LRO_LOLA_DEM_NPole75_30m and LRO_LOLA_DEM_SPole75_30m datasets for topography, and the LPSR_75N_120M_201608.JP2 and LPSR_75S_120M_201608.JP2 datasets for PSR locations.

Ice Prospecting: Your Guide to Getting Rich on the Moon

Version 2.0 // August 2019

Kevin M. Cannon (

Creative Commons License

This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License

PDF Version (12.4 MB)

ZIP Version (63.7 MB), includes PDF and larger map files

Changelog (1 KB)

Full text reproduced below.

Key Terms

Permanently shadowed regions (PSRs): areas on the Moon’s surface that never receive direct sunlight during the 19-year lunar orbital period.
Regolith: the meters-thick layer of finely pulverized rock that coats the surface of the Moon.
Volatiles: elements or compounds with particularly low boiling points, including H2O, CO2, SO2, Hg, etc., that tend to vaporize easily.
Impact gardening: the process by which small meteoroid impacts churn the regolith, mixing it and bringing fresh material to the surface.
LCROSS: A NASA mission that impacted the upper stage of a rocket into a PSR and detected ~5-6% water ice in the ejected plume of material.


Water ice has been detected within and surrounding permanently shadowed regions (PSRs) at both poles of the Moon. This ice is an attractive target for mining to produce liquid oxygen and hydrogen propellants, and water and oxygen for human life support. However, just the presence of ice at the poles does not provide much information: Where is it exactly? How much is there? Is it thick layers of pure ice, small amounts mixed in the soil, or just thin frosts? How hard is it to dig out, and what are the best mission architectures for a mining outpost? This white paper offers answers to these questions based on the best data now available. New prospecting missions in the future–especially landers and rovers–will continue to improve our understanding of ice on the Moon. This guide will be updated on an ongoing basis to incorporate new findings.

Why is there ice on the Moon?

Two factors create conditions that allow ice to pile up and persist at the lunar poles: (1) the Moon has a very small spin axis tilt of 1.54° compared to 23.5° for the Earth, and (2) the Moon has rough topography, mostly due to large craters formed by asteroid and comet impacts. Combined, these factors lead to topographic lows mostly in crater floors near the poles that never receive direct sunlight (the PSRs), or are seasonally shadowed (SSRs, seasonally shadowed regions). As one can imagine, areas of near- permanent darkness will be cold on a planetary body with no atmosphere, and areas at the lunar poles reach temperatures below 40 K. At average temperatures below about 110 K, water ice in a hard vacuum is stable against sublimation on billion-year timescales, allowing ice that accumulated in the past to possibly remain there today. Ice is even more stable when covered with a layer of lunar soil (regolith) that protects against exposure to the space environment.

What is the source of the ice at the poles? There are three main processes that probably all contribute to ice accumulation (Figure 1): (1) Impacts of water- rich asteroids and comets. These high-energy collisions deliver water and other volatiles (sulfur, carbon, etc.) to the Moon, and especially large impacts can form a short-lived atmosphere in the hours to days after the impact event. Some of the H2O and other volatile molecules in this atmosphere are cold trapped at the poles, building up deposits millimeters to centimeters thick for large enough impacts. Cratering rates were many, many times higher in the early history of the solar system (4.5-3.5 billion years ago), so most ices deposited by this process are expected to be old. (2) Volcano-related outgassing. Volcanic vents and cracks erupt lava on the surface, but at the same time spew out gases including water vapor and other volatiles. Volcanic activity has more or less stopped on the Moon, but was much greater in the past with very old eruptions pumping out large amounts of gases. Some of these should have ended up at the poles, and ice deposits centimeters to meters thick could have formed if volcanic activity occurred in brief intense spikes. (3) Solar wind. The solar wind is a stream of electrons, protons and other particles coming from the sun that are constantly smashing into the unprotected surface of the Moon. This interaction can create individual OH and H2O molecules that are able to ballistically hop across the surface, eventually moving to the poles. However, solar wind can also wear away ice at the upper surface of the polar cold traps, and it is debated whether net accumulation should happen in these areas.

Multiple processes change the concentration and distribution of ice once it has accumulated at the surface of cold traps. Smaller impacts churn the soil and mix ice both vertically and horizontally. Given enough time this will break up any pure ice layers at the surface and incorporate this ice into the surrounding dry soil. The result will be patchy and thinly spread ice in the regolith that differs in concentration on scales of meters to kilometers. Under certain conditions H2O molecules at the surface can be “pumped” deeper into the underlying regolith by diffusion, but this process may be inefficient on the Moon. Ice is lost due to sublimation, micrometeorite bombardment, and other processes at the surface, which will reduce the ice concentration over time without constant resupply. A complicating factor is that the current spin axis of the Moon may not have always been located in the same place as it is now. If the spin axis changed in the past as has been proposed, certain areas where ice accumulated will have become warmer and other areas will have become colder, allowing ice to accumulate where it could not before. Another complication is that the large craters that host PSRs all had to have formed at some point in lunar history. Those that formed earliest have had the chance to accumulate more ice over time, but could have been affected by massive nearby impacts that deeply buried their ice deposits. Overall, there has been a complex history of ice accumulating, being diluted into surrounding soils, and being lost over time. Current thinking suggests most major accumulation at the poles happened more than 2-3 billion years ago, and mixing and loss processes have dominated since then. If there ever were meters-thick sheets of nearly pure ice at the surface, they have been deeply buried or battered away.

Ice locations and concentrations

Figure 2 shows where the PSRs are in the north and south polar regions. There are more PSRs at the north pole but they tend to be smaller than those at the south pole. Every PSR on the Moon has the potential to host ice but the available data do not show all of them do. It’s also possible for ice to exist outside the PSRs if it’s buried under a dry layer of regolith, or in the surrounding SSRs during certain seasons. Ice concentrations have been estimated by different instruments on orbiting satellites, and by the LCROSS experiment that impacted into Cabeus crater near the south pole.

Orbital remote sensing instruments have different spatial resolutions and sensing depths: this makes interpreting ice locations and concentrations challenging. Instruments that depend on reflected UV, visible, and near-infrared light only sense to microns or millimeters below the surface, while microwave, radar and neutron spectroscopy penetrate centimeters to meters. So, two different instruments can give different values for ice concentration at the same location and this is not necessarily a contradiction. Table 1 lists the main instruments used to infer or directly detect polar ice on the Moon. The table describes limitations of these measurements, depths below the surface they sense to, and presents ice concentrations as pessimistic, realistic, and optimistic cases, trying to capture different interpretations by the many people who have studied these data. Overall the data are consistent with the presence of two broad groups of ice deposits: surface ice (frost), and deeper diffuse ice.

Table 1. Reported ice concentrations from orbital datasets.

Instrument Spatial Resolutiona Sensing Depth Limitations Ice: pessimistic Ice: realistic Ice: optimistic
Lunar Prospector Neutron Spectrometer 500 m 0-70 cm Cannot distinguish uniform low abundances from buried wet layer, senses all H not just H2O Very low ice concentrations (<1%) spread over wide regions Heterogeneous ice distributions with richest PSRs at 1-4% ice, buried wet layers with >1% ice Ice highly clustered with local regions and buried wet layers at >10% ice
Lunar Crater Observation and Sensing Satellite 20-25 mb 0-3 m Represents single point on the surface; may not be representative 2.7% ice at the impact site 5.6-6.3% ice at the impact site 8.5% ice at the impact site
Mini-RF 15 m 0-10 mc Enhanced circular polarization ratios indicative of subsurface ice are non-unique Enhanced CPRs in polar craters are entirely due to rocks, no ice indicated Upper limit of 5-10% ice in the subsurface, but ice not uniquely identified Meters-thick layers of nearly pure ice in the subsurfaced
Lyman-Alpha Mapping Project 76.4 m μm-mm Relies on ratios/slopes instead of ice spectrum; cannot distinguish H2O from OH Signal can be attributed to OH, very little H2O Frosts covering up to 10% of the surface, or diffuse surface ice with 0.1-2% concentration Even more extensive ice not detected due to low SNR
Lunar Orbiter Laser Altimeter 240 m μm-mm Single wavelength, enhanced albedo not unique to ice High albedos caused by other factors, little to no ice Thin frosts or diffuse ice with 7% concentration Even more extensive ice not detected due to strict criteria used to report detections
Lunar Exploration Neutron Detector 10 km 0-70 cm Instrument may not operate as reported (see discussion in Teodoro et al. 2014) Much lower spatial resolution than reported, no spatially resolved ice detected Up to 0.5% ice over large areas covering the PSRs Ice highly clustered with local regions or buried wet layers at >10% ice
Moon Mineralogy Mapper 280 m μm-mm Uses multiply bounced photons for PSR investigations: low SNR Ice present as thin frosts with low overall yields Up to 20 wt.% ice if present as intimate ice-regolith mixture, or 20 vol.% if pure ice patches Even more extensive ice not detected due to low SNR
Chang'E microwave radiometer 280 m cm-m Unclear how to link the band ratios to physically meaningful properties Dataset not fully understood, ice not uniquely detected Areas of subsurface ice in some large PSRs More extensive ice than reported in initial analysis

aRepresents maximum resolution used in studies reporting ice detections; these are often binned pixels and the maximum resolution of the instrument can be higher. bEstimated diameter of the crater created by LCROSS. cFor S-band. dThis interpretation is widely disputed.

Surface ice and frost

In this paper surface ice is defined as ice within the upper microns to millimeters of the surface, including true frosts that are thin coatings of ice on the surface of regolith grains. If true frosts are present, they probably formed more recently than deeper ice, and newer work has suggested frosts are very short-lived and can only last as long as a few thousand years. This is controversial and will likely continue to be debated in the future. Figure 3 shows locations of inferred surface ice based on orbital data that use reflected light to detect the spectrum of ice or enhanced brightness interpreted to be caused by ice. Ice is detected where current average temperatures are below 110 K, but not all locations this cold show surface ice deposits. The data show the surface ice is very patchy, but cannot distinguish whether it is a true frost or loose ice grains mixed with dry regolith.

Frosts and surface ice can be easily mined by mechanical excavation or thermal methods, but yields will likely be low if ice is only present in thin patchy layers. To estimate yields for frost deposits, one can use the equation:
Mice = 0.00917*APd (1)
where Mice is the mass of ice in metric tons, A is the total area mined in km2, P is the patchiness in percent (i.e., how much of the surface is covered in frost), and d is the thickness of the ice in microns. For example, if all of the 233.7 km2 Shackleton crater PSR is covered in a 10% patchy micron-thick frost layer, this would add up to only ~20 t of ice. Because of this, extreme caution should be taken when using ice concentrations reported from orbital datasets at shorter wavelengths (Table 1) for prospecting purposes. These data should be coupled with neutron spectroscopy or other products that sense deeper, and missions to the surface will be needed to trench or drill beneath any surface deposits to evaluate ice concentration at depth.

Deeper diffuse ice

Evidence for ice deeper than the upper few microns comes from orbital neutron spectroscopy, the LCROSS mission, and statistical analyses of crater shapes at the poles. This ice was probably deposited at the surface via the sources described above, then mixed into the surrounding dry regolith by billions of years of impacts churning the soil. Neutron spectroscopy is sensitive to small amounts of ice in the subsurface but this technique has very poor spatial resolution and cannot be used to figure out ice layer thickness and concentration at the same time. Most reported ice abundances within the lunar poles are 1% ice by weight or less if a uniform meter-thick layer is assumed (Table 1), but there are reasons to be more optimistic: (1) These values are averaged over large areas, and ice could be mainly located in smaller clusters such that local or regional concentrations are much higher. The LCROSS experiment detected 6% ice in an area where neutron data indicated only 1%, showing richer potential resources are in fact present at local scales. (2) Most PSRs are in craters, and the roughness of craters can affect neutron measurements leading to underestimates of ice contents.

A recent analysis of small crater shapes at the lunar poles suggests meters-thick highly concentrated deposits (30-50% ice) fill many craters throughout the south pole but not in the north pole. However, these results seem to conflict with the neutron spectroscopy data and require more analysis. It is also not clear why there would only be thick deposits at the south pole, when ice has been detected at both poles using other techniques. Yields from deeper diffuse ice can be estimated with the equation:
Mice = ACd(16800-113C) (2)
where Mice is the mass of ice in metric tons, A is the total area mined in km2, C is the concentration of ice in weight percent, and d is the thickness of the ice-rich deposit in meters. For example, if the whole Sverdrup crater PSR has a 1-meter deposit with 2.5% average ice concentration, this would equal ~20 million t of ice. Promising ice deposit locations can be estimated based on thermal models of where ice was likely stable in the past and where it is now stable, combined with models of the effects of impact gardening on ice concentration and distribution. Figure 4 shows my maps of the most promising locations for mining, agnostic of the operations scenario. These maps do not use the detected locations of surface ice/frost as an input because frosts are unlikely to form economic deposits.

Outstanding questions

Reasonable interpretations about polar ice deposits can be made when orbital datasets are combined and their limitations are understood (Table 1). That said, there are major outstanding questions needing future orbital and landed missions to fully address. These are some of the more pressing questions related to mining:

Physical properties of lunar ice deposits

As now understood, most orbital data do not support the presence of pure ice sheets or ice- cemented regolith at the poles. These types of deposits could in theory be present especially at depths greater than a meter, but it should not be assumed they exist unless proven true by future missions. Some studies using radar data and small crater shape statistics infer thick pure deposits exist and are buried under the surface, but this conflicts with LCROSS results and neutron spectroscopy data. The current balance of the evidence suggests lunar ice is in the form of frosts, and loose ice grains mixed in the soil at low to moderate concentrations (1-10% by weight) as discussed above. The composition and physical properties of the ice-bearing regolith are probably similar to highlands soils elsewhere on the Moon: these soils are mostly made of the rock type anorthosite and are unconsolidated with high porosity near the surface, and are more compact at depth. Bulk densities for lunar regolith vary between about 1500- 1900 kg/m3. Apollo soils are poorly sorted (well graded), and follow a power law size distribution with average grain diameters of 90-110 μm if highlands soils at the Apollo 16 site are representative (basaltic soils at other landing sites had 40-60 μm average grain diameters). Some people stress almost nothing is known about the physical properties of material within PSRs, and treat the PSRs themselves as a kind of mysterious twilight zone. In fact, the Apollo astronauts encountered mini PSRs beneath large boulders, and noted nothing strange or unique about the soil within them. Recent studies found locations where boulders rolled from non-shadowed areas into PSRs, and found only a small difference in the bearing capacity of the soils. It is true that no rovers or landers have investigated the polar PSRs, but a reasonable geologic inference is that their materials are more or less the same as regolith in other places on the Moon, just colder with more water ice and other volatiles.

In-Situ Resource Utilization (ISRU) studies have created simulated lunar ices for experiments involving drilling and water extraction, but unfortunately these efforts may give wrong impressions about PSR material properties. These experiments mix dry lunar regolith simulant with liquid water then freeze the mixtures. As expected, this leads to hard ice- cemented materials similar to permafrost. Some studies quote compressive strengths over 100 MPa, and these results are used to argue that deposits in the PSRs cannot be excavated with mechanical diggers. These experiments do not reflect the geologic processes that have formed and modified polar volatiles: the same impact cratering that has turned solid basalt into a meters-thick powder on the Moon has been pummeling the polar regions for billions of years, and hard cemented materials are unlikely to have survived intact. Mechanical excavation is probably just as possible at the poles as it is in other places on the Moon.


Operating conditions in the challenging lunar environment can be divided into three categories for simplicity: (1) terrain, (2) power, and (3) communications.

1. Terrain includes topographic slopes, roughness, and rockiness. These are important because most lunar mining activities will be carried out by semi-autonomous to fully autonomous mobile vehicles, and more benign terrains will increase the efficiency of mining operations. Planetary rovers generally cannot operate on slopes steeper than 20-30 degrees, and must travel between large boulders and small craters where they are present (Figure 5). Areas of major terrain hazards can already be identified using existing orbital datasets, and supplemented using rover-based cameras and LiDAR systems. In general, the rims and ejecta of younger craters present the greatest hazards in terms of slopes and boulders.

2. Access to power will probably be a major limiting factor for initial mining operations (see architectures below). Near the poles there are peaks and crater rims with nearly constant sunlight: conventional thinking is to land and establish a solar-powered base in the sunlit areas while conducting mining operations nearby in shadowed crater floors. However, distances between well-illuminated areas and crater floors are often 5-10 km or more, and involve descents down crater walls with >40 degree slopes. Power beaming using solar reflectors is only effective within a few km, and using power cables would require large landed masses for operations. Also, there are only a small handful of large PSRs located near peaks of high illumination, and available data suggest those like Shackleton crater are not favorable for ice accumulation.

3. High-bandwidth communication to Earth will be important for mining operations, especially in early phases before large lunar bases are established. Existing mapping products are available that show the fraction of time Earth is visible from different locations at the lunar poles. Locations on the near side of the Moon are more favorable, but ground or satellite relays will probably be needed to provide connectivity to mine sites within PSRs. This is doubly true for far- side locations. Below, I describe four possible architectures for a mining outpost at the lunar poles, mostly focused on the source of power. Each has pros and cons, and is highly dependent on location. I provide maps for each highlighting where the most promising ice deposits are likely located based on Figure 4.

Architecture A: Buried Treasure
As Figure 4 shows, there are large areas of highly favorable ice located outside the PSRs themselves. Figure 6 shows these areas in isolation. The ice here is likely buried by moderately thick dry layers (10s of cm), but the operational challenges (temperature, solar power, etc.) are much more benign than working in a PSR itself. A non-PSR architecture may have the lowest barrier to entry, but does pass up on richer deposits found in the cold shadows.

Architecture B: Crescent Craters
Some craters such as Amundsen at the south pole have a flat floor that is divided cleanly into permanently shadowed and not permanently shadowed areas. A mining architecture could make use of this fact and place the power generation and main base in the partially lit area with the mining operations nearby in the shadowed areas. This avoids some of the problems with traversing steep crater walls, but the areas where this architecture works are limited (Figure 7), and the fractional time of illumination on the partially-lit floors is not as high as on peaks and crater rims. Extensive power storage would be necessary, otherwise mining operations would be limited by the lunar night.

Architecture C: Sea to Sky
This is the most commonly assumed architecture as described above, involving a well-lit area on a crater rim or peak next to a shadowed crater floor. Figure 8 shows that there are actually very few areas of >70% illumination close to favorable ice locations, and the majority are around Shackleton, a young impact crater that has not had time to accumulate deep ice deposits. The main challenges in this architecture involve moving power, equipment and materials between the raised rim/peak and the crater floor. Some kind of funicular could be built to accomplish this, although at high mass and construction difficulty.

Architecture D: The Nuclear Option
A mining architecture based on fission reactors opens up all the PSRs and surrounding areas and is therefore the most flexible option. Figure 9 shows favorable ice deposits that are not amenable for architectures A-C, and are thus best suited for nuclear. However, current tech such as NASA’s kilopower project only produces about 10 kW of electrical power, whereas megawatts will be needed for an industrial-scale mining outpost. The company or country who develops MW-scale reactors for the Moon will be at a huge advantage compared to those depending on solar alone.


Deposits at the lunar poles are probably not pure water ice. The LCROSS experiment detected a variety of contaminants including hydrogen sulfide, methane, and mercury. Many of these compounds are poisonous in the gas or liquid phase, and processing methods will be needed to filter out a variety of metallic and organic materials to purify the ice. The Paragon Space Development Corporation has developed end- to-end technology for such purification. However, prospecting efforts could also target locations where average temperatures are above the stability temperature of most carbon- and sulfur-bearing ices (~54 K) but still below the stability temperature of water ice (~110 K). Elemental mercury and elemental sulfur will remain an issue between 54-110 K.


Exploring lunar ices at the poles will never be complete from a scientific point of view. But from a mining perspective, the amount of knowledge necessary to start a lunar mine can be quantified based on the risk tolerance of an organization. Have we reached the point where such decisions can be made? Surface ice has been directly detected within multiple PSRs and has been mapped at 240 m resolution (Figure 3). However, it is not known whether this represents a very thin frost or diffuse ice that extends deeper. Ice that likely extends deeper than the shallow surface has been directly detected by the LCROSS experiment at Cabeus crater (~6% ice) and is inferred at many other PSRs (~1-5% ice) from neutron spectroscopy data. The maps in Figure 4 are enough to narrow down sites for future prospecting with higher-resolution orbital instruments and with landed spacecraft. For the highly risk tolerant, these maps combined with a single drilling/trenching campaign may be sufficient to locate the first mining outpost on the Moon.


Much thanks to Ariel Deutsch for fruitful discussions and for reviewing the content in this guide, and to Matt Siegler and Shuai Li for providing data products.

References & further reading

Crash course on lunar ice
Arnold, J. R. (1979), Ice in the lunar polar regions. JGR 84, 5659.
Lawrence, D. J. (2017), A tale of two poles: Toward understanding the presence, distribution, and origin of volatiles at the polar regions of the Moon and Mercury. JGR Planets 122, 21.
Siegler, M. A. et al. (2016), Lunar true polar wander inferred from polar hydrogen. Nature 531, 480.
Watson, K. et al. (1961), On the possible presence of ice on the Moon. JGR 66, 1598.

Ice detections
Colaprete, A. et al. (2010), Detection of Water in the LCROSS Ejecta Plume. Science 330, 463.
Feldman, W. C. et al. (2000), Fluxes of Fast and Epithermal Neutrons from Lunar Prospector: Evidence for Water Ice at the Lunar Poles. Science 281, 1496.
Fisher, E. A. et al. (2017), Evidence for surface water ice in the lunar polar regions using reflectance measurements from the Lunar Orbiter Laser Altimeter and temperature measurements from the Diviner Lunar Radiometer Experiment. Icarus 292, 74.
Hayne, P. O. et al. (2015), Evidence for exposed water ice in the Moon’s south polar regions from Lunar Reconnaissance Orbiter ultraviolet albedo and temperature measurements. Icarus 255, 58.
Li, S. et al. (2018), Direct evidence of surface exposed water ice in the lunar polar regions. PNAS 115, 8907.
Miller, R. S. et al. (2014), Identification of surface hydrogen enhancements within the Moon’s Shackleton crater. Icarus 233, 229.
Teodoro, L. F. A. et al. (2010), Spatial distribution of lunar polar hydrogen deposits after KAGUY A (SELENE). GRL 37, L12201.
Yang, F. et al. (2019), Study of Chang’E-2 Microwave Radiometer Data in the Lunar Polar Region. Hindawi Advances in Astronomy 2019, 3940837.

Modeling papers
Mazarico, E. et al. (2011), Illumination conditions of the lunar polar regions using LOLA topography. Icarus 211, 1066.
Schorghofer, N., and O. Aharonson (2014), The lunar thermal ice pump. ApJ 788, 169.

Recent Publications
Farrell, W. M. et al. (2019), The Young Age of the LAMP-observed Frost in Lunar Polar Cold Traps. Geophysical Research Letters, in press.
Kloos, J. L. et al. (2019), The Temporal and Geographic Extent of Seasonal Cold Trapping on the Moon. JGR Planets, in press.
Qiao, L. et al. (2019), Analyses of Lunar Orbiter Laster Altimeter 1,064-nm Albedo in Permanently Shadowed Regions of Polar Crater Flat Floors: Implications for Surface Water Ice Occurrence and Future In Situ Exploration. Earth and Space Science 6, 467.
Rubanenko, L. et al. (2019), Thick ice deposits in shallow simple craters on the Moon and Mercury. Nature Geoscience 12, 597.

Fa, W., and V. R. Eke (2018), Unravelling the Mystery of Lunar Anomalous Craters Using Radar and Infrared Observations. JGR Planets 123, 2119.
Lamelin, M. et al. (2014), High-priority lunar landing sites for in situ and sample return studies of polar volatiles. Planetary and Space Science 101, 149.
Neish, C. D. et al. (2011), The nature of lunar volatiles as revealed by Mini-RF observations of the LCROSS impact site. JGR 116, E01005.
Teodoro, L. F. A. et al. (2014), How well do we know the polar hydrogen distribution on the Moon? JGR Planets 119, 574.
Thomson, B. J. et al. (2012), An upper limit for ice in Shackleton crater as revealed by LRO Mini-RF orbital radar. GRL 39, L14201.