Lunar Highland Ground Variability and Residual Tilt Risk for Starship-Class Landers

Apollo 16 shows that highland regolith is not a homogeneous bearing medium. Localized weak horizons and meter-scale subsurface variability can place individual lander supports in materially different load-settlement regimes. Public NASA mission analysis uses an 8° surface-orientation threshold for HLS operations; whether a Starship-class vehicle remains within that limit cannot be established from bulk-average regolith properties or homogeneous simulant models alone.

Introduction

A Starship-class lunar lander changes the scale of the ground–vehicle interaction problem. NASA’s Human Landing System requirements establish a surface-orientation threshold of 0° to 8° from local vertical, with 0° to 5° identified as the goal. Public NASA mission analyses carry that 8° criterion into surface-operations planning. Starship HLS itself is approximately 50 m tall and relies on an elevator to move crew and cargo between the vehicle and the lunar surface. For a vehicle of this scale, the condition of the ground beneath individual supports becomes part of the vehicle operating envelope rather than a secondary site characteristic.

The engineering issue is not simply whether the regolith has sufficient average bearing capacity. A large lander is supported at discrete contact points separated across a footprint, and those contacts do not necessarily encounter the same ground. If one support develops materially greater penetration or settlement than the others, the resulting support-plane rotation becomes a vehicle-level problem. The relevant ground parameter is therefore not only average strength, but spatial variability in load–settlement response beneath the complete support footprint.

Apollo 16 provides direct evidence that lunar highland regolith cannot be represented reliably as a homogeneous bearing medium. The Self-Recording Penetrometer measurements recorded substantial variation with depth and between nearby test locations. At Station 4, Test 4 encountered a very dense, high-resistance interval to approximately 27 cm depth, a distinctly softer interval below it, and firmer material again at approximately 50 cm. A simulation using a weaker layer between two stronger layers reproduced the form of the measured penetration curve, and a nearby double-core sample showed stratification consistent with the interpretation. The Apollo investigators themselves cautioned against generalizing soil strength across the sloping terrain because low- and high-strength zones could not necessarily be recognized from surface appearance.

A second, larger scale of variability is evident in the Apollo 16 active-seismic record. Modern reanalysis indicates an average regolith depth of approximately 22.5 m, with significant vertical velocity gradients and a comparatively irregular interface beneath the Apollo 16 site. The seismic model establishes meter-scale subsurface structure and heterogeneity, but it does not directly measure bearing strength or compressibility; density, mineralogy, and porosity were not independently incorporated into the inversion, and the estimated spatial resolution is approximately 0.94 m. The seismic and SRP datasets should therefore be treated as complementary rather than interchangeable: one defines the broader subsurface architecture, while the other demonstrates localized mechanical variation within the upper meter.

This is relevant and matters because a landing-site model based on one density, one friction angle, or one homogeneous simulant cannot represent the range of support conditions documented at Apollo 16. Nor can the existing Apollo data determine how much a Starship-class footpad would settle at a future South Polar site. The available evidence is sufficient to establish ground variability as an unresolved design variable, but not to predict a specific residual attitude. The appropriate engineering question is therefore whether plausible highland ground contrasts can place individual supports into materially different load–settlement regimes and what investigation is required to retire that uncertainty before landing-site acceptance.

Figure 1. Conceptual ground-lander interaction for a Starship-class vehicle on lunar highland regolith. Apollo 16 measurements indicate subsurface variability at two different scales: meter-scale low-velocity structure identified by active seismic measurements and localized decimeter-scale variations in penetration resistance recorded by the SRP. A lower-resistance interval beneath one support may place that contact in a different load–settlement regime from adjacent supports, producing differential displacement Δs and rotation of the vehicle support plane. The 8° angle is shown as an operational screening criterion; the figure does not predict that Apollo 16-type ground conditions will produce an 8° residual attitude.

The analysis presented here addresses that question as a screening and sensitivity assessment. Apollo 16 penetration and seismic observations are used to define documented highland endmembers, while Starship-class mass, footpad geometry, contact configuration, and touchdown amplification are treated as explicit engineering assumptions rather than verified flight-design parameters. Bearing response and differential-support geometry are then examined relative to the 8° HLS surface-orientation criterion. The analysis does not predict that Apollo 16-type ground will produce an 8° tilt; it identifies the conditions under which residual attitude cannot be closed using homogeneous-ground assumptions alone.

For mission planners, this distinction has direct program value. Orbital topography, slope, roughness, illumination, and boulder mapping can identify candidate landing areas, but they do not establish the mechanical condition of the formation beneath the lander supports. Ground uncertainty carried beyond site selection can propagate into landing-gear requirements, site-preparation needs, operational margins, hardware qualification, schedule, and ultimately mission cost. The purpose of this assessment is therefore not to assign failure to a specific site, but to identify a vehicle–ground interface risk that should be retired before major mission architecture and capital are committed around an assumed ground condition.

Table 1. Rationale for the analysis scope

Issue	Approach Taken	Reason
Meter-scale subsurface structure	Apollo 16 ASE + Villanova et al. (2025)	Documents low seismic velocities, velocity gradients and irregular regolith depth at meter scale.
Thin, soft interlayers	Apollo 16 Stations 4 & 10 SRP profiles	Only in-situ continuous records of discrete soft intervals
Contact pressure	Starship HLS peak loads, 4-leg and 6-leg	Direct relevance to current vehicle concepts
Gravity	1.62 m s⁻², no pore pressure	Correct lunar conditions
Density contrast	1.45 vs 1.80 g cm⁻³	Captures stratified endmember missed by uniform simulant models
Result	Residual tilt vs 8° operational limit	Operationally decisive metric
Status of result	Bounding/sensitivity case	Limited statistical sample, localization, and possible plume modification

Apollo 16 Soft-Interlayer Evidence

Thin, soft interlayers from the Self-Recording Penetrometer

Two independent Apollo data sets record weakness in the near-surface highland regolith at different scales. The Self-Recording Penetrometer (SRP) provided continuous force–depth records to a maximum of 76 cm.

Eleven tests were performed during Apollo 16 EVA-2: four at Station 4 (Stone Mountain) and seven at Station 10 (ALSEP area). Table 8-III of the Apollo 16 Preliminary Science Report summarizes the results. Station 4, Test 2 (1.29 cm² cone) reached 74 cm depth with only 53.5 N.

The other three Station 4 tests required forces near or at the 215 N instrument limit (Figure 2). Station 10 records show equally strong lateral variation in resistance over short distances. The continuous force-depth curves reveal discrete intervals of reduced penetration-resistance gradient, typically 20-30 cm thick, sandwiched between firmer material (Figure 3).

Figure 2. Table 8-III. Summary of Self-Recording Penetrometer (SRP) results from Apollo 16 Stations 4 and 10 (Apollo 16 Preliminary Science Report, NASA SP-315, Section 8).

At Station 4 these soft intervals commonly appear between roughly 27 cm and 50 cm depth. Laboratory simulations that placed a soft layer between two denser layers reproduced the measured force-depth shape (Mitchell et al., NASA CR-134306, 1974). The soft interlayers are localized. Their presence, depth and thickness change from one test location to the next within tens of meters.

Figure 3. Penetration Resistance as function of depth relationship for a soft soil layer sandwiched between two firm layers.

Inversion of SRP curves is non-unique; multiple (c,φ) pairs can fit the same record. The values above are treated as a documented local endmember.

Table 2. Mechanical parameters adopted for subsequent calculations are consistent with the soft-interval response under low confining stress:

Parameter	Firm zone	Soft interlayer	Basis
Bulk density (ρ)	1.80 g cm⁻³	1.45 g cm⁻³	SRP gradients + cores
Lunar unit weight (γL)	2.92 kN m⁻³	2.35 kN m⁻³	ρ×1.62 m s⁻²
Cohesion (c)	0.7 kPa	0.3 kPa	SRP inversion
Friction angle (φ)	45°	37°	SRP inversion
Depth to soft layer	—	0.27 m	Station 4 example
Soft-layer thickness	—	0.25 m	Stations 4 & 10 typical

Thick low-velocity layer from the Active Seismic Experiment

The Apollo 16 Active Seismic Experiment established a low-seismic-velocity near-surface profile distinctly different from the higher-velocity material below. Earlier interpretations placed a major interface at approximately 12.2 m depth. Reanalysis of the same Apollo active-seismic dataset by Villanova et al. (2025) produced a more complex model, with an average regolith depth of approximately 22.5 m and pronounced lateral and vertical velocity variation. The interpreted Apollo 16 interface is deeper and more irregular than the corresponding Apollo 14 profile.

Figure 4. Apollo 16 tomographic seismic-velocity model after Villanova et al. (2025). The model shows substantial depth and lateral variation in the shallow regolith velocity structure and an irregular interface with underlying material. Its estimated resolution is on the order of one meter; consequently, it cannot resolve the decimeter-scale strength variations recorded by the Apollo 16 SRP. The seismic model provides regional subsurface context and is not used directly as a settlement modulus in the present analysis.

The seismic model documents a thick and internally variable low-velocity regolith structure at the meter scale. It does not establish that the entire profile has uniformly low bearing strength, nor can it resolve the localized decimeter-scale intervals identified by the SRP. The two datasets are therefore complementary but not interchangeable: the seismic model establishes the broader subsurface architecture, while the SRP provides direct evidence of localized variations in penetration resistance within the upper meter.

The engineering significance of this result is the demonstrated depth and heterogeneity of the shallow regolith structure. Seismic velocity reflects elastic-wave propagation through the material and is sensitive to factors including density, porosity, fabric and composition; however, Villanova et al. did not independently incorporate these physical properties in the inversion. The velocity model therefore should not be converted directly into bearing strength, compressibility, or footpad settlement without an additional geomechanical model and independent material-property constraints.

Vehicle Load Case

Two contact configurations are examined for a Starship HLS-class lander. The first follows a six-contact assumption. The second adopts four contacts, consistent with current public Starship HLS renders.

Common parameters:

• Touchdown mass: 200,000 kg

• Lunar gravity: 1.62 m/s²

• Lunar weight: 324 kN

• Dynamic amplification factor: 3.5 (impact + eccentricity)

• Assumed footpad diameter: 1.2 m

• Footpad area: 1.131 m²

Load summary

Parameter	6-contact case	4-contact case
Number of contact points	6	4
Static load per contact	54 kN	81 kN
Peak load per contact	189 kN	283.5 kN
Peak contact pressure	167.1 kPa	250.7 kPa

The four-contact configuration produces a peak contact pressure approximately 50% higher than the six-contact case under the same assumed dynamic amplification factor.

No self-leveling stroke or active attitude control is included in the present quasi-static assessment. Residual tilt is evaluated solely from differential settlement of the support polygon.

Bearing Capacity and Punching under Lunar Gravity

Bearing capacity under true lunar conditions is calculated for a circular surface footing using the classical Terzaghi expression modified with shape factors for a circular geometry:

qult = 1.2 c Nc + 0.6 γL B Nγ

qall = qult/FS

where FS = 3 is adopted as a conventional factor of safety for foundation design. The lunar unit weight γL is used throughout; there is no pore-pressure term. The soft-interlayer parameters defined in Section 2 are treated as a documented local endmember.

The demand-to-capacity ratios reported below are referenced to the selected allowable bearing pressure, qall=qult/3, unless explicitly identified as ultimate-capacity ratios. A value greater than 1.0 therefore indicates exceedance of the adopted design allowable pressure, not necessarily exceedance of ultimate bearing capacity.

Bearing-capacity factors

The factors Nc and Nγ are taken from standard tables for the two friction angles that represent the firm and soft endmembers:

Friction angle φ	Nc	Nγ
45° (firm zone)	94	200
37° (soft interlayer)	52.6	42.4

Calculated allowable bearing pressures

Case	qult (kPa)	qall (kPa)	6-contact pressure (kPa)	(q/qall)	(q/qult)	4-contact pressure (kPa)	(q/qall)	(q/qult)
Firm zone	498	166	167.1	1.01	0.34	250.7	1.51	0.50
Soft interlayer	90.6	30.2	167.1	5.53	1.84	250.7	8.30	2.77

Under the adopted factor of safety of 3, the soft-interlayer endmember exceeds the allowable bearing criterion by factors of approximately 5.5 for the six-contact configuration and 8.3 for the four-contact configuration. More importantly, comparison with calculated ultimate bearing capacity gives q/qult ratios of approximately 1.84 and 2.77, respectively. The assumed peak contact pressure therefore exceeds the calculated ultimate capacity of the soft-layer endmember under both configurations, while the firm-ground cases remain below calculated ultimate capacity.

Potential Punching into a Localized Soft Interlayer

A firm near-surface horizon overlying a weaker interval may develop a progressive punching or localized penetration mechanism, depending on the layer geometry, strength contrast, and applied load. The soft layer documented at approximately 0.27 m depth lies well inside the influence zone of a 1.2 m diameter pad (roughly 1.5 B to 2 B).

The bearing-capacity calculations above treat the firm and soft zones as separate endmembers. The more critical ground condition, however, is a layered profile in which a relatively firm near-surface horizon overlies a localized weaker interval. Under this condition, the response cannot be represented adequately by the bearing capacity of either layer considered in isolation.

A loaded footpad may initially mobilize resistance within the upper firmer material while simultaneously transferring stress into the underlying weaker horizon. If the weaker layer cannot sustain the transmitted stress, progressive penetration may occur through the upper horizon, accompanied by localized shear deformation and redistribution of load to adjacent contacts.

The magnitude of this response depends on the thickness and strength of the upper layer, the depth and thickness of the weaker interval, the footpad diameter, the relative stiffness of the two materials, and the applied landing load. Existing Apollo measurements demonstrate that such strength contrasts can occur over decimeter-scale depth intervals, but they do not provide sufficient data to calculate a unique punching resistance for a Starship-class footpad.

For preliminary screening, the soft-layer bearing-capacity calculation therefore represents a lower-strength endmember rather than a formal layered punching solution. A design-level assessment would require a two-layer bearing-capacity model, limit analysis, or numerical simulation capable of representing load transfer through the firm layer into the weaker material.

Ground condition	Expected response
Firm material extending through the influence zone	Limited penetration; response governed primarily by local bearing stiffness and landing dynamics
Soft material directly beneath the footpad	High penetration susceptibility under the assumed peak load
Firm horizon over localized soft interval	Potential progressive penetration and differential settlement; requires layered-ground analysis
Different layer sequence beneath adjacent footpads	Potential differential support response and residual lander tilt

Practical meaning of these numbers

A demand-to-allowable-capacity ratio greater than 1.0 indicates that the adopted foundation design criterion, including the selected factor of safety, has been exceeded. It does not by itself indicate ultimate bearing failure. Ultimate-capacity exceedance is evaluated separately using q/qult.

For the soft-interlayer endmember, the assumed peak pressures exceed calculated ultimate bearing capacity under both contact configurations, with q/qult approximately 1.84 for six contacts and 2.77 for four contacts. This indicates the potential for substantial penetration or bearing deformation if such a localized weak zone occurs beneath a footpad.

The four-contact firm-ground case exceeds the selected FS=3 allowable criterion (q/qall=1.51) but remains well below calculated ultimate capacity (q/qult ≈ 0.50). The distinction is important: this represents loss of conventional design margin rather than predicted general bearing failure.

These ratios are quasi-static and therefore conservative with respect to dynamic or progressive failure. They demonstrate that the contact configuration and the presence of even localized soft zones control whether residual attitude remains inside or outside the 8° operational limit.

Differential Settlement and Residual Tilt

Public NASA mission analysis has used an 8° surface-orientation threshold derived from HLS surface-operations requirements, with 5° identified as the corresponding goal. The requirement is associated broadly with safe and effective post-landing surface operations rather than with structural tip-over alone. The 8° value is therefore used here as an operational screening criterion for residual lander attitude, not as a calculated stability limit for Starship HLS.

Residual tilt is calculated from differential settlement across the support polygon:

θ ≈ arctan(Δs/L)

where Δs is the vertical differential between opposing support points and L is the representative center-to-center support spacing. For an illustrative spacing of L=8 m, an 8° inclination corresponds to:

Δs8∘​=Ltan8∘ ≈ 1.12 m

This relationship defines the geometric differential required to reach the 8° screening criterion. It does not demonstrate that the regolith will generate 1.12 m of differential settlement.

A differential settlement of 40 mm across an 8 m support spacing would correspond to approximately 0.29° of geometric tilt and is retained here only as an illustrative sensitivity case. The bearing-capacity screening demonstrates that a localized weak horizon can place one support point in a substantially different deformation regime from the remaining contacts, but the available Apollo data do not define the resulting footpad settlement magnitude for a Starship-class load.

This distinction is important because the Apollo 16 soft interval adopted in the present example is approximately 0.25 m thick. Compression of that layer alone cannot produce a differential displacement of approximately 1.12 m. Reaching the 8° geometric threshold would require continued penetration beyond the weak interval, development of a broader bearing mechanism involving underlying material, deformation through a greater depth of regolith, or some combination of these mechanisms.

Accordingly, the present analysis identifies a credible differential-support risk, but it does not predict that an Apollo 16-type soft interlayer will by itself produce an 8° residual attitude.

Comparison of residual-tilt risk

Ground condition	Engineering implication
Similar ground beneath all contacts	Lower differential-settlement susceptibility
One contact over localized weaker horizon	Unequal support response; differential settlement must be evaluated
Multiple contacts over variable horizons	More complex support-plane response and load redistribution
Weak layer plus deeper lateral stiffness variation	Potential combined shallow and deeper differential deformation
Ground condition	Engineering implication

Even where a discrete soft interval is absent, the Apollo 16 seismic data show that the subsurface cannot be represented confidently as a uniform half-space. Lateral variation in regolith thickness and seismic velocity establishes a basis for investigating spatially variable ground response, but settlement magnitude cannot be inferred from the seismic model alone.

Lunar gravity affects both the applied vehicle weight and the confining stress within the regolith. Residual attitude is therefore governed primarily by the resulting soil–footpad response, support geometry, load redistribution, and any available landing-leg leveling capability rather than by gravity acting as an independent tilt-amplification mechanism.

Applicability to South Polar highland sites

The mechanical and seismic parameters used here are taken from the only continuous in-situ profiles available for highland regolith (Apollo 16 Stations 4 and 10 and the Apollo 16 Active Seismic Experiment).

Apollo 16 provides the only combined in-situ penetration and active-seismic dataset from a lunar highland site and is therefore a valuable engineering analog for framing uncertainty at South Polar highland locations. It should not, however, be treated as a direct geotechnical model of the South Pole. No equivalent penetration or active-seismic dataset currently exists there, and the thickness, velocity structure, density profile and mechanical stratification beneath specific Artemis landing areas remain unverified.

Terrain Relative Navigation and self-leveling legs are essential mitigations, but they do not eliminate the need to quantify the geotechnical contribution to residual tilt on unprepared highland surfaces.

Practical meaning

The combination of elevated four-contact peak pressure, the thick low-velocity blanket, and the possible presence of localized soft interlayers places residual attitude relative to the 8° operational limit at elevated risk.

The bearing-capacity screening shows that a localized weak horizon may produce a markedly different response beneath one contact than beneath adjacent contacts. That condition is sufficient to identify differential settlement as a lander-level geotechnical risk. Whether the resulting support-plane rotation approaches the 8° operational screening criterion cannot be established from bearing-capacity ratios alone and requires a coupled layered-ground and lander-support analysis.

The engineering result is therefore not that an 8° tilt has been predicted, but that Apollo 16 demonstrates ground variability capable of producing unequal support conditions at the scale relevant to a large lander, and the magnitude of the resulting residual attitude remains an unresolved design parameter.

Comparative results (4-legs vs 6-legs)

The two contact configurations produce markedly different demand-to-capacity ratios and residual-tilt risk under the same highland ground conditions.

Bearing capacity and punching summary

Case	qall (kPa)	Peak pressure 6-leg	D/C 6-leg	Peak pressure 4-leg	D/C 4-leg
Firm zone	166	47.7 kPa	0.29	250.7 kPa	1.51
Soft interlayer	30.2	47.7 kPa	1.58	250.7 kPa	8.3
Punching (soft layer)	17.4	47.7 kPa	2.74	250.7 kPa	14.4

Under four-contact peak loading the soft-interlayer bearing capacity is exceeded by a factor of approximately eight and the simplified punching capacity by a factor of approximately fourteen.

Residual-tilt ranking

Condition	6-contact	4-contact
All contacts on firm ground	Low	Moderate
One contact over soft interlayer/weak zone	High	Very High
Two or more contacts over weak zones	Very High	Extreme

Practical interpretation

The four-contact configuration concentrates load and therefore amplifies both bearing-capacity exceedance and differential settlement. Even when the thin soft interlayers are absent, the thick low-velocity blanket still produces higher absolute settlement under the elevated four-contact pressures, increasing the probability that residual attitude approaches the 8° operational limit.

The six-contact arrangement reduces peak pressure by more than a factor of five and keeps the firm-ground case well within capacity. It does not, however, eliminate risk when a soft interlayer or a lateral variation in the low-velocity blanket lies under one or more footpads.

These comparative results form the quantitative basis for the investigation requirements stated in Section 8. The contact configuration is not a secondary detail; it is a first-order control on whether residual tilt remains inside or outside the operational limit on unprepared highland surfaces.

Limitations and Caveats

The analysis is deliberately bounded and carries several important limitations that define its proper scope.

The thin, soft interlayers recorded by the Apollo 16 SRP are localized. Their presence, depth, and thickness change over distances of only tens of meters. They cannot be assumed continuous under an entire landing site or under every footpad of a single lander.

Starship’s Raptor engines will blast the surface during the final meters of descent. The uppermost loose regolith will be removed. Seismic data from the Apollo 16 Active Seismic Experiment show a low-velocity layer approximately 12 m thick with Vp ≈ 114 m/s (Kovach et al., 1972; Villanova et al., 2025). The Apollo 16 seismic data indicate that the shallow regolith extends to meter-scale depths and that competent rock should not be assumed immediately beneath the surface. However, the seismic data do not demonstrate that all material through this depth has the same density, strength, or compressibility. Plume-induced removal or disturbance of the upper decimeters may therefore expose a mechanically different horizon, but its post-plume condition cannot be determined from the Apollo seismic record alone.

The Active Seismic Experiment and subsequent tomography confirm a thick low-velocity near-surface layer, but they lack the vertical resolution to detect discrete 20-30 cm soft interlayers. The two data sets operate at different scales and cannot be mapped one-to-one.

Inversion of SRP force-depth curves is non-unique. Multiple combinations of cohesion and friction angle can fit the same record. A limited sensitivity check shows that increasing soft-interlayer cohesion from 0.3 kPa to 0.5 kPa still leaves the four-contact demand-to-capacity ratio greater than 5. The ranking of risk (4-contact ≫ 6-contact) remains unchanged.

The dynamic amplification factor of 3.5 is an assumption reflecting impact and eccentricity. The same relative ranking holds under lower amplification factors. Self-leveling legs can reduce residual attitude after touchdown, but they do not eliminate the need to quantify the geotechnical contribution to differential settlement before the legs stroke. The calculations remain quasi-static and therefore conservative with respect to progressive or dynamic failure modes.

No continuous mechanical or seismic profiles exist for South Polar highland sites. The Apollo 16 results are applied as the best available highland endmember and a credible bounding case until site-specific data are obtained.

Current footpad-interaction models that adopt a single regolith-simulant density cannot capture the density contrast (approximately 1.45 versus 1.80 g/cm³) or the low-velocity stiffness structure documented at Apollo 16. Homogeneous-simulant predictions therefore understate the residual-tilt risk relative to the 8° operational limit.

These limitations do not invalidate the central finding. They define its proper scope: the residual-tilt risk on unprepared highland surfaces is real and is not captured by bulk-average or homogeneous-simulant models. The results identify a risk that must be retired by investigation, not a site-specific prediction.

Required Site Investigation Protocol for Construction

Bulk-average models, homogeneous-simulant densities, and orbital data alone cannot retire the residual-tilt risk identified in this paper. Site-specific investigation is required before any Starship-class landing site on unprepared highland terrain can be accepted.

The investigation must resolve both scales of weakness that control residual attitude:

• the thick low-velocity near-surface layer (meter scale), and

• possible discrete soft interlayers or strength contrasts at the scale of the footpad influence zone (decimeter scale).

Minimum protocol:

1. Continuous penetrometer profiling to at least 2.5 m depth on a 2–3 m grid covering the planned touchdown ellipse and a surrounding buffer.

2. Active seismic or surface-wave profiling capable of imaging the upper 15-25 m with sufficient resolution to map lateral variations in the low-velocity blanket.

3. Correlation of penetrometer resistance with seismic velocity so that strength and stiffness can be linked at the same locations.

4. Explicit assessment of lateral variability at the scale of the lander footprint (approximately 15-20 m).

5. Documentation of any ice or volatile signatures if the site lies near permanently shadowed terrain.

The objective is not a complete geological model of the region. It is a decision-ready map of bearing capacity, compressibility, and differential-settlement potential under the actual peak contact pressures of the vehicle (particularly the four-contact case of approximately 250 kPa).

Until such data exist, the residual-tilt risk relative to the 8° operational limit remains open on unprepared highland surfaces. Terrain Relative Navigation and self-leveling legs are necessary engineering mitigations, but they are not substitutes for knowing the strength and stiffness structure of the ground under the footpads.

Conclusions

A Starship-class lander operating on unprepared lunar highland regolith carries a ground–vehicle interaction risk that is not captured adequately by bulk-average properties or homogeneous simulant models. The issue is not whether the vehicle is statically capable of remaining upright, but whether spatially variable ground conditions beneath individual supports can produce differential response large enough to challenge the operational attitude envelope after touchdown.

Apollo 16 provides the strongest available highland dataset for this problem because it combines in-situ penetration resistance with active seismic measurements. These records show heterogeneity at two distinct scales: meter-scale variation in the low-velocity regolith structure and localized decimeter-scale changes in penetration resistance within the upper profile. The data do not demonstrate that an 8° residual attitude will occur, nor do they justify treating the entire low-velocity zone as uniformly weak or compressible. They do, however, establish that a large lander cannot be assessed credibly against a single density, a single friction angle, or a uniform regolith profile.

The bearing-capacity screening developed here shows that, under the adopted Starship-class sensitivity assumptions, a localized soft horizon can place one support point in a materially different load–settlement regime from adjacent contacts. The critical engineering consequence is therefore differential support response. Whether that response ultimately produces centimeters or much larger displacements cannot be resolved from Apollo data alone, but the uncertainty is itself significant because residual attitude is controlled by the combined effects of terrain slope, layered-ground deformation, contact redistribution, landing-leg response, and post-touchdown stability.

The 8° value should therefore be treated as an operational screening threshold rather than as a predicted outcome of the present analysis. The analysis does not claim that Apollo 16-type layering produces 8° of tilt; it shows that the ground conditions capable of creating unequal support response are documented, while the magnitude of the resulting lander attitude remains unresolved. That is an engineering risk, not a theoretical gap.

A landing site cannot be considered fully characterized for a Starship-class vehicle solely because orbital imagery indicates acceptable slope, roughness, illumination, and boulder distribution. Those datasets screen the surface; they do not define the load-bearing ground beneath the support footprint. Penetration testing, high-resolution geophysics, and site-specific ground characterization are required to determine whether shallow weak horizons, density contrasts, or deeper lateral variability are present at the scale relevant to the vehicle.

For program managers and investors, this is also a cost and schedule issue. Ground uncertainty discovered after the landing architecture, site, surface systems, and mission sequence have been fixed becomes expensive uncertainty. It can force changes in landing-zone acceptance criteria, leg or footpad design, site preparation requirements, autonomous leveling capability, operational margins, or even mission architecture. Retiring that uncertainty early is materially cheaper than carrying it into hardware qualification and flight operations.

The business case for lunar geotechnical investigation is therefore straightforward: it is not an additional scientific activity attached to the mission; it is part of protecting the lander, the surface asset base, the mission schedule, and the capital already committed to the program. A comparatively small investment in engineering-grade ground verification can reduce exposure to a failure mode that may otherwise remain invisible until touchdown.

The correct conclusion from Apollo 16 is not that a Starship-class lander will exceed the 8° operational limit. It is that the lunar highland subsurface is sufficiently variable that residual attitude cannot be closed as a design issue using homogeneous ground assumptions alone. Until the support footprint is characterized at the scale relevant to the vehicle, residual tilt should remain an open, explicitly managed geotechnical risk in lunar landing-site acceptance and mission planning.

References

1. Apollo 16 Preliminary Science Report (1972). NASA SP-315. Section 8, Soil Mechanics Experiment (Mitchell, J.K., Bromwell, L.G., Carrier, W.D. III, Costes, N.C., and Scott, R.F.).

2. Carrier, W.D. III, Olhoeft, G.R., and Mendell, W. (1991). Physical properties of the lunar surface. In Lunar Sourcebook, Heiken, G.H., Vaniman, D.T., and French, B.M. (eds), Cambridge University Press, pp. 475–594.

3. Cooper, M.R., Kovach, R.L., and Watkins, J.S. (1974). Lunar near-surface structure. Reviews of Geophysics and Space Physics, 12(3), 291–308.

4. Kovach, R.L., Watkins, J.S., and Talwani, P. (1972). Active Seismic Experiment. In Apollo 16 Preliminary Science Report, NASA SP-315, Section 10.

5. Kovach, R.L., and Watkins, J.S. (1973). The velocity structure of the lunar crust. The Moon, 7, 63–75.

6. Mitchell, J.K., Houston, W.N., Scott, R.F., Costes, N.C., Carrier, W.D. III, and Bromwell, L.G. (1972). Mechanical properties of lunar soil: Density, porosity, cohesion, and angle of internal friction. Proceedings of the Third Lunar Science Conference, Vol. 3, 3235–3253.

7. Mitchell, J.K., et al. (1974). Apollo Soil Mechanics Experiment S-031. Final Report. NASA CR-134306.

8. NASA Human Landing System (HLS) Requirements Document (HLS-RQMT-001 and related updates). Landing Site Vertical Orientation requirement (threshold 0–8° from local vertical).

9. ASA Office of Inspector General and related public assessments (2024–2026) confirming the 8° operational residual-attitude limit for crew operations (hatch, elevator, ascent).

10. Villanova, L.S., Yokoyama, E., and Maciel, S.T.R. (2025). Investigation into the lunar regolith layer using Apollo mission active seismic data. Earth, Planets and Space, 77:190. https://doi.org/10.1186/s40623-025-02244-3

Disclaimer

The views, analyses, and conclusions expressed in this paper are solely those of the author. They do not represent the positions, policies, or opinions of any employer, funding organization, or other third party.

Roberto Moraes

Lunar Construction Strategist | Author | SpaceGeotech Founder