Geology of the InSight landing site on Mars

The Interior Exploration using Seismic Investigations, Geodesy and Heat Transport (InSight) spacecraft landed successfully on Mars and imaged the surface to characterize the surficial geology. Here we report on the geology and subsurface structure of the landing site to aid in situ geophysical investigations. InSight landed in a degraded impact crater in Elysium Planitia on a smooth sandy, granule- and pebble-rich surface with few rocks. Superposed impact craters are common and eolian bedforms are sparse. During landing, pulsed retrorockets modified the surface to reveal a near surface stratigraphy of surficial dust, over thin unconsolidated sand, underlain by a variable thickness duricrust, with poorly sorted, unconsolidated sand with rocks beneath. Impact, eolian, and mass wasting processes have dominantly modified the surface. Surface observations are consistent with expectations made from remote sensing data prior to landing indicating a surface composed of an impact-fragmented regolith overlying basaltic lava flows.


Supplementary Note 1 Perched versus Partially Buried Rocks
A sense of where exhumation or burial has occurred can be gained by characterizing the exposed cross-section of rocks as a proxy for visually determining whether they are mostly exposed, partially exposed, or mostly buried. This approach assumes the shape of fragments larger than ~5 cm across at Homestead hollow is broadly similar to the mostly equant shape of rocks at Lonar crater that formed into basalt 5 (same as the landing site). A mosaic covering approximately 180 degrees from the north to east to south side of lander in Homestead hollow (from left to right) is shown in Supplementary Figure 8. Colored dots denote the relative distribution of buried (only top exposed with minimal relief, red), embedded (multiple faces exposed with moderate relief, yellow), and perched (nearly complete exposure with portion of base visible, green) rocks in and around the hollow. 1180 rocks were examined within and around the half circumference of the hollow (333 on rim, 847 in the interior). Large rocks are easier to detect on the rim, whereas smaller rocks are easier to detect in the hollow and closer to the lander (and contribute to the greater number of rocks detected within the hollow). Perched rocks represent ~70% of those seen on the rim relative to a combined 30% buried (1%) and embedded (29%) rocks. By contrast, the combined 58% of buried (9%) and embedded (49%) rocks is greater than the 42% perched rocks mapped inside the hollow. Note that rocks in foreground can block those in background, and the view favors detection of small rocks within hollow and hampers detection of perched and buried rocks at a distance (where the base is often obscured by other rocks). Moreover, the viewing angle may preclude detection of buried rocks near and beyond edge of the hollow and probably explains the apparent paucity of buried rocks beyond the near field and especially along the rim. Nevertheless, the large increase in perched rocks relative to embedded and buried rocks along and beyond the hollow rim and the comparable dominance of buried and embedded rocks relative to perched rocks inside the hollow appears real (Supplementary Figure 8).

Supplementary Figure 8. Perched versus buried rocks.
Mosaic covering 180° showing buried, embedded and perched rocks. Colored dots denote the relative distribution of buried (only top exposed with minimal relief, red), embedded (multiple faces exposed with good relief, yellow), and perched (nearly complete exposure with portion of base visible, green) rocks in and around the hollow. The purple line is the edge of the hollow. Portion of IDC Mosaic D_LRGB_0014_RAS030100CYL_R__SCIPANQM1.

Supplementary Note 2 Terrain Strength from Elastic Response to Wind Vortices
Convective vortices (named dust devils when the vortex transports dust particles) are detected as a sharp dip in local pressure in the time series. The decrease in pressure by the convective vortex pulls the elastic ground up during its passage causing the surface to tilt away from the vortex. This leads to a tilt signature on the horizontal component of a seismometer in contact with the ground. The isolated seismic signature of a convective vortex was first detected on Earth in 2015 6  By studying the direction of the ground tilt (measured by SEIS 10 ) at the closest approach (when radial tilt amplitude is largest) it appears that the majority of seismically detected vortices have a closest approach to the east of SEIS and that there is a distinct lack of detections to the west (Supplementary Figure 9).
Wind drag 11 from the vortex winds was considered as a possible source of this azimuth bias. On the horizontal axes, the effect of the wind drag will likely not be the same for a dust devil passing on one side or the other, if there is an a priori dominant rotation direction of the dust-devil convective vortices. There is, however, no such dominant rotation as was hinted at from the early field observations of dust devils in terrestrial deserts 12 and Viking Lander observations on Mars 13,14 . In other words, vortex rotation is equally divided between clockwise and counterclockwise. This is consistent with vorticity being generated at very local scales where the Coriolis force is negligible even for the largest dust-devil-like convective vortices. In addition, at closest approach, the tangential wind velocity is perpendicular to the direction of the tilt and any drag force would not influence the seismic measurement in the tilt direction. Given the position of SEIS to the south of the InSight lander, it is also possible that the lander itself may perturb the vortex trajectories. However, if this effect exists, it should only be important for direct vortex encounters (i.e., when the center of the vortex passes directly over the lander).
Lower strength ground will deform more as a vortex passes, leading to a larger ground deformation 15 . This larger ground deformation will be detected more easily by SEIS. Therefore, our preferred explanation for the observed close-approach azimuth bias is that the ground has lower strength to the east thus increasing the number of seismic detections of vortices on that side. Such an interpretation of lower strength ground to the east is consistent with the infill of Homestead hollow with predominantly sand by eolian activity and the position of the lander (and SEIS) on the western edge of the hollow.

Supplementary Note 3 Soil Cohesion Estimates from Slope Stability Analysis
Cohesion estimates are obtained using slope stability analysis for the pits observed beneath the InSight lander thrusters and near the HP 3 mole. We assume plane strain conditions, that the material is homogeneous, and that a Mohr-Coulomb failure criteria is satisfied along the failure surfaces. Two failure mechanisms are examined: a straight-line failure plane (Supplementary Figure 10a) using the Culmann method 16 and a circular arc failure plane (Supplementary Figure  10b) using stability factor charts 16,17 ; both failure planes pass through the toe of the slope. The Culmann method considers that static equilibrium conditions are satisfied and provides a relationship between the resisting force (i.e., the shear strength of the soil along the failure surface) and the driving force (i.e., the weight W of the region isolated by the failure surface). As a result, the stability of the entire slope can be simply represented by:

Supplementary
where is the bulk density of the soil, g is the gravitational acceleration, and H is the height of the slope, c is the cohesion, is the angle of internal friction, and is the slope inclination angle. The term / is a dimensionless expression called the stability factor and describes an equilibrium criterion for which the slope is marginally stable. For a circular-arc failure plane, we use the chart derived in Moore et al. 17 [ Figure 66] that plots the stability factor as a function of the slope angle for different values of internal friction angle.
Slope inclination angle and height H of the pits underneath the lander are taken from elevation profiles in stereogrammetric digital elevation models 18 (Supplementary Figure 11). Slope stability analysis is conducted on two profiles, which have the largest slope inclination angle ( = 66°, H = 0.048 m; G in Supplementary Figure 6) and the largest slope height ( = 57°, H = 0.073 m; H in Supplementary Figure 11). Cohesion estimates are deduced from the stability factor assuming a range of internal friction angles between 30° and 50° and bulk density between 1200 kg/m 3 and 1600 kg/m 3 . We emphasize that the computed cohesion values are minimum estimates. The results show that a circular-arc failure plane yields higher minimum values of cohesion than a straight-line failure plane, which are typically used for steep slopes. For a circular-arc failure plane, the cohesion must exceed 5 to 24 Pa in order to maintain the slopes observed in the pits underneath the lander thrusters. Figure 11. Topography beneath lander. Digital elevation model of part of the area underneath the lander. Gaps in the map are where stereo correlations are poor due to highly oblique viewing geometry and lander hardware obscurations. Where stereo correlations were calculated but sparse, elevations were interpolated from neighboring points within 3 cm. Profiles were taken in the direction of steepest topographic gradient with average slopes calculated for the steepest sections of each profile; the steepest measured were profiles G and H at 66° and 57°, respectively. The map has elevation postings every 1.2 mm and was created from images D001L0018_598131526EDR_F0606_0010M2 and D001R0018_598131636EDR_F0606_0010M2.

Supplementary
Cohesion estimates are also obtained for the pit that formed around the HP 3 mole. On Sol 240, the flat part of the IDA scoop was used to apply a preload at the edge of the HP 3 pit in an attempt to cause failure of the western wall, as shown in Supplementary Figure 12a. The IDA flight software 19 determined that the force applied by the scoop was Fz = 59 N in the vertical direction and Fr = 40 N in the radial direction, which did not cause slope failure. The force Fr, which acts away from the lander, does not affect the stability and only the vertical force Fz is considered in the analysis. The force P acting on the region isolated by the failure surface due to the scoop preload is given by: where, as depicted in Supplementary Figure 12b, p is the uniform stress exerted by the scoop given by = 8 : : ⁄ and d is the length of the region isolated by the failure surface on which the scoop preload is applied given by = Dcot E FGH I J − cot K, : = 0.071 m is the width of the scoop, and : = 0.092 m is the length of the scoop. Considering that the force P adds to the downward force, a closed-form expression for the cohesion c derived using the Culmann method is: where the weight W of the region isolated by the failure surface is given by: The HP 3 pit is assumed to have a slope inclination angle = 80° and a height H = 0.05 m. Minimum estimates of the cohesion obtained using equation (3) are calculated for a range of internal friction angles between 30° and 50° and bulk densities between 1200 kg/m 3 and 1600 kg/m 3 . The results imply that a minimum cohesion c between 1 and 1.9 kPa is required for the slope to be marginally stable when a force P is applied by the IDA scoop. Slope stability analysis for the HP 3 pit performed without the IDA scoop preload or considering the load due to the right front foot of the HP 3 instrument, immediately adjacent to the pit, yields lower cohesion estimates. H is the height of the slope, is the slope inclination angle, d is the length of the region isolated by the failure surface on which the scoop preload is applied, and the red arrow corresponds to the direction of the movement.

Supplementary Note 4 RAD Measurements and Thermal Inertia
We use brightness surface temperatures acquired by the HP 3 radiometer at 8-14 µm from under the lander deck to derive the thermal inertia of the soil in the far RAD spot, which is least influenced by the lander 20 . First, we average the measurements in hourly bins (1 hour is 1/24 th of a sol) to minimize spread and processing time. Then, we fit the temperatures acquired at local times most diagnostic of thermal inertia (e.g., 12AM-Dawn and 11AM-2PM) with a numerical model's output 21 where the surface albedo and regolith thermal inertia are free parameters. Key inputs include the atmospheric dust opacity as derived from images of the sky 8 , the local slope, slope azimuth, latitude, elevation etc. An example of diurnal temperature fit is presented in Supplementary Figure 13. Figure 13: Thermal inertia from diurnal temperature. Example of diurnal temperature fit, sol 103, albedo = 0.15, thermal inertia = 196 J m -2 K -1 s -1/2 . Error bars correspond to reported instrumental uncertainty. Data acquired near the AM and PM crossover times, i.e., when the temperature is rapidly changing are not used for the fit.

Supplementary
A similar model based on the single column version of the LMD General Circulation Model full physics package as described in Forget et al. 22 and Pottier et al. 23 arrives at similar values of thermal inertia and albedo. Best fits are obtained with a homogeneous regolith characterized by typical thermal inertia values in the range of 160-230 J m -2 K -1 s -1/2 and albedo of 0.13-0.15.
Inertia derivations are generally accurate within ±15% 24,25 , although in the present case, the excellent knowledge of the surface properties and multiple observations at various local times certainly indicate an accuracy closer to 10%. Based on in situ observations 25,26,27 and laboratory measurements 28 , such values are associated with unconsolidated fine to medium sand size material, with typical grain sizes of ~130-350 µm. The volumetric amount of cementing phase can hardly be constrained for such low inertia values, but it must be <1% in volume (or more if not located in the inter-grain regions, i.e., not participating in the thermomechanical properties of the regolith). For comparison, ~1% cement in volume raises thermal inertia values by large factors (i.e., ~600 J m -2 K -1 s -1/2 ) 29,30 . No indication of layering is discernable from the diurnal temperature trends, although very fine (sub mm) or deep layering (sub dm) may be unveiled from future analysis of eclipse or seasonal data trends.