Assessing the Sampleability of Bennu's Surface for the OSIRIS-REx Asteroid Sample Return Mission

Abstract

NASA's first asteroid sample return mission, OSIRIS-REx, collected a sample from the surface of near-Earth asteroid Bennu in October 2020 and will deliver it to Earth in September 2023. Selecting a sample collection site on Bennu's surface was challenging due to the surprising lack of large ponded deposits of regolith particles exclusively fine enough ( $\le 2\text{cm}$ diameter) to be ingested by the spacecraft's Touch-and-Go Sample Acquisition Mechanism (TAGSAM). Here we describe the Sampleability Map of Bennu, which was constructed to aid in the selection of candidate sampling sites and to estimate the probability of collecting sufficient sample. "Sampleability" is a numeric score that expresses the compatibility of a given area's surface properties with the sampling mechanism. The algorithm that determines sampleability is a best fit functional form to an extensive suite of laboratory testing outcomes tracking the TAGSAM performance as a function of four observable properties of the target asteroid. The algorithm and testing were designed to measure and subsequently predict TAGSAM collection amounts as a function of the minimum particle size, maximum particle size, particle size frequency distribution, and the tilt of the TAGSAM head off the surface. The sampleability algorithm operated at two general scales, consistent with the resolution and coverage of data collected during the mission. The first scale was global and evaluated nearly the full surface. Due to Bennu's unexpected boulder coverage and lack of ponded regolith deposits, the global sampleability efforts relied heavily on additional strategies to find and characterize regions of interest based on quantifying and avoiding areas heavily covered by material too large to be collected. The second scale was site-specific and used higher-resolution data to predict collected mass at a given contact location. The rigorous sampleability assessments gave the mission confidence to select the best possible sample collection site and directly enabled successful collection of hundreds of grams of material.

Keywords: Asteroid exploration; Bennu; Landing site selection; Spacecraft safety; Surface topography.

Fig. 1

(top) The TAGSAM sample collection device, with its key dimensions indicated, showing the 32-cm outer diameter, the 21-cm inner diameter (orifice), and the 3-cm gap, behind which is a mylar flap to trap particles that pass through (figure adapted from DellaGiustina et al. 2018). The annulus between the 32-cm outer diameter and the 21-cm inner orifice is used later in calculations of tilt induced by tall boulders on the surface of Bennu (Sect. 3.4). The contact pads are visible along the outer circumference of the baseplate. (bottom) Cross-section of the TAGSAM head showing the pathway for gas to be forced down around the head, and escapes from the bottom. Material redirected into the TAG head passes through the mylar flaps (indicated in red) to be captured in the outer annular ring. (Figure adapted from Lauretta et al. 2017)

Fig. 2

An example of a Bennu DTM that has 49,152 facets. This example shows the triangular facets on top of a shaded surface

Fig. 3

OCAMS image of the Nightingale site taken during Recon A with average pixel scale of 0.01 m/pixel (left), overlain with mapping of the resolvable particles’ longest visible axis (right). Image is adapted from Burke et al. (2021)

Fig. 4

The best-fit line (blue, thin) to data from basalt and wooden spheres (points not shown) serves as the nominal sampleability function, providing a predicted collected volume as a function of the reduced quantity of ${(D_{min}\ast D_{max})}^{(\text{PSFD slope})}$, where $D_{min}$ and $D_{max}$ are minimum and maximum grain size diameter. The uppermost thick red is the fit for the “high mobility” scenario, whereas the best fit (blue) is considered “low mobility”, and the lower-bounding fit (lowest mobility, lower red line) was never deployed in flight because the observed surface properties of Bennu indicated that it was not realistic

Fig. 5

(Left) The collection efficiency function as applied when the TAGSAM head is tilted by a rock on the surface, where the efficiency starts to drop immediately from 1.0. (Middle) Collection efficiency as a function of tilt from the orientation of the TAGSAM head with respect to local asteroid terrain, where the compliance of TAGSAM allows the first ${14}^{\circ }$ of tilt with no change in efficiency. (Right) The collection efficiency as a function of the fractional exposed area of the TAGSAM head due to flat obstructions

Fig. 6

Facet tilt (left) only applies if the tilt relative to the topography exceeds the ${14}^{\circ }$ that the TAGSAM head can accommodate, whereas rock tilt (right) is strictly a function of the height of the tilting particle

Fig. 7

The measured height over length of particles at ROIs DL15. The average value was 0.42

Fig. 8

An example of the rock tilt efficiency values at the Nightingale sample collection site. Colors indicating the calculated efficiency are overlaid on an OCAMS image mosaic of the sample site

Fig. 9

Example of the ArcMap technique applied to site DL11, one of the top 16 candidate sample sites identified at the global survey stage. The red outline on the left shows the region used for calculation. On the right, the red straight lines indicated mapped particles, and the green area is the surface area they covered within the boundaries of the sample site. This site was 49.4 square meters, and 72% of it was covered by unresolved material

Fig. 10

(Left) A cartoon of a particle is overlaid on a regular grid to show how the facet of interest a distance $d$ from the center of the rock with radius $r$ would be tested for masking (where it would be masked if $d<r$ in this case). (Right) Extra distance calculations are needed for the case of an elliptical particle, where distances $d_2$ and $d_3$ are calculated from the facet center to each foci of the ellipse. The outline of the ellipse is $(2\times r)<(d_2\times d_3)$, and a facet center within that boundary would result in it being masked. This example shows square facets for simplicity, whereas the actual DTM utilizes triangular facets, but the math and concept are identical

Fig. 11

Examples of particle masks at Osprey, the ROI that was ultimately selected as the back-up sampling site. Masked facets are red and unmasked facets are blue. The mask on the left considers circular particles, and the mask on the right considers elliptical particles. The most notable differences are the shape of the masking around the large light-colored rock located at the top-center of the image

Fig. 12

The measured ellipticity ($b/a$ axis ratio) of particles at ROIs DL06 and EX07 by two different mappers on images obtained during Detailed Survey with pixel scale of $\sim 5\text{cm}$. A total of 315 measurements were made, and the different mappers’ efforts at different sample sites are indicated in the stacked histogram, where mapper ERJ mapped DL06 and EX07 and mapper KJW mapped at DL06

Fig. 13

A particle that is smaller than a characteristic triangular facet does not overlap with the center of any facet and thus could be missed by the standard masking technique. The masking technique for small particles calculates which facet contains the center of this particle by using the geometry of the three vertices (red points) that define the boundaries of the facet

Fig. 14

The progression of key data products (blue) and sampleability assessments (black) over the course of the OSIRIS-REx mission

Fig. 15

The top 16 ROIs shown on the global basemap of Bennu (Bennett et al. 2021). For scale, the small crater of DL06 is 20 m in diameter, ${10}^{\circ }$ of longitude at $0^{\circ }$ latitude is $\sim 43\text{m}$, and the yellow circles in the bottom panes are 10 m in diameter

Fig. 16

Candidate site Sandpiper with a mosaic colored by the tilt-weighted unresolved material score. An unresolved and low-tilt facet will have an efficiency score of 1, and a resolved or high-tilt facet will have a score of 0. The regions surrounding masked particles show the most common areas that are not masked by a particle but have intermediate scores due to tilt values being elevated by the edges of rocks that were not perfectly masked

Fig. 17

The PSFD power-law slope for each facet for the primary (Nightingale, left) and backup (Osprey, right) sample sites, calculated by tabulating all particles within a 1.5 m radius from each facet to a completeness limit of 4 cm. Spatial scales are not identical, as Nightingale is a much larger site; the radius of the color overlay is 4.23 m for Nightingale and 3.017 m for Osprey

Fig. 18

The rock tilt efficiency values for each facet at the (Nightingale, left) and backup (Osprey, right) backup sample sites

Fig. 19

The distribution of facets with particles larger than 2 cm, smaller than 2 cm, or with no mapped material (unresolved) for Osprey (top) and Nightingale (bottom). Figure adapted from Cambioni et al. (2021)

Fig. 20

The trend of the percentage of unresolved facets at Osprey (red) and Nightingale (blue) as a function of the particle size completeness limit used in the calculation. The data points at 16 cm were from the global analysis, where the two sites had similar values, but as the completeness limit decreased in site-specific analyses (Recon A and Recon C), Nightingale maintained a much higher fraction of unresolved facets