The LCROSS Cratering Experiment

See allHide authors and affiliations

Science  22 Oct 2010:
Vol. 330, Issue 6003, pp. 468-472
DOI: 10.1126/science.1187454


As its detached upper-stage launch vehicle collided with the surface, instruments on the trailing Lunar Crater Observation and Sensing Satellite (LCROSS) Shepherding Spacecraft monitored the impact and ejecta. The faint impact flash in visible wavelengths and thermal signature imaged in the mid-infrared together indicate a low-density surface layer. The evolving spectra reveal not only OH within sunlit ejecta but also other volatile species. As the Shepherding Spacecraft approached the surface, it imaged a 25- to-30-meter–diameter crater and evidence of a high-angle ballistic ejecta plume still in the process of returning to the surface—an evolution attributed to the nature of the impactor.

Prior studies from instruments on the Apollo (1) and Lunar Prospector (2) missions indicated the presence of mobile volatiles on or around the Moon. Multiple spacecraft recently confirmed these observations through direct spectroscopic measurements of OH and H2O (35). In contrast, the Lunar Crater Observation and Sensing Satellite (LCROSS) mission used a kinetic probe (the emptied stage of the Centaur rocket) to excavate H-bearing compounds from a permanently shadowed region (PSR) near the south pole of the Moon. As the Centaur collided with the lunar surface, the trailing Shepherding Spacecraft (SSc) measured the evolution and composition of the resulting ejecta with a series of instruments. A separate contribution specifically examines H-bearing molecular species observed with LCROSS instruments (6). Here, we describe the Centaur collision with implications for the ejected mass, excavation depth, and regolith composition.

The mid-infrared cameras (MIR1 and MIR2) recorded the “first light” in the frame coinciding with the moment of impact that remained visible for the next 10 s (Fig. 1A). At that time, the resolution of the MIR instrument was approximately 1 km/pixel. Consequently, the thermal radiance generated just by the heated crater floor should have covered less than a single pixel at that range; instead, it spanned multiple pixels corresponding to 3 to 4 km in diameter before fading with time.

Fig. 1

(A) Evolution of the impact flash from MIR2 starting at about 1 s before impact (top left) in 2-s increments. (B) Evolution of the impact flash from the NSP-1 spectrometer in flash mode (intensity shown in arbitrary units). The moment of impact occurred 0.3 s before the rise in signal. The first exposure bracketing the impact in the VSP (VSP-1) also captured the flash. The beginning of the second exposure (VSP-2) is also shown. (C) Spectra from the Visible Spectrometer before and after the Centaur impact (all 2-s exposures): ~1.2 s before impact (black); 0.0 to 0.8 s after impact (red); 1.1 to 3.1 s (green); 3.4 to 5.4 s (blue); and 16.3 to 18.3 s (gray). The solar spectrum (taken from above the Earth’s atmosphere) is shown at top. Absorption lines (Fraunhofer lines) at this time result from sunlight reflecting off the ejecta cloud as it rose above the shadows. Spectra are offset vertically for clarity (colors on ordinate correspond to colors in figures). (D) Differences in spectra from the VSP referenced as shown in (A). The first spectrum after impact indicates several emission lines in the UV (arrows). The overall radiance difference increased in the visible wavelength range as ejecta emerged into sunlight after the end of the first exposure (green). The isolated strong line near 375 nm (and probably 601 nm) resulted from a cosmic ray hitting the charge-coupled device of the VSP detector. Dashed lines connect emissions over successive exposures, which are described in more detail in (8).

The Total Luminance Photometer (TLP) was designed to detect a sudden change in brightness over visible wavelengths from a distance of more than 600 km (7), but this signal has not yet been unequivocally identified. Nevertheless, the Near-Infrared Spectrometer (NSP1), covering longer wavelengths, operated in a flash mode at 72 Hz (five wavelength channels plus a dark measurement measured every 13.8 ms); hence, it functioned as a thermal flash detector (Fig. 1B). This instrument recorded a 0.4-s rise in background intensity, followed by a 0.7-s decay. The onset for the rise in radiance in the NSP1, however, was delayed about 0.3 s from the moment of impact, on the basis of our analysis of flight data for the trajectory and topography.

The UV/VIS Spectrometer (VSP) captured both emissions from the impact flash and ejecta rising into sunlight. A large number of weak [but significant (2σ)] emission lines emerged within the first 0.8 s after impact, during the flash mode of the NSP1 instrument. The spectral resolution of the VSP is better than 1 nm, as demonstrated through clear identification of fine structure in the solar spectrum due to scattered light. Although many emission lines have not yet been identified with confidence, possible identifications include CN, NH, NH2, CO2+, and CS (8). The overall radiance levels increased dramatically during the next exposure (1.1 to 3.1 s after impact), indicating the arrival of ejecta into sunlight. Prominent emissions at 598 nm (Na) and a line pair at 328 and 338 nm (possibly Ag) also emerged, along with other species, such as H2S and H2O+.

The Visible Camera (VIS) started its sequence ~8 s after impact, well after ejecta had reached sunlight (Fig. 2A). The ejecta cloud increased from ~4 km (8 s) to ~8 km (20 s) in diameter and remained visible for about 42 s before dropping below the sensitivity threshold of the instrument (Fig. 2B). Both near-infrared cameras (NIR1 and NIR2) also captured the expanding ejecta cloud well after (~8 s) the moment of impact but with less dynamic range than the visible camera, thereby limiting their use for comparing dimensions.

Fig. 2

(A) Image from visible camera (VIS) showing sunlit ejecta about 20 s after impact. Inset shows a close-up with the direction of the Sun and the Earth (indicated by arrows). Asymmetry of the ejecta reflects, in part, the projected shadow over the crater (from the edge of Cabeus) and across the ejecta cloud. Dotted circle represent the fields of view of the visible and near-infrared spectrometers. (B) Evolving diameter of ejecta from the VIS camera as a function of Coordinated Universal Time (UTC) beginning 11:31:XX (XX corresponding to seconds on the abscissa where 11:31:27.093 UTC is the time stamp of the first frame). Error bars indicate the range of values measured from a given image. VIS camera began its sequence ~7.8 s after impact noted on plot. The dash-dotted curve represents the expected diameter for the base of the ring of a nominal ejecta curtain as it falls into the shadows [as predicted by (20, 26)]. Ejecta velocity distribution is scaled to LCROSS dimensions and ballistically projected (assuming a nominal 45° ejection angle) to the sunlight horizon above the impact site.

In the final 10 s before impact, changes in exposure time and pixel gain for the NIR2 camera allowed the lunar surface to be imaged from scattered light off nearby relief. As a result, “shadowed” and “illuminated” areas are opposite to the direction of direct solar illumination (Fig. 3A). These images (higher in spatial resolution than the MIR images) reveal a region around the point of impact that is typical of the lunar surface: an undulating but relatively flat surface with few large craters. The last three frames from NIR2, starting at a range of 11 km above the surface, included a feature that correlates with a small thermally warm region in the MIR. This region is identified as the crater through correlation produced by the Centaur impact and its surrounding ejecta through correlation of telemetry and registration with the hot spot located in the MIR data. Just before the SSc collision, the NIR2 camera also recorded a diffuse disk-like feature that moved through the field of view in four successive images during approach (Fig. 3B). This feature appeared suddenly as a result of the camera exposure and gain settings.

Fig. 3

(A) Centaur crater captured in the NIR2 camera 10 km above the surface. Inset shows close-up of crater (top arrow) with extension to the bottom right (bottom arrow). Because of the NIR2 wavelength range (0.9 to 1.7 μm), brighter areas are either higher in albedo or relatively warmer. Consequently, the crater is interpreted as the darker area in the center (25 to 30 m in diameter) and is either darker or cooler than the surroundings. (B) Diffuse elliptical feature (inset) associated with Centaur impact crater (open circle) in the permanently shadowed floor of Cabeus crater. Scattered sunlight off the sun-facing Cabeus wall (Fig. 1A) produced sufficient light for this NIR2 image (taken 13.75 km above the surface). The Sun-facing side of the diffuse feature, however, is brighter and indicates direct sunlight off an ejecta cloud. Dotted ellipses outline the cloud in prior (small inset below) and following frames (above, to the right). Scale for the inset (far left) is adjusted to 833 m above the surface as viewed from the SSc. Dashed lines and arrows represent surface normals extending from the Centaur impact crater (dot) to the each feature. The location of the crater is not directly below ejecta because of the offset trajectory of the trailing SSc. Filled circle indicates location and field of view of the VSP and NSP spectrometers.

The evolving ejecta cloud and final crater produced by the Centaur impact place constraints on the nature of the surface, the depth of any released volatiles, and context for observations by other instruments. The prolonged spectral radiance in the NSP1 data as well as NIR1 and the MIR images establish that the Centaur impact fully coupled with a particulate surface, not bedrock. An impact into bedrock would have resulted in a sudden rise in intensity coincident with the moment of impact, a much shorter thermal signal in the MIR, a short-lived ejecta cloud, and much less total ejected mass than that recorded by the NSP and VIS instruments.

The absence of an obvious impact flash in the TLP data is attributed to properties of the target. First, the highest-temperature materials occur at the contact between the Centaur and the regolith. In highly porous (>70% porosity) targets, material is compressed and driven downward, partly masking the view from above (911). Second, energy partitioned into visible light is suppressed by phase transitions in porous volatile-rich targets, even for much higher-speed impacts (10). Third, rapid quenching due to the heat capacity of the very cold 37-K regolith (12) would further suppress emitted light in visible wavelengths.

Both MIR cameras, however, detected a large heated region immediately after impact (~1 s) that persisted in successive images. At the distance of the SSc, such a large source region could not be the result of impact-heated crater floor or near-rim materials because both would be much smaller than a single pixel. This heated area is also inconsistent with the gradual thermal decay in the NSP (Fig. 1B). Consequently, the thermal source in the MIR images is attributed to impact-heated, low-angle ballistic ejecta remaining below the solar horizon over the first second (13).

During the first 0.8 s, heat generated by the Centaur impact also resulted in numerous transient emission lines in the VSP spectra. One example is the prompt emission of OH (313 nm) from the breakdown of H2O by thermal dissociation or excited OH desorbed from grain surfaces. Because this line disappears in subsequent exposures, it is attributed to heat created as the Centaur first made contact with the surface. Because of the low impact speed, molecular species that first appear (0.8 s) must have been weakly attached to near-surface regolith grains. Impact-liberated volatiles would rapidly expand away from the nascent ejecta plume.

Numerous other emission lines emerged within the first 0.8 s and increased in strength with time (Fig. 2). These atomic and molecular species are interpreted as coming from the regolith, rather than any residual propellant in the Centaur (14). Such a contribution should have rapidly dispersed below the detection limits. Instead, most NH and NH2 emission lines persist or strengthen as greater amounts of ejecta reach sunlight. This is more consistent with a contribution from a lunar regolith source.

Between 1.1 and 3.1 s after impact, the increasing wavelength-dependent radiance and deepening Fraunhofer absorption lines both indicate sunlight scattered off ejecta above 833 m (Fig. 2). At the same time, strong emission lines attributable to Na and possibly Ag also emerged. The delayed appearance of Na is enigmatic because of its low excitation energy, even at the relatively low speed of the Centaur impact. The Na emission could have arisen from either the target or the Centaur impactor. Na has been identified in the lunar atmosphere (15); consequently, it is an expected constituent of the regolith along with other cold-trapped volatiles. Because the Na appeared later, either the Na atoms were depleted near the surface or a layer of other volatiles partly shielded them from first-contact heating (10). The elevated Na line faded after 5 s, whereas weak Na emission filled the solar absorption line 40 s after impact. This suggests a concentration of volatile Na (and perhaps Ag) near but not on the surface.

Alternatively, the Na could have come from the Centaur. The pair of emission lines at 328 and 338 nm parallels the evolution of the Na line. Because the low impact speed would have induced little to no shock melting, most of the thermal signal could have arisen from frictional shear heating. During penetration, the Centaur surface would have been scoured and initially mixed with the ejected regolith. In this interpretation, the Na and line-pair emissions would be tracing the evolution of near-surface materials. A source of Na or Ag in the Centaur, sufficient to produce emission lines, however, has not been identified.

Most emission lines developed between 1.1 and 3.1 s and strengthened with time (Fig. 2). Several other lines appeared during the next exposure (3.4 and 5.4 s): 405.3, 426.8, and 461.4 nm (8). Although most lines strengthened with time, the strong emission line at 405.3 nm appeared only once. Another emission (OH at 290.4 nm) appeared 3.4 s after impact, became stronger, and then faded by 18.3 s. A line coincident with H2O+ (313 nm) also became prominent after 16.3 s. This evolving pattern illustrates changes in concentrations with depth or a time delay between heating and sublimation once exposed to sunlight (6).

The diffuse elliptical feature moving across four successive NIR2 images near the end of transmission (Fig. 3A) is interpreted as high-angle (>80° launch angle from the horizontal) ballistic ejecta still returning to the surface. The SSc struck the lunar surface 3 km southeast of the Centaur crater. As the SSc approached the surface, perspective would have caused a sunlit ejecta cloud 833 m above the crater to move across the background. Modeling of a suspended disk at the sunlight horizon in the field of view of the NIR2 during the final approach is generally consistent with the motion of this feature.

The apparent evolution of the shape of the feature and its curved path is attributed to the changing perspectives of the sunlit portion of the cloud during SSc approach (Fig. 3A). The changing dynamic range of the imaged field of view controlled when this cloud could first be detected (Fig. 3B, bottom inset). At this time, the SSc was still well above the surface and viewed the ejecta cloud nearly in line with the crater. The sunlit ejecta then formed a nearly circular cloud, ~820 m in diameter. When the crater appeared in the next frame, the impact point, ejecta cloud, and SSc were no longer directly in line. As the SSc approached, the cloud in the next image became fainter, smaller, and more offset. The solar phase angle (the angle between the Sun, ejecta cloud above the surface, and spacecraft) increased with decreasing altitude, and the SSc viewed shadowed ejecta. As a result, the center of the brightest portion of the cloud faded and shifted in the last moments as the SSc moved over the crater.

The dark central area in the last three transmitted NIR2 images is interpreted as the actual crater (about 25 to 30 m across). The darkness may represent an exposed colder substrate (perhaps enhanced by the very low heat capacity of aluminum debris from the Centaur) or represents a lower albedo. The slightly brighter crescent-shaped area indicates either a higher albedo near-rim ejecta or slightly warmed ejecta due to latent heat released during phase changes, such as condensation on the ejecta.

The size of the Centaur impact crater is consistent with predictions, but the evolution of the ejecta requires explanation. Before the encounter, the total mass of ejecta predicted to reach an altitude of 2 km (or higher) ranged from 6 × 103 to >106 kg (16). Moreover, a canonical ejecta curtain would form an expanding annulus as the ejecta returned into shadow (17). Even though the inner ring of the annulus should have been resolvable (Fig. 2B, dash and dotted line), the ejecta annulus never emerged; rather, ballistic ejecta formed a diffuse and relatively symmetric cloud throughout.

Recent laboratory experiments revised limits on the amount of the ejecta reaching sunlight from the Centaur impact. Solid projectiles striking a lunar regolith-like particulate surface at 2.5 km/s (Fig. 4A) initially produce ejection angles of ~30° (18) before evolving into nominal ejection angles of ~45° (19, 20). The emptied Centaur rocket, however, was effectively hollow, with a total density of about 0.03 g/cm3. Impact experiments by means of hollow projectiles into a regolith-like target result in a very different ejection sequence (Fig. 4B). In addition to an early-time low-angle ejecta component (18), a high-angle ejecta component emerged initially at high speeds (>1 km/s), decreasing with time (21). Such a high-angle ejecta component would account for the absence of the TLP signal, the diffuse cloud in the VIS images (Fig. 2), the inferred ejecta cloud in the NIR2 (Fig. 3B), the long-lasting spectral signatures (VSP and NSP), and the elevated radiance throughout nearly the entire approach (6). Telescopic observations of such a cloud would have been challenging because of the low optical depth in that orientation (22).

Fig. 4

Contrast in evolution of high-velocity ejecta caused by experimental impacts by (left) solid and (right) hollow aluminum projectiles at about the same speed as the LCROSS impact (2.5 km/s). Just after the moment of impact (frame 1, right), the hollow projectile (1.27 cm in diameter) temporarily obscured the heated contact surface. This impact generated a near-vertical ejecta plume, in contrast with the solid sphere.

Ejecta still in sunlight for more than 4 min (composing the cloud about 820 m in diameter) represent material ejected at nearly vertical angles at speeds greater than 400 m/s. This component would not follow the standard mass-velocity scaling relations (20, 23). In laboratory experiments, hollow spheres failed to eject tracers placed one projectile diameter below the surface. Consequently, the observed concentrations of volatiles (6) must be considered highly conservative values and would depend on their distribution with depth in the regolith.

In summary, our results show that a volatile-rich and porous regolith resulted in a fainter-than-expected impact flash in the visible spectra but a longer-than-expected flash in the near-infrared. Low-angle incandescent ejecta resulted in a larger-than-expected feature in the mid-infrared. Although the observed ejecta plume in the visible spectra never developed into the expected sunlit ejecta ring, ejecta did remain evident in both the VSP and NSP. This was long after the canonical outward-moving ejecta curtain should have disappeared below the solar horizon. High-angle ballistic ejecta were still returning as the SSc approached the surface and contributed to sunlit volatile species detected in the NSP throughout approach. LCROSS provided a close view of excavated ejecta and volatiles, whereas the Lyman Alpha Mapping Project (LAMP) had a different viewpoint detecting other species (CO, Ca, Mg, and Hg) rising above 7 km (24). The Centaur impact produced a crater about 25 to 30 m across, with about 4000 to 6000 kg of ejecta reaching sunlight. Going by the evolving spectra, the impact released a variety of volatiles besides water, which suggests a variety of possible sources besides solar wind implantation (25).

Supporting Online Material

SOM Text

Figs. S1 to S3

Table S1


References and Notes

  1. The TLP covered a spectral range from 350 to 975 nm with a wide field of view (11°) that would have included sunlit regions. The TLP was designed, however, to detect sudden changes in light (from a 1000 K source) down to a level of about 1 nW at the detector with a signal-to-noise ratio (S/N) > 5. The intensity and evolution of the flash in this unit provides constraints about the nature of the struck surface. Preimpact laboratory impact experiments over a range of impact speeds into particulate targets yielded predictions for luminous efficiencies of 10−5 to 10−6 (ratio of luminous flux to impact energy) for the LCROSS impact speed with initial temperatures of <2500 K (9).
  2. The supporting online material (SOM) text and (6) provide more detailed discussion about the VSP spectra.
  3. As part of the mission design, the Centaur upper stage had been purged for more than 3 months during its orbit around the Earth (6). Mission engineers have estimated that less than 3 kg of propellant [from the O2, H2, and hydrazine (N2H4) tanks] remained with other volatile components in the Centaur (insulating foam and paint), contributing an additional 35 kg (6). The impact speed would not fully vaporize this material, except scoured surfaces directly in contact with the regolith during penetration. Even then, only a small fraction of this component would have been ejected at (or excited to) sufficient speeds to reach sunlight.
  4. Estimates for the ejected mass covered a wide range depending on models and assumptions. Extrapolations from late-stage scaling relations (for 45° ejection angles) predicted that only about one projectile mass (~2000 kg or about 1% of the total ejected mass) would reach an altitude of 2 km (21, 26), which is about an order of magnitude less than estimates from smooth-particle-hydrocode models (20). For standard ejection models (23), one projectile mass is probably an overestimate because such extrapolations do not include the effect of early time-energy losses (compressible target) and the nature of the impactor (hollow). A high-angle plume, however, substantially offsets this reduction. Because energy and momentum partitioning still limits the maximum ejected mass, not much more than two to three projectile masses (4000 to 6000 kg) should have reached sunlight at ~0.8 km. Derivations of the total ejecta mass in sunlight from LCROSS VSP and Lunar Reconnaissance Orbiter (LRO) LAMP observations are consistent with this prediction (6, 26). It also should be noted that scattering in the near infrared was five times less than in the visible, which is about a factor of two times less than expected or modeled (6).
  5. The expanding annulus of ejecta arises from the evolution of ballistic ejecta. With time, ejection velocities systematically decrease, launch positions increase in distance from the point of impact, and ejection angles remain nearly constant. The evolution of ballistic ejecta in a given direction produce an inclined thin sheet called the ejecta curtain that advances across the surface away from the crater rim. In all directions, this curtain resembles an inverted cone that expands with time around the impact point. It represents the locus of individual ejecta ballistic particles at a given time, as revealed in laboratory experiments (19) and models (20).
  6. The VSP radiance flattened as its narrow field of view subsampled the sunlit ejecta cloud (6). After about 180 s, the Centaur crater (and ejecta in the foreground) was no longer directly centered, and the VSP and NSP instruments no longer had the remaining high-angle component in their fields of view (Fig. 3B). Consequently, radiance levels dropped rapidly during final approach. Nevertheless, still-returning ejecta still in sunlight continued to liberate adsorbed volatiles, which detached as an expanding, tenuous vapor cloud. Ejecta and volatiles within this narrow high-angle plume would have been difficult to detect from the Earth. Even though some ejecta did reach sufficient altitudes, the low column density and limited extent (as viewed from the side) made telescopic observations difficult.
  7. Although trapped CO, OH, and H2O in the regolith could come from solar wind processes (35), other sources include recent gas release from the lunar interior (1, 15, 27) and cometary or asteroid impacts. Although the high impact speeds of cometary impacts generate gas expansion speeds exceeding the lunar escape velocity, recent models predict that a small fraction could be retained (28). Such contributions to polar cold traps from infrequent cometary sources require a replenishment rate that exceeds the loss rate caused by the continuous “gardening” by small impacts. It is also possible, however, that a cometary impact (or impacts) recently resupplied the current inventory, which will gradually disappear until the next volatile-rich collision. Two other processes can also temporarily sequester volatiles within the regolith. First, not all volatiles released (or generated) by small cometary impacts escape as expanding vapor; rather, a fraction will be injected into the regolith below the surface before migrating upward. Second, impact-trapped pockets of hydrous phases (concentrations as high as 20 weight percent) occur in both natural and experimental impact glasses (29). Such high concentrations are possible because of the high solubility of glass at high pressures and temperatures, combined with rapid quenching. In both cases, slow diffusion and subsequent impact-gardening would gradually release the entrained volatiles and allow their migration to cold traps.
  8. We thank the LCROSS Project and NASA’s Exploration Systems Mission Directorate (ESMD) and NASA Science Mission Directorate (SMD; Planetary Geology and Geophysics) for support. We are also very grateful to the NASA Ames Vertical Gun Range technical team and the Thermophysics Facilities Branch at NASA Ames for their continued support in experiments. B.H. was supported through a NASA Rhode Island Space Grant fellowship for part of this study.
View Abstract

Navigate This Article