Accepted Manuscript
Detection and Characterization of Buried Lunar Craters with GRAIL Data Rohan Sood, Loic Chappaz, Henry J. Melosh, Kathleen C. Howell, Colleen Milbury, David M. Blair, Maria T. Zuber PII: DOI: Reference:
S0019-1035(17)30158-6 10.1016/j.icarus.2017.02.013 YICAR 12374
To appear in:
Icarus
Received date: Revised date: Accepted date:
29 February 2016 7 February 2017 9 February 2017
Please cite this article as: Rohan Sood, Loic Chappaz, Henry J. Melosh, Kathleen C. Howell, Colleen Milbury, David M. Blair, Maria T. Zuber, Detection and Characterization of Buried Lunar Craters with GRAIL Data, Icarus (2017), doi: 10.1016/j.icarus.2017.02.013
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Highlights
2
• Gravity mapping observations from NASA’s GRAIL mission are employed
3
to detect, characterize and validate the presence of large impact craters
4
buried beneath the lunar maria. • Gravity gradiometry detection strategy is applied to both free-air and Bouguer
6
gravity field to identify gravitational signatures that are similar to those ob-
7
served over buried craters.
CR IP T
5
8
• The presence of buried craters is further supported by individual analysis
9
of regional free-air gravity anomalies, Bouguer gravity anomaly maps, and
11
• Other large, still unrecognized, craters undoubtedly underlie other portions of the Moon’s vast mare lavas.
AC
CE
PT
ED
M
12
forward modeling.
AN US
10
1
ACCEPTED MANUSCRIPT
Detection and Characterization
13
14
of Buried Lunar Craters with GRAIL Data
Rohan Sood a,b , Loic Chappaz a , Henry J. Melosh c ,
16
Kathleen C. Howell a , Colleen Milbury c , David M. Blair d , and
17
Maria T. Zuber e
Purdue University, West Lafayette, IN 47907 (U.S.A.)
19
20
of Aeronautics and Astronautics,
AN US
a School
18
CR IP T
15
b Current
affiliation: Department of Aerospace Engineering and Mechanics,
21
The University of Alabama, Tuscaloosa, AL 35487 (U.S.A.)
22
c Department
d Haystack
24
M
PT
e Department
26
of Earth, Atmospheric and Planetary Sciences,
CE
Massachusetts Institute of Technology, Cambridge, MA 02139-4307 (U.S.A.) c 2015 Rohan Sood, Loic Chappaz, Henry J. Melosh, Copyright
Kathleen C. Howell, Colleen Milbury, David M. Blair, and Maria T. Zuber
AC
29
Observatory, Massachusetts Institute of Technology, Westford, MA 01886-1299 (U.S.A.)
25
28
ED
Purdue University, West Lafayette, IN 47907 (U.S.A.)
23
27
of Earth, Atmospheric, and Planetary Sciences,
30
Number of pages: 46
31
Number of tables: 4
32
Number of figures: 31
Preprint submitted to Icarus
23 February 2017
33
Proposed Running Head:
34
Buried Lunar Craters.
35
Please send Editorial Correspondence to:
36
Rohan Sood
38
Department of Aerospace Engineering and Mechanics
39
The University of Alabama
40
238 Hardaway Hall
41
401 7th Ave., Box 870280
42
Tuscaloosa, AL 35487, USA
Email:
[email protected]
45
Phone: (765) 409-2871
ED
44
M
43
AN US
37
AC
CE
PT
46
CR IP T
ACCEPTED MANUSCRIPT
3
ACCEPTED MANUSCRIPT ABSTRACT
48
We used gravity mapping observations from NASA’s Gravity Recovery and
49
Interior Laboratory (GRAIL) to detect, characterize and validate the pres-
50
ence of large impact craters buried beneath the lunar maria. In this paper
51
we focus on two prominent anomalies detected in the GRAIL data using the
52
gravity gradiometry technique. Our detection strategy is applied to both free-
53
air and Bouguer gravity field observations to identify gravitational signatures
54
that are similar to those observed over buried craters. The presence of buried
55
craters is further supported by individual analysis of regional free-air gravity
56
anomalies, Bouguer gravity anomaly maps, and forward modeling. Our best
57
candidate, for which we propose the informal name of Earhart Crater, is ap-
58
proximately 200 km in diameter and forms part of the northwestern rim of
59
Lacus Somniorum, The other candidate, for which we propose the informal
60
name of Ashoka Anomaly, is approximately 160 km in diameter and lies com-
61
pletely buried beneath Mare Tranquillitatis. Other large, still unrecognized,
62
craters undoubtedly underlie other portions of the Moon’s vast mare lavas.
AC
CE
PT
ED
M
AN US
CR IP T
47
63
Keywords: MOON; MOON, INTERIOR; MOON, SURFACE; CRATERING
64
4
ACCEPTED MANUSCRIPT 1
Introduction
66
NASA’s Discovery Program has supported a series of low-cost scientific and
67
exploration missions. Among these, the Gravity Recovery and Interior Lab-
68
oratory (GRAIL) mission focused on high quality mapping of the Moon’s
69
gravitational field. Launched in 2011, the sister spacecraft, Ebb and Flow,
70
entered lunar orbit on December 31, 2011 and January 1, 2012, respectively,
71
after a 3.5 month low-energy cruise phase. GRAIL’s Primary Mission lasted
72
for a period of 88 days during which the two spacecraft mapped the lunar
73
gravity (Zuber et al., 2012) from an average altitude of 55 km. The technique,
74
satellite-to-satellite tracking, employed for mapping the Moon is similar to
75
that used by Gravity Recovery and Climate Experiment (GRACE) to map
76
the Earth’s gravitational field. GRAIL’s Extended Mission began on August
77
8, 2011 during which the altitude of the dual spacecraft was lowered to ap-
78
proximately 23 km and additional gravitational data was collected. A final
79
phase at still lower altitude mapped the Orientale basin at particularly high
80
resolution before the two spacecraft exhausted their remaining propellant and
81
on December 17, 2012, were steered to an impact on the lunar surface.
82
One of the overarching objectives of the investigation reported in this paper is
83
the detection, characterization and validation of the existence of buried craters
84
within lunar maria. These craters form a subset from among numerous other
AC
CE
PT
ED
M
AN US
CR IP T
65
85
types of subsurface features on the Moon. Although the surface topography is
86
easily destroyed by later impact events, the configuration of the crust-mantle
87
interface is much less easily altered and thus will survive even if the surface
88
has been modified beyond easy recognition. The quasi-circular anomalies de-
89
tected are consistent with the work carried out by Neumann et al. (2015) 5
ACCEPTED MANUSCRIPT and Evans et al. (2016) that have identified buried craters due to the density
91
contrast between the surrounding crust and mantle material uplifted by the
92
excavation of large impact craters. The origin of the impact plays a critical role
93
in determining the shape of the density contrast along with any subsequent
94
events that may alter the gravitational footprint of the anomaly. In particular,
95
the proposed Earhart crater predated Serenitatis and was almost completely
96
overprinted by the impact that resulted in the formation of Serenitatis that
97
was subsequently followed by volcanism, thus flooding the basin with lava
98
basalts.Previous work by Chappaz et al. (2014) employed a detection strategy
99
based on gravity gradiometry to detect subsurface features such as buried lava
CR IP T
90
tubes.
101
2
102
The gradiometry method has long been used for prospecting for oil, gas and
103
minerals on the Earth (Mueller, 1964). It focuses on determining the second
104
horizontal derivatives of the gravitational potential, which can then be inter-
105
preted to delineate subsurface density variations. Our implementation begins
106
with the spherical harmonic gravity field of the Moon derived from GRAIL
107
data (Zuber et al., 2013). The spherical harmonic data are truncated and ta-
108
pered to a predetermined degree and order to enhance the short wavelength
AN US
100
AC
CE
PT
ED
M
Technique
109
structures of interest. For any scalar field, a widely employed method to detect
110
or highlight ridges or valleys within the field of interest involves the computa-
111
tion of the Hessian matrix, and consequently, the eigenvalues and eigenvectors
112
that are associated with the Hessian of the scalar field. The Hessian of the
113
gravitation potential is defined as the matrix of second partial derivatives of 6
ACCEPTED MANUSCRIPT
114
the potential function with respect to the spherical coordinates of the position
115
vector, i.e.,
Hij =
∂ 2U ∂xi ∂xj
(1)
where xi , xj = (λ, φ, r), and, λ is longitude, φ is latitude and r denotes the
117
radial distance. In essence, the eigenvalue of largest magnitude and the corre-
118
sponding eigenvector are associated with the direction of maximum gradient in
119
the field. In this investigation, similar to the development in Andrews-Hanna
120
et al. (2013), eigenvalue maps that depict the magnitude of the largest mag-
121
nitude eigenvalue for each point on a grid of the lunar surface are produced.
122
Either the free-air potential or the Bouguer potential (which corrects the free-
123
air potential for topography and terrain by assigning a nominal density of 2560
124
kg/m3 , the Moons average crustal density) can be employed in the analysis, de-
125
pending on the objective. For the purpose of this investigation, localized maps
126
that focus on specific regions are most relevant. A negative density anomaly
127
(that creates a trough in the potential) corresponds to a positive eigenvalue on
128
the gradiometry map, or eigenvalue map, because of the additional derivative
129
of the potential function in the computation of the Hessian.
130
This detection strategy, based on maximum eigenvalue, is used in conjunction
131
with traditional free-air and Bouguer gravity maps. Related technique relying
AC
CE
PT
ED
M
AN US
CR IP T
116
132
on anti eigenvalue has also been proposed to detect buried craters (Evans
133
et al., 2016). To further support our interpretation we use forward modeling to
134
validate the existence of a buried crater, as well as comparison with the gravity
135
signatures of known unburied craters of similar size. Our analysis focuses on
136
two features within Mare Tranquillitatis and Mare Serenitatis. The first feature 7
ACCEPTED MANUSCRIPT is completely buried by subsequent volcanic flows, whereas the second feature
138
is only partially buried, although it was never previously recognized as a crater
139
until gravity data became available. The density contrast between the buried
140
feature and the overlying mare basalt, in addition to a central anomaly created
141
by the mantle uplift beneath these large craters, produce gravity signatures
142
that permit the identification of these distinct quasi-circular anomalies.
143
3
144
Because the Moon lacks an atmosphere, as well as fluvial and sedimentologi-
145
cal processes and plate subduction, impact craters are typically well preserved
146
by terrestrial standards. Nevertheless, later impacts may alter a preexisting
147
crater or even completely remove any visible topographic evidence of its former
148
presence or may be covered under the ejecta blanket of the nearby impacts.
149
Additionally, volcanic activity and subsequent emplacement of thick piles of
150
mare basalt flows can be especially effective in burying existing craters ei-
151
ther partially or completely. A widely cited lunar ‘ghost’ crater is the 112-km
152
diameter Flamsteed P ring located in southern Oceanus Procellarum as il-
153
lustrated in Figure 1. The interior of Flamsteed P is completely filled with
154
mare basalt. The discontinuous ring of hills are believed to be remnants of
CR IP T
137
AC
CE
PT
ED
M
AN US
Buried Craters
155
the crater rim, making the Flamsteed P ring a prime example of a crater that
156
was nearly, but not quite completely, buried by basalt flows. Smaller craters,
157
craters with lower rims, or those buried by thicker basalt piles, however, may
158
be completely obscured. Thus, visual detection based on regional imagery may
159
not be effective in identifying completely buried craters. 8
CR IP T
ACCEPTED MANUSCRIPT
AN US
Fig. 1. Flamsteed P ring forming a discontinuous ring of hills. Courtesy: NASA/Lunar and Planetary Institute, Lunar Orbiter 4, image 143, h3.
4
Detection of Buried Craters
161
The non-circularity of the gravitational footprint of the features can be further
162
analyzed within the context of the interpreted impact origin. The anomalies
163
under investigation in this paper closely resemble approximately 100 large
164
quasi-circular positive Bouguer anomalies revelaed by GRAIL data. Neumann
165
et al. (2015) interpreted these non-circular and circular anomalies as due to the
166
density contrast between the surrounding crust and mantle material uplifted
167
by the excavation of large impact craters. A large number of these anomalies
AC
CE
PT
ED
M
160
168
are unambiguously associated with known impact basins or peak-ring craters,
169
although 16 (Neumann et al., 2015, table S6) of these anomalies are described
170
as the traces of otherwise obliterated ancient impact craters. Evans et al.
171
(2016) further identified 104 circular (positive and negative) Bouguer anoma-
172
lies in the Moons mare regions which they interpreted as complex impact 9
ACCEPTED MANUSCRIPT craters flooded by basalt, an interpretation based on the circularity and size
174
of the anomalies, although not supported by the type of detailed modeling
175
reported in this paper. Both of these papers assumed that the anomalies were
176
associated with impact craters rather than volcanic intrusions because of the
177
near-circular pattern of the anomalies, which ranged in diameter from 30 to
178
150 km (Evans et al., 2016) and 100 km or more (Neumann et al., 2015),
179
and in magnitude up to several hundred mgal. An alternative interpretation
180
is that these anomalies are caused by subsurface igneous intrusions. However,
181
volcanic intrusions only rarely approximate a circular disk of the observed di-
182
ameter (most such intrusions are linear) and the magnitude of the anomalies,
183
typically more than 100 mgal, would require thickness of intrusive magma of
184
more than 5 km, assuming a density contrast between intrusive rock and the
185
surrounding crust of 500 kg/m3 . This thickness greatly exceeds that of most
186
terrestrial intrusions and, moreover, would be expected to uplift the overlying
187
crust into a positive relief feature resembling a broad laccolith, which is at
188
odds with the lack of observed topographic expression for our two features.
189
The gravity signatures of volcanic intrusions most similar to those we report
190
are found in floor-fractured craters, which are believed to form when igneous
191
intrusions uplift the floors of craters ranging between 10 and 200 km in di-
192
ameter. A recent study of such craters by Jozwiak et al. (2016) shows that
193
the Bouguer anomalies of the intrusions are seldom circular, are associated
AC
CE
PT
ED
M
AN US
CR IP T
173
194
with non-axisymmetric domed uplifts and volcanic features, are usually offset
195
from the crater center, and are typically only a few tens of mGal, approaching
196
100 mGal only in the case of Humboldt crater. Even in this case, Jozwiak
197
et al. (2016) attribute most of the anomaly to mantle uplift beneath this large
198
(207 km diameter) crater. The best-known volcanic intrusion on the Moon 10
ACCEPTED MANUSCRIPT is probably the highly volcanic Marius Hills region, which is marked by a
200
broad topographic swell that rises about 500 m above the surrounding plains
201
and possesses a quasi-circular positive Bouguer anomaly that ranges up to
202
about 100 mGal. This positive anomaly is surrounded by an annular negative
203
anomaly of about -50 mGal that resembles a flexural moat. As we show in
204
the Supporting Material, a gravity gradient analysis of the Marius Hills re-
205
gion does not exhibit the same degree of circularity of either of our Earhart
206
or Ashoka anomalies.
207
In the current investigation, two features are presented as a part of detec-
208
tion and validation efforts. The apparent features investigated include a com-
209
pletely buried anomaly within northern Mare Tranquillitatis and the partially
210
buried rim of what is almost certainly an impact crater lying in the north-
211
western region of Lacus Somniorum (northeast of Mare Serenitatis). In this
212
paper, the provisional name “Ashoka” is adopted for the Mare Tranquillitatis
213
anomaly and “Earhart” for the barely-visible crater in Lacus Somniorum. Ap-
214
parently,“Earhart” has never been previously recognized as a crater but, based
215
on the knowledge of the quasi-circular gravity anomaly location, it is possible
AC
CE
PT
ED
M
AN US
CR IP T
199
216
to visualize an incomplete quasi-circular structure in the surface topography
217
(Figure 7). These provisional names were selected to honor Ashoka Maurya,
218
in whose kingdom was situated Lonar, India’s first known impact crater (and
219
one of a very few terrestrial craters in basaltic rock), and Amelia Earhart, a
220
female explorer and early aviatrix, respectively. 11
ACCEPTED MANUSCRIPT 4.1 Speculated Buried Crater 1 (Ashoka)
222
The first subsurface anomaly under investigation lies at the northern edge of
223
Mare Tranquillitatis. The vicinity around the Ashoka anomaly is filled with
224
mare basalt and has been subjected to subsequent impacts as apparent in
225
Figure 2 using topography data collected by LOLA. The figure is centered
CR IP T
221
M
AN US
around the speculated buried crater. Recent work by Robinson et al. (2012)
Fig. 2. Local topography in the vicinity of the Ashoka anomaly. The color bar
ED
represents the elevation in meters with respect to a lunar reference sphere of radius
PT
1738 km.
226
confirmed the presence of a sublunarean void at coordinates 8.335o N and
228
33.222o E marked by a red dot and commonly known as the Mare Tranquil-
229
litatis skylight. For this investigation, the authors will limit the discussion to
AC
CE
227
230
buried craters, however, additional work investigating skylights and lava tubes
231
is also being carried out. Based on solely visual analysis of the topography for
232
the region of interest, there appears to be no evidence of any buried structure.
233
Thus, to detect the presence of a subsurface feature, free-air and Bouguer
234
gravitational data is exploited. 12
ACCEPTED MANUSCRIPT 4.1.1
Gradiometry, Free-air and Bouguer Gravity Anomalies
236
For this application, two gravity models up to degree and order 900 are con-
237
sidered (Lemoine et al., 2014). Each model is truncated on both ends of the
238
SH expansion and tapered to attenuate the resulting ringing. The gradiometry
239
technique is then applied to the localized region for the two gravity models.
240
Figure 3 illustrates the corresponding local averaged maximum eigenvalues
241
for the free-air gravity, Bouguer gravity, and the correlation between the two
CR IP T
235
M
AN US
gravitational fields. The figure suggests some form of quasi-circular anomaly
Fig. 3. Local gradiometry maps of the free-air gravity field (left), the Bouguer gravity
ED
anomaly field (center), and the correlation between the free-air and Bouguer gravity fields (right) for the Ashoka anomaly. The maps overlay local topography and the
PT
color scale represents the signed magnitude corresponding to the largest eigenvalue
242
243
CE
of the Hessian derived from the gravitational potential.
about 120 km in diameter with its approximate central coordinates at 8.9o N and 31.0o E.
245
The Ashoka anomaly is further explored by generating regional free-air and
246
Bouguer gravity maps for the area under investigation. Figure 4 illustrates lo-
247
cal maps for the two gravity models. The figure on the left reflects the free-air
248
gravity anomaly field and the figure on the right represents Bouguer gravity
AC 244
13
CR IP T
ACCEPTED MANUSCRIPT
Fig. 4. Regional free-air and Bouguer gravity anomaly maps for Ashoka.
anomaly field over Ashoka. A subsurface quasi-circular anomaly is clearly ap-
250
parent in both anomaly fields, further supporting the existence of a buried
251
crater at this location. The non-circularity of the gravitational footprint may
252
be attributed to later impacts, thus altering altering the preexisting crater or
253
even completely remove any visible topographic evidence of its former presence
254
by being covered under the ejecta blanket of the nearby impacts. Additionally,
255
volcanic activity and subsequent emplacement of thick piles of mare basalt can
256
be especially effective in burying and altering the circularity of the craters.
257
4.1.2 Buried Peak-ring Basin: Ashoka
258
To better visualize the gravity signature and any form of existing topographic
259
evidence that may still exist after being partially buried under lava deposition
AC
CE
PT
ED
M
AN US
249
260
or under ejecta from later impacts, a larger area surrounding the anomaly is
261
investigated. The free-air and Bouguer gravity anomaly fields appear in Figure
262
5 for 540 km by 510 km wide region in the vicinity of Ashoka overlying regional
263
topography. The two gravitational maps clearly illustrate a central anomaly
264
with a slight gravitational ‘high’ relative to its surrounding. To investigate the 14
ACCEPTED MANUSCRIPT possibility of the central anomaly being associated with potentially a larger
266
buried structure, we employed the power-law fit to ring diameters (Dring ) and
267
rim-crest diameters (Dr ) of peak ring basins provided by Baker et al. (2011).
268
The power law trend follows the equation Dring = [0.14 ± 0.10](Dr )1.21±0.13 .
269
With the knowledge of the inner anomaly being approximately 120 km in di-
270
ameter as evident from Figure 3, the rim-crest of the basin can be calculated to
271
be approximately 265 km in diameter. The black circle on the Bouguer gravity
CR IP T
265
M
AN US
map outlines the inner anomaly ring, approximately 120 km in diameter, and
ED
Fig. 5. Free-air gravity map (left) and Bouguer gravity map (right) for Ashoka. The color represents free-air and Bouguer gravitational acceleration in the respective
PT
maps. The positive anomaly at 4.8o N and 23.4o E is part of (Neumann et al., 2015, table 1), and the two positive anomalies are #43 and #48 of Evans et al. (2016).
CE
The reason for picking the Ashoka anomaly, and not the two small but stronger positive anomalies, is because of the gravity gradient data in Figure 3, suggesting
AC
that Ashoka is larger and deeper than these two other anomalies. 272
273
the dotted magenta circle marks the supposed crater rim-crest, approximately
274
265 km in diameter. Comparing the two maps, the dotted magenta boundary
275
coincides, over a small fraction, with topographic feature in the lower half of
276
the circle. Although, the gradiometry technique along with regional free-air 15
ACCEPTED MANUSCRIPT
277
and Bouguer gravity maps aid in recognizing subtle features associated with
278
the buried anomaly, additional analysis using forward modeling technique is
279
employed to further verify the existence of a buried basin.
280
4.2
281
Similar to the detection and validation efforts outlined for Ashoka, the sec-
282
ond anomaly, Earhart, lying in the northwestern region of Lacus Somniorum
283
(northeast of Mare Serenitatis) is also investigated. To detect and confirm the
284
presence of the subsurface anomaly, gradiometry based analysis is carried out
285
for which free-air and Bouguer gravitational data is once again exploited. The
286
proposed Earhart crater predated Serenitatis and was almost completely over-
287
printed by the impact that resulted in the formation of Serenitatis that was
288
subsequently followed by volcanism, thus flooding the basin with lava basalts,
289
as well as the ejecta blanket of the Serenitatis impact.
290
4.2.1 Gradiometry, Free-air and Bouguer Gravity Anomalies
291
The gradiometry detection technique is applied to the localized region. Gravity
292
models up to degree and order 900 with predetermined truncation and tapers
293
are utilized (Lemoine et al., 2014). Figure 6 illustrates the corresponding local
294
averaged maximum eigenvalues for the free-air gravity, Bouguer gravity, and
AC
CE
PT
ED
M
AN US
CR IP T
Speculated Buried Crater 2 (Earhart)
295
the correlation between the two gravitational fields, respectively. The maps
296
depicts a subsurface quasi-circular anomaly of about 80 km in diameter with
297
its center at 41.2o N and 21.8o E. Even though Earhart is a partially buried
298
crater, so that its topography does not reflect its circularity, both free-air and
299
Bouguer eigenvalue maps illustrate a circular-shaped gravity anomaly. 16
CR IP T
ACCEPTED MANUSCRIPT
Fig. 6. Local gradiometry maps for the free-air gravity anomaly field (left), Bouguer gravity anomaly field (center), and the correlation between the free-air and Bouguer gravity anomalies (right) for the Earhart anomaly. The maps overlay local topog-
raphy and the color scale represents the signed magnitude corresponding to the
AN US
largest eigenvalue of the Hessian derived from the gravitational potential. 300
Continuing the validation to support the existence of Earhart, regional free-
301
air and Bouguer gravity anomaly maps are generated. Figure 7 illustrates
302
local maps for the free-air gravity anomaly field on the left and Bouguer grav-
CE
PT
ED
M
ity anomaly field on the right. Evidence for the presence of a quasi-circular
AC
Fig. 7. Regional free-air and Bouguer gravity maps for Earhart.
303
304
anomaly is apparent in the two gravity maps, a small fraction of which coin-
305
cides with the partial rim structure in the topography. In total, the detection
306
process and analysis supports the existence of the Earhart buried crater but 17
ACCEPTED MANUSCRIPT additional evidence is warranted that is later accomplished by forward mod-
308
eling.
309
4.2.2 Earhart Crater’s Regional Topography
310
The region is presently flooded with mare basalt that has been subjected to
311
a number of small impacts; the region appears in Figure 8a using topography
CR IP T
307
(b) Earhart depicted by black circle.
M
(a) Topography near Earhart.
AN US
data collected by LOLA. Based on visual analysis of Figure 8a, it appears
ED
Fig. 8. Speculated Buried Crater 2.
PT
312
that there is a partially buried, discontinuous ring structure interior to the
314
black circle as illustrated in Figure 8b. The heavy modification of Earhart also
315
suggests that it predated the Serenitatis impact and is overlain by Serenitatis
316
ejecta.
317
4.2.3
318
A similar analysis for buried peak-ring basin is also completed for a wider
319
area surrounding Earhart. Figure 9 reflects free-air and Bouguer gravity for
AC
CE
313
Buried Peak-ring Basin: Earhart
18
ACCEPTED MANUSCRIPT
CR IP T
a 900 km by 900 km wide region in the vicinity of Earhart. Once again,
Fig. 9. Free-air gravity map (left) and Bouguer gravity map (right) for Earhart. The
AN US
color represents free-air and Bouguer gravitational acceleration in the respective maps. 320
gravity maps are overlaid on regional topography for better visualization. The
322
gravitational maps clearly illustrate a bulls-eye shaped anomaly with a gravity
323
‘high’ relative to its surrounding. As evident from the Bouguer gravity map,
324
no other gravitational footprint is as distinguishable as the Earhart anomaly.
325
Neumann et al. (2015) clearly identified and used the edge of the Bouguer
326
high to locate the inner ring of a crater. Applying the similar strategy to
327
Earhart anomaly, the black circle on the Bouguer map outlines the inner
328
ring, approximately 80 km in diameter. The dotted magenta circle outlines
329
the boundary of the crater rim-crest, approximately 190-200 km in diameter
AC
CE
PT
ED
M
321
330
calculated using power-law fit as given by Baker et al. (2011). Other than
331
the few weak signatures scattered along the map, additional anomalies to the
332
south are simply gravitational highs at the boundary between mare Serenitatis
333
and topographic structures. In the case of Earhart, some topographic features
334
along the southwestern edge of the dotted magenta circle are apparent. There 19
ACCEPTED MANUSCRIPT is also the possibility of the crater being buried under the ejecta of mare
336
Imbrium and Serenitatis as evident from additional topographic structures in
337
the region. Once again, the topography alone does not aid in the detection
338
process for the circular anomaly, i.e., the bulls-eye pattern marked by black
339
and magenta circles. The gradiometry based technique assists in the detection
340
process. The free-air and Bouguer gravity maps also support the verification
341
process for the anomalies but additional forward modeling carried out in the
342
subsequent sections will assist in strengthening the case for the buried crates.
343
5
344
The gradiometry strategy relies on the gravity models derived from the GRAIL
345
data to detect features of interest. It is essential to assess the validity of such
346
detections. To adequately apply the concept that serves as a basis for the
347
characterization of a feature is based on a model that describes the gravita-
348
tional signature of an anomaly and estimates the required parameters from
349
the gradiometry maps. Following the initial setup, the gravitational potential,
350
gravity anomaly, and Hessian matrix are computed for the forward model
351
and compared with the initial simulation for each candidate structure. The
352
performance of the forward model is assessed by its ability to match the ob-
353
served signatures that correspond to the feature of interest on the gradiometry
AC
CE
PT
ED
M
AN US
Forward Model Development
CR IP T
335
354
maps. The objective is the development of a simple and computationally in-
355
expensive strategy to describe the gravitational anomaly that is associated
356
with a buried crater. It is assumed that the crater is buried beneath mare
357
flood basalts, ejecta blanket of subsequent impacts, and has no topographic
358
expression. Then, the gravity anomaly for a given buried crater arises from 20
ACCEPTED MANUSCRIPT the density contrast between crustal material, mare material, and mantle ma-
360
terial. A simple diagram that represents the assumed layering outside of the
361
crater is depicted in Figure 10. The formation of a crater disturbs this initial
362
layering, creating density contrasts that result in gravitational anomalies. It is
363
important to note that the objective of the forward model is not to provide a
364
perfectly matching model but a rapid and simple modeling approach to serve
365
as a basis for the validation of potential buried features. Additionally, the tool
366
that is used for such detections is primarily the gradiometry. The gradiom-
367
etry technique is most sensitive to density gradients directly, not simply the
368
magnitude of a given contrast. In this analysis, the gradient of interest is the
369
radial density contrast across the crater. Thus, some simplifying assumptions
370
are adopted, and in particular, the details of the composition of the floor of the
371
crater are not included. The model is formulated in terms of simple density
372
contrast between two types of material: crust and mantle/mare.
PT
ED
M
AN US
CR IP T
359
Fig. 10. Layering scheme outside the crater employed in the forward models, as
AC
CE
assumed to characterize an uncratered portion of the mare.
373
5.1 Construction of a Shape Model for Buried Craters
374
The morphology and characteristics of lunar craters as a function of their size
375
is fairly well-known. As crater size increases, different crater morphologies 21
ACCEPTED MANUSCRIPT appear, running upward in size from simple craters, to complex craters, and
377
finally peak ring basins (Baker et al., 2011; Pike, 1980; Allen, 1975; Melosh,
378
2011). In this analysis, the focus is on complex crater and peak ring basin mor-
379
phologies, as these types are most similar in size to the candidate detections.
380
The objective is to construct a relatively simple geometrical shape model for
381
such structures in terms of simple elementary blocks to allow rapid evaluation
382
of the corresponding gravity anomaly. In this analysis, any crater feature in
383
the forward model is a series of discrete annular rings or a stack of disks.
384
5.1.1 Buried complex crater model
385
On the moon, complex craters are features of diameter, Df , larger than 20 km
386
and smaller than about 200 km. Such craters are characterized by a relatively
387
flat floor, terraced walls and small central peak or peak cluster (Baker et al.,
388
2011; Pike, 1980; Allen, 1975; Melosh, 2011). Our focus is on buried features,
389
hence it is assumed that the crater is filled with lava material and the entire
390
structure is buried under a layer of mare, as illustrated in Figure 11. Note that
391
to construct a model that describes the gravity anomaly due to the presence
392
of a buried crater, the relevant information is not the physical structure of
393
the crater directly but the density contrast that is associated with the buried
394
structure. Within this context, the lava flooding in the crater cavity, up to the
395
initial mare/crust interface, is a mass surplus and corresponds to a positive
396
gravity anomaly. Similarly, the rim of the crater, because it is surrounded by
397
mare material, corresponds to a mass deficit and a negative gravity anomaly.
398
The forward model for a complex crater is then comprised of two building
399
blocks: the rim and lava fill. The lava fill is modeled as a disk with positive
AC
CE
PT
ED
M
AN US
CR IP T
376
22
ACCEPTED MANUSCRIPT
CR IP T
Fig. 11. Complex crater geometry.
density contrast ∆ρ = ρmare − ρcrust ≈ 640kg/km3 (Kiefer et al., 2012; Wiec-
401
zorek et al., 2013; Gong et al., 2015), a diameter equal to the assumed crater
402
diameter and a constant thickness. The rim is constructed as a series of rings
403
with an inner diameter equal to the assumed crater diameter. The width of
404
each ring in the rim model is fixed and the thickness of a given ring is com-
405
puted such that the height of the rim is proportional to 1/r3 ,(Melosh, 2011)
406
where r is the radius of a given ring. The density contrast of the rim is opposite
407
to that of the lava fill value. For simplicity, the rim structure is constructed
408
such that the volume of the fill and the rim are equal.(Melosh, 2011) A sample
409
forward model for a 100 km diameter crater is illustrated in Figure 12 where
410
the density contrast that is associated with the presence of a buried lava fill
411
and rim is represented in red and blue, respectively. It is also assumed that
412
the model is axisymmetric with respect to the center of the crater, i.e., the
413
left edge of the lava fill.
CE
PT
ED
M
AN US
400
0
AC
depth, km
5
−5
−10
0
20
40
60 80 Radial distance, km
100
120
Fig. 12. Axisymmetric forward model for complex crater including lava fill (red) and rim (blue)
23
ACCEPTED MANUSCRIPT 5.1.2
Buried complex crater model with mantle uplift and central peak
415
Based on the initial simple forward model for a complex crater, additional
416
crater features can also be included. A central peak may form, and similar
417
to the rim, such a peak corresponds to a mass deficit or a negative gravity
418
anomaly (Allen, 1975). Also, depending on the crustal thickness, the size of
419
the crater, and other parameters, a mantle uplift in the center region of the
420
crater may appear. Similar to the lava fill, the mantle uplift is a mass surplus
421
because denser mantle material replaces crustal rocks. The same density con-
422
trast between mare fill and crust is assumed for the crust/mantle interface,
423
that is, ∆ρ = ρmantle − ρcrust ≈ 640kg/km3 . A downward crustal bulge of-
424
ten forms outboard of the mantle uplift, constituting a mass deficit (Milbury
425
et al., 2015), as illustrated in Figure 13. Note that while we model this mass
426
deficit as crustal thickening, higher porosity may also be a contributor.
PT
ED
M
AN US
CR IP T
414
Fig. 13. Assumed geometry for a buried complex crater with mantle uplift, crustal
Our forward model for a complex crater, composed of a lava fill and a rim
AC
427
CE
bulge and central peak in addition to a lava fill and crater rim.
428
structure, is further augmented with a central peak, mantle uplift, and crustal
429
bulge, as shown in Figure 13. The central peak is constructed as a conical
430
stack of disks with fixed thickness that originates on the crater floor with a
431
given basal diameter. The subsequent disks are added in the model such that
432
the final structure satisfies the prescribed final central peak diameter at the 24
ACCEPTED MANUSCRIPT chosen height. The mantle uplift model is constructed similarly. The series
434
of disks originates at the interface between the crust and the mantle with
435
a diameter equal to Df /2. The thickness of individual disks, the height and
436
apex diameter of the mantle uplift are prescribed to construct the mantle up-
437
lift model. In the complex crater model, the mantle uplift apex diameter is
438
assumed to be equal to zero (that is, the mantle uplift is a cone terminating
439
in a point). Finally, a model for the crustal bulge is also incorporated. The
440
crustal bulge is constructed similarly to the rim, that is, a series of rings with
441
fixed radii that originate at Df /2 and terminate at Df . The height of indi-
442
vidual rings is governed by either a short or long wavelength sine function
443
to produce a lobe-shaped structure. The density contrast for each of these
444
structures is assigned consistent with the gravity anomaly type, that is, pos-
445
itive or negative. A sample forward model of a complex crater with mantle
446
uplift, crustal bulge and central peak is illustrated in Figure 14. While these
447
features are detectable in the gravity signature, the mantle uplift dominates.
448
The other smaller features are more apparent in the gradiometry profiles as
449
the eigenvalue is most sensitive to gradients in the field, even of smaller mag-
450
nitude. While these higher-order features may contribute to a better fit, they
ED
M
AN US
CR IP T
433
PT
0 −10
CE
depth wrt R0, km
do not dominate the traits of the gravity and gradiometry curves. Finally, the
−20
AC
−30
0
20
40
60
80 100 120 Radial distance, km
140
160
180
200
Fig. 14. Forward model for complex crater.
451
452
gradiometry technique is very sensitive to sharp changes in the gravitational
453
potential function. In the initial forward model for a complex crater, note the 25
ACCEPTED MANUSCRIPT sharp transition between the lava fill and the rim. In reality, a transition region
455
between the cavity and the rim exists. The forward model is thus augmented
456
with such a transition region to alleviate the artificial sharpness of the shape
457
model. This transition region, similar to the rim, is constructed as a series
458
of rings located between the lava fill disk and the first ring of the rim. The
459
height of each ring is determined such that a smooth transition between the
460
depth of the fill and the height of the rim is achieved. Note that, while sharp
461
discontinuities do not necessarily reflect in the gravity anomaly depending on
462
the magnitude, they are more significant in the gradiometry technique. Thus,
463
this transition region is primarily included for numerical purpose rather than
464
in a pursuit of physical fidelity.
465
5.1.3 Buried peak ring crater model
466
Larger crater structures on the Moon are called peak ring basins. The overall
467
morphology of peak ring basins is similar to complex craters but a peak ring
468
may form inside the crater rim with a diameter roughly equal to Df /2. Also,
469
such craters may exhibit a broader plateau-shaped mantle uplift, as illustrated
470
in Figure 15. Then, the forward model for a peak ring basin is constructed
471
similarly to the complex crater scenario and the model is augmented with a
PT
ED
M
AN US
CR IP T
454
AC
CE
peak ring. Similar to the crustal bulge, the peak ring is constructed as a stack
Fig. 15. Peak ring crater geometry. 472
26
ACCEPTED MANUSCRIPT
473
of annular rings that lies on the crater floor. The width of each ring is fixed
474
and the height is determined with a sine function to produce a bell-shaped
475
structure. The overall width and height of the peak are prescribed to complete
0 −10 −20 −30 0
50
100
150 200 Radial distance, km
250
CR IP T
depth wrt R0, km
the model, as illustrated for a sample peak ring basin in Figure 16.
300
350
Fig. 16. Peak ring crater geometry. The center of the axisymmetric structure is on the left and the outer part of the crater is on the right of this profile.
AN US
476
Note that the different elements of the forward model may imply specific as-
478
pects of the rheology of mantle. However, while the rheology of the mantle is
479
important, to first order, the density contrast dominates the driving factors
480
that affect the gravity signature of an anomaly. Additional details leading to
481
forward model development and steps towards computation of the gravita-
482
tional potential, acceleration, and the Hessian matrix from a shape model for
483
buried craters is provided in Supplementary material.
484
5.2
CE
PT
ED
M
477
AC
Forward Model Validation
485
The forward modeling approach is validated by applying the strategy to two
486
known exposed impact structures, Compton crater and Schroedinger basin.
487
The capability of the models to represent the gravity anomalies that are as-
488
sociated with these features serves as a basis to evaluate the performance of
489
the forward model. 27
ACCEPTED MANUSCRIPT 5.2.1
Compton Crater
491
Compton is a complex crater located on the far side of the Moon, at 55.3◦ N,
492
103.8◦ E, with a diameter Df approximately equal to 162 km. The free-air
493
and Bouguer anomalies that are associated with the crater are illustrated on
494
the left and right images, respectively, in Figure 17. The free-air anomaly
495
is characterized by a large positive gravity anomaly that corresponds to the
496
crater rim, while the Bouguer anomaly is poorly correlated with the crater,
CR IP T
490
ED
M
AN US
although it is, in general, smaller inside the crater than exterior to it. The
PT
Fig. 17. Free-air (left) and Bouguer (right) gravity anomaly for Compton crater. 497
absence of a positive Bouguer anomaly in the center of the crater suggests
499
an absence of mantle uplift. To compare the gravity data and the forward
500
model, an azimuthal average profile of the gravity anomalies is computed, as
AC
CE
498
501
illustrated on the left plot in Figure 18. A series of concentric great circles
502
for a predetermined diameter range are constructed at a pole and rotated
503
to the center of the crater or anomaly of interest. For each rotated circle,
504
the gravity, or eigenvalue, anomaly is sampled along the circle and averaged.
505
This series of points constitute the azimuthally averaged profile as a function 28
ACCEPTED MANUSCRIPT
506
of radial distance from the anomaly center. All subsequent gravity profiles
507
discussed in this analysis are azimuthally averaged. The gradiometry technique
508
is also applied to this region and a similar profile for the maximum magnitude eigenvalue of the Hessian is displayed in the left plot in the figure. Recall Gravity Anomaly
Eigenvalue
100
Free−air Bouguer
eigenvalue, eotvos
20
0
0
−20
−40
−50
AN US
gravity anomaly, mgal
50
CR IP T
40
−60
−100
−80
100 150 radial distance, km
200
50
100 150 radial distance, km
200
M
50
Fig. 18. Azimuthal average for the free-air and Bouguer gravity anomaly (left) and
ED
eigenvalue (right) profile for Compton crater. 509
that a mass deficit corresponds to a negative gravity anomaly and a positive
511
eigenvalue, while a mass surplus is associated with a positive gravity anomaly
512
and a negative eigenvalue. Note that Compton is an exposed crater, thus, the
513
objective with the forward model is to represent the free-air gravity anomaly
514
and the eigenvalue.
515
A forward model for Compton is constructed assuming that the crater can
516
be reasonably represented as a large complex crater with a central peak. In
517
fact, Compton is understood to be transitional to a peak-ring basin. Some-
518
times denoted a “protobasin” (Baker et al., 2011), the central peak cluster is
AC
CE
PT
510
29
ACCEPTED MANUSCRIPT surrounded by a very subdued ring of hills with a diameter equal to about
520
half the rim diameter. However, these hills are too small to contribute sub-
521
stantially to the gravity anomaly and our model neglects them. Compton is
522
not flooded with lava, so the density contrast is accordingly adjusted. In this
523
case the rim and the central peak reflect a mass surplus and the crater cavity,
524
lacking any lava fill, corresponds to a mass deficit. The parameters for each
525
structure in the forward model are summarized in Table 1, where the symbols
526
hRim , hF ill , hCP , wCP , and hCrust denote the height of the rim, lava fill, central
527
peak, and crust, respectively; the resulting shape model is displayed in Figure
528
19. Note that the parameter hF ill in this context corresponds to the crater
529
cavity height, as the crater is assumed to be exposed rather than buried and filled with lava material. Table 1
AN US
CR IP T
519
Compton forward model characteristics hF ill
0.75 km
1.2 km
hCP
M
hRim
hCrust
13 km
30 km
ED
1.7 km
wCP
530 1
PT
0.5
−0.5
AC
CE
depth with respect to reference sphere, km
0
−1
−1.5
−2
−2.5
−3
−3.5
−4
0
50
100 Radial distance, km
150
200
Fig. 19. Forward model for Compton crater.
30
ACCEPTED MANUSCRIPT The gravity anomaly and the eigenvalues that are associated with the for-
532
ward model are computed and compared with the free-air profiles from the
533
GRAIL data. These profiles are represented in Figure 20 in red and black
534
for the data and the forward model, respectively. Also, to better assess the
535
fit of the forward model, a red envelope that depicts the 1-σ azimuthal dis-
536
persion of the gradiometry and gravity data is overlaid on the figure. Note
537
that the parameters in the forward model could be further tuned to achieve
538
an even better agreement, but, overall the forward model is successful in re-
539
producing the gravity signature that corresponds to Compton crater. Recall
540
that the objective is not to provide a perfectly matching model but, rather, a
541
rapid and simple modeling approach to serve as a basis for the validation of
542
potential buried feature detection. Interestingly, the gradiometry fit requires
543
a small mantle uplift beneath the crater center, which is not apparent from
544
the Bouguer or Free Air data contours alone. It is simply that the mantle
545
uplift that is incorporated in the forward model is not significant enough to
546
have an expression in the gravity anomaly but does reflect in the eigenvalue
547
technique, as the gradiometry technique is most sensitive to variations in the
548
gravity field.
549
5.2.2 Schroedinger Basin
550
A second simulation is completed for a larger structure, that is, Schroedinger
AC
CE
PT
ED
M
AN US
CR IP T
531
551
basin. This feature is a peak ring basin located on the far side of the Moon, at
552
75◦ S, 132.4◦ E, with a diameter Df approximately equal to 312 km. The free-
553
air and Bouguer anomalies associated with the basin are illustrated on the
554
left and right images, respectively, in Figure 21. The free-air anomaly is again
555
characterized by a large positive gravity anomaly that corresponds to the rim 31
AN US
CR IP T
ACCEPTED MANUSCRIPT
Fig. 20. Free-air gravity anomaly (left) and eigenvalue (right) azimuthal profiles for Compton crater computed with GRAIL data (red) and forward model (black).
AC
CE
PT
ED
M
of the basin, but also features a prominent central positive anomaly. Unlike
Fig. 21. Free-air (left) and Bouguer (right) gravity anomaly for Schroedinger basin. 556
557
Compton, the Bouguer gravity signature is characterized by a pronounced 32
ACCEPTED MANUSCRIPT positive anomaly in the center of the basin, indicative of a substantial mass
559
surplus, that may correspond to features such as a mantle uplift or thick
560
section of volcanics. The azimuthal average profile for the gravity anomalies
561
and the eigenvalues are computed and illustrated on the left and right images,
562
respectively, in Figure 22. Note the steep gradient in the eigenvalue profile
563
toward the center of the crater. Such a negative eigenvalue is indicative of a
564
mass surplus consistent with mantle uplift or lava fill. However, the slope of
565
the profile dictates a sharp feature toward the center of the anomaly, only
566
consistent with mantle uplift. The objective is then to represent the free-air
CR IP T
558
gravity anomaly and eigenvalue with the forward model strategy. Eigenvalue
AN US
Gravity Anomaly 250
60
200
Free−air Bouguer
40
150
0
ED
−50
−100
PT
−150
eigenvalue, eotvos
50
M
gravity anomaly, mgal
20
100
0
−20
−40
−60
−80
−200
100 150 200 250 radial distance, km
300
50
100 150 200 250 radial distance, km
300
CE
50
Fig. 22. Azimuthal average for the free-air and Bouguer gravity anomaly (left) and
AC
eigenvalue (right) profiles for Schroedinger basin.
567
568
A forward model for Schroedinger is constructed assuming the feature is rea-
569
sonably represented as a peak ring basin. In addition to the central peak, the
570
forward model includes a mantle uplift, crustal bulge, and peak ring. Similarly 33
ACCEPTED MANUSCRIPT
571
to Compton, the crater is exposed, for which we adjust the density contrast.
572
The peak ring, similar to the rim and central peak, is a mass surplus and
573
the crater cavity is a mass deficit. The parameters for each structure in the
574
forward model are summarized in Table 2, where the additional symbols hM U ,
575
hCB , and hP R denote the height of the mantle uplift, crustal bulge, and peak
Table 2 Schroendiger forward model characteristics.
CR IP T
ring, respectively; the resulting shape model is displayed in Figure 23.
hRim
hF ill
hCP
wCP
hCrust
hM U
1.25 km
1.85 km
0.4 km
30 km
30 km
15 km
0
4.3 km
0.4 km
−10
M
−15
−20
−25
ED
depth with respect to reference sphere, km
−5
PT
−30
−35
hP R
AN US
576
hCB
0
50
100
150 200 Radial distance, km
250
300
Then, the gravity anomaly and the eigenvalue that are associated with the
AC
577
CE
Fig. 23. Forward model for Schroedinger basin.
578
forward model are computed and compared with the free-air profiles from the
579
GRAIL data. These profiles are illustrated in Figure 24 in red and black for the
580
data and the forward model, respectively. Similar to Compton, in fact slightly
581
better, the forward model provides a good overall agreement with GRAIL data
582
and is successful in reproducing Schroedinger’s gravity signature, although it 34
ACCEPTED MANUSCRIPT
583
requires a much larger mantle uplift, consistent with the basin’s much larger
AN US
CR IP T
size.
M
Fig. 24. Azimuthal average for the free-air gravity anomaly (left) and eigenvalue (right) profiles for Schroedinger basin computed with GRAIL data (red) and forward
ED
model (black).
6
Forward Model Application to Buried Features
AC
585
CE
PT
584
586
After initially validating our forward modeling strategy with two known lunar
587
craters, the next step in the analysis is an application of the forward mod-
588
eling approach to the speculated buried craters that we detected with our
589
gradiometry technique. The objective of subsequent investigation is to con-
590
struct a forward model that captures the overall gravitational signature of the 35
ACCEPTED MANUSCRIPT potential features, both in terms of the gravity anomaly and the eigenvalue.
592
Compared to the validation scenarios, where the size and overall features of
593
the craters were known, an additional challenge in characterizing the Ashoka
594
and Earhart anomalies is the absence of any a priori information other than
595
the gravity filed itself.
596
6.1
597
The first speculated buried crater, informally named Ashoka, is centered at
598
8.7◦ N, 31.2◦ E, and the diameter of the positive Bouguer anomaly is approxi-
599
mately equal to 100-120 km. The size of the anomaly is not, however, necessar-
600
ily the size of the structure. Similar to Compton and Schroedinger, azimuthal
601
average profiles for the gravity anomalies and eigenvalues are produced to
602
compare with the forward model, as illustrated on the left and right images,
603
respectively, in Figure 25. In contrast to the previous two craters, however,
604
Taking into consideration that Ashoka is a buried feature and, thus, the ob-
605
jective is to represent the Bouguer gravity anomaly and eigenvalue with the
606
forward model strategy.
607
As the exact size of the potential crater is unknown several forward models for
608
the Ashoka anomaly are constructed. The model that seems to best capture
609
the overall anomaly is a complex crater morphology with a pronounced mantle
610
uplift. The parameters for the forward model are summarized in Table 3, where
611
hM are denotes the thickness of the lava layer; the corresponding shape model
612
is then illustrated in Figure 26.
613
The gravity anomaly and eigenvalue for the forward model are computed and
CR IP T
591
AC
CE
PT
ED
M
AN US
Ashoka anomaly
36
ACCEPTED MANUSCRIPT
Gravity Anomaly
Eigenvalue Free−air Bouguer
120 10 100 0
60
40
−10
−20
20 −30 0 −40 −20
−50 100 150 radial distance, km
200
50
100 150 radial distance, km
200
AN US
50
CR IP T
eigenvalue, eotvos
gravity anomaly, mgal
80
Fig. 25. Azimuthal average for the free-air and Bouguer gravity anomaly (left) and eigenvalue (right) profile for Ashoka crater. Table 3
Ashoka crater forward model characteristics. hRim
hF ill
hCP
wCP
hCrust
hM U
hCB
hM are
160 km
1 km
2 km
0 km
0 km
20 km
12 km
4 km
3.5 km
ED
M
Df
displayed in black on the left and right plots, respectively, in Figure 27. For
615
comparison, profiles from GRAIL data are also overlaid in blue, and although
616
the agreement between the data and the forward model is not as complete
617
as in the validation scenarios, the overall fit is satisfying. Recall that the ob-
618
jective is not to provide a perfectly matching model but, rather, a rapid and
AC
CE
PT
614
619
simple modeling approach to serve as a basis for the validation of potential
620
buried feature detection. The construction of the shape model in this scenario
621
is driven by several factors. First, note the steep gradient in the eigenvalue
622
profile toward the center of the crater. Such a negative eigenvalue is indicative
623
of a mass surplus consistent with mantle uplift or lava fill. However, the slope 37
ACCEPTED MANUSCRIPT
0
−10
−15
−20
−25
−30
0
20
40
60
80 100 Radial distance, km
120
140
CR IP T
depth with respect to reference sphere, km
−5
160
180
CE
PT
ED
M
AN US
Fig. 26. Forward model for Ashoka crater.
Fig. 27. Azimuthal average for the Bouguer gravity anomaly (left) and eigenvalue
AC
(right) for Ashoka crater computed with GRAIL data (blue) and forward model (black).
624
of the profile dictates a sharp feature toward the center of the anomaly, only
625
consistent with mantle uplift. The distinction between the two mantle uplift
626
models, without or with a plateau, is reflected in the slope of the eigenvalue
627
curve. The plateau-shaped model is associated with a flatter eigenvalue curve 38
ACCEPTED MANUSCRIPT in the innermost region of the crater as the mass surplus no longer varies radi-
629
ally. In this scenario, the slope of the eigenvalue curve remains steep up to the
630
cater center. Also, the location of the positive peak in the eigenvalue profile,
631
corresponding to a mass deficit, hence, consistent with the rim of the crater,
632
constrains the diameter of the feature to some range. There is, however, no
633
signature consistent with a central peak or a peak ring; such features would
634
be expressed as a central positive eigenvalue peak and another positive peak
635
halfway between the center and the rim. The accumulation of this informa-
636
tion further constrains the possible morphology of the buried structure. The
637
anomaly appears to be consistent with a crater that is sufficiently large to
638
possess a pronounced mantle uplift, yet not so large that the mantle uplift
639
exhibits a plateau shape or a peak ring.
640
6.2
641
The second speculated buried crater, informally named Earhart, is centered
642
at 41.5◦ N, 21.7◦ E, and the diameter of the Bouguer anomaly is similar to
643
that of Ashoka. The azimuthal average profiles for the gravity anomalies and
644
eigenvalues that are associated with Earhart crater are illustrated on the left
645
and right images, respectively, in Figure 28. The profiles appear very similar to
646
the Ashoka anomaly scenario, but with a stronger positive Bouguer anomaly
647
at the center of the crater. A similar morphology to Ashoka is adopted to
648
construct the forward shape model, with parameters summarized in Table 4.
649
The Curve Fitting strategy detailed in the supplementary material is applied
650
following which optimization is performed on the initial solution as initial guess
651
and the corresponding optimized shape model is illustrated in Figure 29. The
AN US
CR IP T
628
AC
CE
PT
ED
M
Earhart anomaly
39
ACCEPTED MANUSCRIPT
Gravity Anomaly
Eigenvalue
200 Free−air Bouguer
10
0 150
100
−20
−30
50 −40
0
−50
100 150 200 radial distance, km
250
50
100 150 200 radial distance, km
250
AN US
50
CR IP T
eigenvalue, eotvos
gravity anomaly, mgal
−10
Fig. 28. Azimuthal average for the free-air and Bouguer gravity anomaly (left) and eigenvalue (right) profile for Earhart crater. Table 4
180 km
1 km
hF ill
hCP
ED
hRim
2 km
0 km
wCP
hCrust
hM U
hCB
hM are
0 km
20 km
17 km
5 km
2.5 km
PT
Df
M
Earhart crater forward model characteristics.
profiles for the forward model and those from the GRAIL data, represented
653
in black and blue, respectively, are overlaid in Figure 30 on the left and right
654
plots, respectively.
AC
CE
652
655
Overall, the size of the crater is similar to the initial simulation, Df ≈ 200
656
km and, notably, the mare fill is thinner such that the crater rim lies just
657
underneath the surface. Also, the agreement between the eigenvalue profiles
658
is improved, although some discrepancy in the inner part of the crater still
659
remains. 40
ACCEPTED MANUSCRIPT
0
−10
−15
−20
−25
−30
0
20
40
60
80 100 Radial distance, km
120
140
CR IP T
depth with respect to reference sphere, km
−5
160
180
CE
PT
ED
M
AN US
Fig. 29. Forward model for Earhart crater after curve fitting.
Fig. 30. Azimuthal average for the Bouguer gravity anomaly (left) and eigenvalue
AC
(right) for Earhart computed with GRAIL data (blue) and forward model (black). 660
7
Summary
661
Gravitational analysis using gradiometry technique lead to the detection of
662
buried craters that is consistent with the observations and buried craters iden-
663
tified by other authors (Neumann et al., 2015; Evans et al., 2016), however, 41
ACCEPTED MANUSCRIPT the interpretation is non-unique due to the nature of gravity data. As a part
665
of the validation process, a partially and a completely buried structure are
666
investigated. Gravitational signatures that correspond to two putative buried
667
craters, here identified as Ashoka and Earhart craters, are corroborated by
668
evidence from free-air and Bouguer gravity anomaly maps, further sharpened
669
by our gradiometry technique. Forward modeling further supports the pres-
670
ence of a completely buried Ashoka crater and a partially buried Earhart
CR IP T
664
(b) Earhart crater.
ED
(a) Ashoka crater.
M
AN US
crater, shown in Figure 31. Preliminary analysis based on gradiometry tech-
Fig. 31. Regional Bouguer maps: Black circle outlines the mantle uplift and the
PT
dotted magenta circle marks the crater rim derived from the forward model for the two craters. The forward modeling technique revealed the true size of the two
nique suggested that the two buried craters were 120 km and 80 km in diam-
AC
672
CE
craters under investigation. 671
673
eter. However, further analysis to complete the study, encompassing forward
674
modeling, revealed that the true size of the two buried craters is indeed ap-
675
proximately 160 km (Ashoka crater) and 200 km (Earhart crater) in diameter.
676
The forward modeling approach is validated by applying the strategy to two
677
known exposed impact structures, Compton crater and Schroedinger basin; 42
ACCEPTED MANUSCRIPT the strategy is successful in reproducing the gravity signatures that corre-
679
spond to both features. The strategy offers an initial assessment of the gravity
680
anomalies that are detected, and generally matches the observed values for
681
Earhart and Ashoka. Further refinement of the forward model, in particular,
682
the mantle uplift model, may be required in future work. Current ongoing
683
efforts include the implementation of automatic curve fitting techniques to
684
improve the agreement between the forward model and the GRAIL data.
685
Acknowledgements
686
The research work described in this paper made use of GRAIL data and was
687
carried out at Purdue University. The GRAIL mission is supported by the
688
NASA Discovery Program under contract to the Massachusetts Institute of
689
Technology and the Jet Propulsion Laboratory, California Institute of Tech-
690
nology.
691
References
692
Allen, C. C., Apr. 1975. Central peaks in lunar craters. Moon 12, 463–474.
693
Andrews-Hanna, J. C., Asmar, S. W., Head, J. W., Kiefer, W. S., Kono-
694
pliv, A. S., Lemoine, F. G., Matsuyama, I., Mazarico, E., McGovern, P. J.,
AC
CE
PT
ED
M
AN US
CR IP T
678
695
Melosh, H. J., Neumann, G. A., Nimmo, F., Phillips, R. J., Smith, D. E.,
696
Solomon, S. C., Taylor, G. J., Wieczorek, M. A., Williams, J. G., Zuber,
697
M. T., 2013. Ancient igneous intrusions and early expansion of the moon
698
revealed by GRAIL gravity gradiometry. Science 339 (6120), 675–678.
699
Baker, D. M. H., Head, J. W., Fassett, C. I., Kadish, S. J., Smith, D. E., 43
ACCEPTED MANUSCRIPT
700
Zuber, M. T., Neumann, G. A., 2011. The transition from complex crater
701
to peak-ring basin on the moon: New observations from the lunar orbiter
702
laser altimeter LOLA instrument. Icarus 214 (2), 377–393.
703
Chappaz, L., Sood, R., Melosh, H., Howell, K., 2014. Buried empty lava tube
704
detection with GRAIL data. In: AIAA SPACE 2014 Conference and Expo-
705
sition. Evans, A. J., Soderblom, J. M., Andrews-Hanna, J. C., Solomon, S. C., Zuber,
707
M. T., 2016. Identification of buried lunar impact craters from grail data
708
and implications for the nearside maria. Geophysical Research Letters .
CR IP T
706
Gong, S., Wieczorek, M., Nimmo, F., Kiefer, W., Head, J., Smith, D., Zuber,
710
M., 2015. Constraints on the distribution and thickness of mare basalts and
711
cryptomare from grail. In: Lunar and Planetary Science Conference. Vol. 46.
712
p. 2691.
AN US
709
Jozwiak, L., Head III, J., Neumann, G., Wilson, L., 2016. Observational con-
714
straints on the identification of shallow lunar magmatism: insights from
715
floor-fractured craters. Icarus .
M
713
Kiefer, W. S., Macke, R. J., Britt, D. T., Irving, A. J., Consolmagno, G. J.,
717
2012. The density and porosity of lunar rocks. Geophysical Research Letters
718
39 (7).
PT
ED
716
Lemoine, F. G., Goossens, S., Sabaka, T. J., Nicholas, J. B., Mazarico, E.,
720
Rowlands, D. D., Loomis, B. D., Chinn, D. S., Neumann, G. A., Smith,
721
D. E., et al., 2014. Grgm900c: A degree 900 lunar gravity model from grail
AC
CE
719
722
primary and extended mission data. Geophysical research letters 41 (10),
723
3382–3389.
724
725
726
Melosh, H., 2011. Planetary Surface Processes. Cambridge University Press, Cambridge, UK New York. Milbury, C., Johnson, B., Melosh, H., Collins, G., Blair, D., Soderblom, J., 44
ACCEPTED MANUSCRIPT
727
Nimmo, F., Bierson, C., Phillips, R., Zuber, M., 2015. Preimpact porosity
728
controls the gravity signature of lunar craters. Geophysical Research Letters
729
.
730
731
Mueller, I., 1964. The Horizontal Gradients of Gravity in Geodesy. Ohio State University, Department of Geodetic Science. Neumann, G. A., Zuber, M. T., Wieczorek, M. A., Head, J. W., Baker, D. M.,
733
Solomon, S. C., Smith, D. E., Lemoine, F. G., Mazarico, E., Sabaka, T. J.,
734
et al., 2015. Lunar impact basins revealed by gravity recovery and interior
735
laboratory measurements. Science advances 1 (9), e1500852.
CR IP T
732
Pike, R. J., 1980. Control of crater morphology by gravity and target type
737
- Mars, earth, moon. In: Bedini, S. A. (Ed.), Lunar and Planetary Science
738
Conference Proceedings. Vol. 11 of Lunar and Planetary Science Conference
739
Proceedings. pp. 2159–2189.
AN US
736
Robinson, M. S., Ashley, J. W., Boyd, A. K., Wagner, R. V., Speyerer, E. J.,
741
Hawke, B. R., Hiesinger, H., van der Bogert, C. H., Aug. 2012. Confirmation
742
of sublunarean voids and thin layering in mare deposits. Planetary and
743
Space Science 69, 385–391.
ED
M
740
Wieczorek, M. A., Neumann, G. A., Nimmo, F., Kiefer, W. S., Taylor, G. J.,
745
Melosh, H. J., Phillips, R. J., Solomon, S. C., Andrews-Hanna, J. C., Asmar,
746
S. W., Konopliv, A. S., Lemoine, F. G., Smith, D. E., Watkins, M. M.,
747
Williams, J. G., Zuber, M. T., 2013. The crust of the moon as seen by
748
GRAIL. Science 339 (6120), 671–675.
AC
CE
PT
744
749
Zuber, M. T., Smith, D. E., Lehman, D. H., Hoffman, T. L., Asmar,
750
S. W., Watkins, M. M., jan 2013. Gravity recovery and interior labora-
751
tory (GRAIL): Mapping the lunar interior from crust to core. Space Sci
752
Rev 178 (1), 3–24.
753
Zuber, M. T., Smith, D. E., Watkins, M. M., Asmar, S. W., Konopliv, A. S., 45
ACCEPTED MANUSCRIPT Lemoine, F. G., Melosh, H. J., Neumann, G. A., Phillips, R. J., Solomon,
755
S. C., Wieczorek, M. A., Williams, J. G., Goossens, S. J., Kruizinga, G.,
756
Mazarico, E., Park, R. S., Yuan, D.-N., dec 2012. Gravity field of the moon
757
from the gravity recovery and interior laboratory (GRAIL) mission. Science
758
339 (6120), 668–671.
AC
CE
PT
ED
M
AN US
CR IP T
754
46