1 2 Geophysical Research Letters 3 Supporting Information for 4 5 Regolith stratigraphy at the Chang’E-3 landing site as seen by Lunar Penetrating Radar 6 Wenzhe Fa1*, Meng-Hua Zhu2, Tiantian Liu1, Jeffery B. Plescia3 7 8 9 10 1Institute of Remote Sensing and Geographical Information System, Peking University, Beijing, China. Science Institute, Macau University of Science and Technology, Macau, China. 3 The Johns Hopkins University Applied Physics Laboratory, Laurel, Maryland, USA. *E-mail: wzfa@pku.edu.cn. 2Space 11 12 13 14 15 16 17 18 19 20 Contents of this file Text S1 to S5 Figures S1 to S11 21 22 Introduction 23 This supporting information provides: 24 25 26 27 28 29 30 31 32 33 S1: Optical images of the CE-3 landing site; S2: Regolith thickness over the CE-3 landing site; S3: Ages of small craters; S4: LPR data processing and subsurface structure identification; S5: Turnover times of the regolith; Figures S1-S11: Figures for text S1-S5. 34 35 S1. Optical images of the CE-3 landing site In the optical image, there are two distinct geologic units in the northern region of Mare 36 Imbrium (Fig. S1a), the dark I22 and the bright I5 [Hiesinger et al., 2000]. The CE-3 landing 37 site is in the northern portion of unit I22, about 10-15 km from the boundary between the 38 geologic units I22 and I5 (Fig. S1b). The CE-3 landing site is about 50 m from the rim of the 39 500 m CE-3 crater (Fig. S1c). In the Clementine Ultraviolet/Visible (UVVIS) image (Fig. S1d) 40 [Eliason et al., 1999], unit I22 shows a very blue color, indicating a high titanium abundance, 41 whereas unit I5 has a red color, corresponding to a low titanium abundance. 42 43 S2. Regolith thickness over the CE-3 landing region 44 In this study, regolith thickness over the CE-3 landing region (unit I22) is estimated using 45 two methods: morphology of small fresh craters [Quaide and Oberbeck, 1968; Fa et al., 46 2014] and the excavation depth of craters with the threshold diameter between rocky and 47 rock-free craters [Wilcox et al., 2005]. 48 49 High-resolution optical images show that the morphology of small fresh craters can be 50 classified into four types: normal, central mound, flat-bottomed, and concentric [Quaide and 51 Oberbeck, 1968]. Laboratory impact experiments found that these crater types could be 52 used to estimate regolith layer thickness [Quaide and Oberbeck, 1968]: a normal crater 53 forms when D/d (D: crater diameter; d: regolith thickness) is smaller than a value between 54 3.8 and 4.2, a concentric crater forms when D/d is larger than a value between 8 and 10, and 55 a central mound or flat-bottomed crater forms when D/d has a value between that forms a 56 normal and a concentric crater. 57 58 Here we used two LROC Narrow Angle Cameras (NACs) images (M1144922100LE and 59 M1144922100RE) with a spatial resolution of ~1.6 m/pixel and an incidence angle of 56.5° 60 for crater counting. The study area is 8.1 km × 12.2 km (Fig. S2). In total, 1557 small fresh 61 craters with diameters from 11 to 145 m were counted, of which 1189 are normal craters, 62 334 are flat-bottomed craters, and 34 are concentric craters (Fig. S3). Because the 63 illumination angle (33.5°) is larger than typical value of regolith repose angle (31°), shadow 64 effect correction is not required in regolith thickness estimation [Quaide and Oberbeck, 65 1968; Fa et al., 2014]. Using the relation between regolith thickness and crater morphology, 66 Fig. S4 shows the cumulative distributions of regolith thickness estimated from normal 67 (red) and concentric (blue) craters. From Fig. S4, the median regolith thickness for the mare 68 around the CE-3 landing site is about 8 m. 69 70 Based on rock population, small craters can be classified as distinctly blocky (type A), a few 71 blocks (type B), and no blocks (type C) (Fig. S5) [Wilcox et al., 2005]. In a cratering event, a 72 crater of type A or B will have penetrated the entire regolith layer and excavated the 73 coherent substrate to some extent, bringing at least a few rocks (arrows in Fig. S5a and b) to 74 the surface. In contrast, a crater of type C excavates only regolith, and therefore no blocks 75 from the substrate occur on the surface. Regolith breccias may form in any of these cases. 76 Regolith thickness can be estimated using the crater excavation depth with the threshold 77 diameter between type A and B craters and type C craters. Here, the excavation depth is 78 simply assumed as 1/10 of the crater diameter [Pike, 1974]. 79 80 The same region in the same LROC NAC images was selected for rocky and rock-free crater 81 counting. In total, 318 craters with diameters from 30 to 770 m were counted, of which 33 82 are type A, 73 are type B, and 212 are type C (Fig. S6). Fig. S7 shows the relative 83 distribution of the counted A, B, and C type craters, and the threshold diameter between 84 type A and B and type C is about 85 m. Therefore, the regolith thickness is estimated to be 85 about 8.5 m. 86 87 S3. Ages of small craters 88 According to Basilevsky [1976], the lifetime of a crater can be estimated as 89 𝑇(𝐷) = { 2.5𝐷, 𝐷 < 160 m 8𝐷 − 900 , 𝐷 > 160 m (1) 90 where T is the crater lifetime in Myr, and D is the diameter of a crater in m. Using crater 91 morphology, density of surface rocks, inner wall slope, and depth/diameter ratio, a typical 92 bowl-shaped crater can be classified into morphological class A, AB, B, BC, and C. Each 93 morphological class represents a different degradation stage, with class A representing 94 young fresh and class C representing old and degraded. For each morphological class, a 95 lifetime fraction is further assigned (A+AB: 0-3%; B: 3-20%; BC: 20-50%; C: 50-100%). 96 Then, the age of a crater is estimated as the product of the lifetime and its fraction. The 97 accuracy of the estimated age is about 30% [Basilevsky, 1976]. 98 99 Using the LROC DTM, slope of the inner wall and depth/diameter ratio of the CE-3 crater are 100 calculated to be about 18° and 1/10, respectively. In addition, images from the CE-3 rover- 101 mounted panorama cameras and LROC show that there are abundant rocks in the interior 102 region, whereas almost no meter-scale rocks beyond the crater rim. Therefore, the CE-3 103 crater is classified as morphologic class B, and it is very close to class AB. The lifetime of a 104 500 m crater is about 3.1 Gyr, and hence the age of the CE-3 crater is 3.1 Gyr × 3-20% = 93- 105 620 Myr. Since morphological prominence of the CE-3 crater is very close to class AB, its age 106 can be approximated as 100 Myr with an uncertainty of 30 Myr. 107 108 S4. LPR data processing and subsurface structure identification 109 For LPR high frequency channel, the transceiver antenna is chosen as bow tie antenna [Fang 110 et al., 2014]. The bow tie antenna is designed with a half-ellipse shape, and is loaded by 111 parallel resistors in order to absorb the reflected current at the edge of the antenna. 112 Laboratory measurements of the antenna system [Fang et al., 2014] show that the antenna 113 can produce a good waveform and antenna ringing can be effectively reduced. Ground 114 experiments of the LPR proto [Zhang et al., 2014] show that the LPR performance can meet 115 the designed specifications (detection ability and range resolution). In this study, the LPR 116 raw data were processed through repetitive observation removal, noise reduction, 117 horizontal band removal, band-pass filtering (Fig. 2a), compensation of geometrical 118 spreading and dielectric attenuation (Fig. 2b), and range migration (Fig. 2c) in sequence. 119 There were LPR observations when the Yutu rover stopped, and repetitive observation 120 removal is applied to remove the nearly identical LPR traces and other erroneous traces. 121 Horizontal band removal is conducted to remove the influence of complex coupling effect 122 between the antenna and the lunar surface. To reduce the lower and higher frequency 123 noises, band-pass filtering is further applied. Geometrical spreading makes the amplitude of 124 the LPR echo fall off in proportion to the traveled distance, and dielectric loss can attenuate 125 LPR signals when propagate in the lunar subsurface. Compensation of geometrical 126 spreading and dielectric attenuation is applied to balance the amplitudes of the LPR echoes 127 at different depth. A time-domain diffraction stacking range migration method is further 128 used to reduce the clutters caused by subsurface discrete targets. 129 130 In the band-pass filtering, a sine-squared function is used 0, 1 sin2 [𝜋2(𝑓𝑓−𝑓 )], 2 −𝑓1 1, 𝐻(𝑓) = 131 sin2 [𝜋( 𝑓4−𝑓 )], { 𝑓 < 𝑓1 𝑓1 ≤ 𝑓 ≤ 𝑓2 𝑓2 ≤ 𝑓 ≤ 𝑓3 2 𝑓4 −𝑓3 𝑓3 ≤ 𝑓 ≤ 𝑓4 0, 𝑓 > 𝑓4 (2) 132 where the lower and higher cutoff frequencies are selected as 𝑓1 = 50 MHz and 𝑓4 = 650 133 MHz, and 𝑓3 and 𝑓4 are chosen as 150 and 500 MHz. All the LPR data were processed using a 134 set of homebrew codes. 135 136 Fig. 2a shows the observations after noise reduction, and is suitable for near surface (<3 m) 137 structure identification according to echo strength. Because of the geometrical spreading 138 and dielectric attenuation, the strength of radar echoes decreases dramatically with depth. 139 These two losses were compensated so deep subsurface echoes become stronger and more 140 obvious (Fig. 2b). As the CE-3 crater is young and the landing site is only ~50 m from its 141 rim, the LPR survey line is within the continuous ejecta region. Buried rocks in the ejecta 142 make the observed LPR image chaotic, and cause additional scattering loss of the incident 143 radar waves. Range migration in time domain (aka diffraction stacking) is further applied to 144 reduce clutters caused by buried rocks and make deep subsurface feature clearly (Fig. 2c). 145 146 The high-frequency antennas are about 0.273 m from the surface, and the one-way travel 147 time is offset in the processed LPR images so that zero time (and also zero depth) 148 corresponds to the level of the surface. Rectification of surface topography in LPR image is 149 impossible because no topographic data at centimeter scale are available. In Fig. 2a, depths 150 of some radar returns are smaller than zero. Compensation of dielectric attenuation and 151 range migration were conducted for those echoes with depth larger than zero. Therefore, in 152 Fig. 2b and c, there are no echoes for depth smaller than zero. At the navigation points (C, D, 153 E/S1, S2, S3, and S4), there were no LPR observations, and the rover may have moved a 154 small distance. As a result, when these individual observations were mosaicked, there are 155 unavoidable vertical strips at these navigation points (Fig. 2b and c). However, these 156 artifacts do not affect the identified subsurface structure in any significant way. The lateral 157 distance in Figs. 2 and S8 is calculated from the coordinates of the navigation points in Fig. 158 1, which were given in the LPR raw data. 159 160 Subsurface structure was identified from all the three images in Fig. 2. Given the large 161 dynamic range of the received radar echoes, the brightness and contrast of local image were 162 adjusted in some cases in order to confidently identify subsurface features. The reworked 163 zone and the ejecta were identified in Fig. 2a and b. Though there are numerous, chaotic 164 layers in the reworked zone and the ejecta, lateral continuity of layers in the reworked zone 165 is generally larger than those in the ejecta (Fig. S8a and b). Layers in the reworked zone 166 were identified mainly based on Fig. S8a. The ejecta layer contains more subsurface features 167 than the paleoregolith. In addition, as seen in Fig. S8b, radar echo strength from the ejecta is 168 much stronger than that from the paleoregolith. The interface between the paleoregolith 169 and the transition zone is identified by the increase in the strength of the radar echoes (cyan 170 lines in Fig. 2b as a conservative base, and reflector 2 as the deepest base). It should be 171 noted that from the LPR image, the transition from one zone to another is gradational. Since 172 these zones were mainly formed from continuous impacts of meteoroids, there should be no 173 abrupt transition from one zone to another from the viewpoint of geology. 174 175 In GPR images, the echo of a discrete scatterer (e.g., rock, void) is a hyperbola [Daniels, 176 2004]. In this study, a rock is identified as the vertex of a hyperbola (Fig. S9). Given to the 177 large dynamic range of the received echo, the contrast and brightness of local LPR image 178 were adjusted in order to identify a rock confidently. In Fig. 3a, the dots represent the 179 identified rocks. The number of rocks at a given depth along the survey line is counted from 180 Fig. 3a, then normalized by the length of the survey line (85.5 m), and finally, the histogram 181 of rock number as a function of depth (Fig. 4b) is obtained by binning the normalized 182 number of rocks along depth with a width of 0.5 m. 183 184 In GPR image, the trace of the received radar echo from a buried rock is a hyperbola (Fig. 185 S9d). The shape of a hyperbola depends on the antenna height, the depth of buried rock, and 186 the real part of the dielectric permittivity [Daniels, 2004]. Fig. S8a-c shows four typical 187 hyperbolas in the LPR images that are produced by buried rocks. Fig. S8d show the 188 simulated traces of LPR echoes with dielectric permittivity of 1.0, 2.0, 3.0, 5.0 and 7.0 for the 189 subsurface, where the antenna height is 0.273 m, and the depth of the buried rock is 1.5 m. 190 By comparing the shape of the observed LPR echo with the simulated trace [Arcone et al., 191 1998], real part of the dielectric permittivity can be estimated. The real part of the dielectric 192 permittivity of the subsurface materials is estimated as 2.4 for the hyperbola in Fig. S8a, 2.5 193 for Fig. S8b, and 3.4 for Fig. S8c. In total, we analyzed more than 50 typical hyperbolas in the 194 LPR images, and the results show that the real part of the dielectric permittivity is 3.2 on 195 average, corresponding to a bulk density of 1.8 g/cm3. 196 197 S5. Turnover times of the regolith 198 Regolith turnover rate at different depth is estimated based on the model proposed by 199 Shoemaker et al. [1970] and Gault et al. [1974], and summarized by Melosh [1989]. If a 200 crater of diameter D disturbs the regolith to a depth of 𝑑 = 𝐷⁄4, then the average length of 201 time between subsequent turnovers roughly equals the time that must elapse before the 202 fractional area covered by craters of diameter 𝐷 = 4𝑑 or larger equals one. Overturn time, 203 tovr, of the regolith, can be written as, 204 2(𝑏−2) 𝑑 𝑏−2 𝑡𝑜𝑣𝑟 ≈ 3𝜋𝑏𝑐 (ℎ ) 𝑒𝑞 𝑡𝑠𝑢𝑟 𝑒𝑞 (3) 205 where tsur is the surface age, b and c are the power and the coefficient in the cumulative 206 distribution of crater number, 𝑁 = 𝑐𝐷 −𝑏 , and ceq is the coefficient in the crater number in 207 the equilibrium population,𝑁𝑒𝑞 = 𝑐𝑒𝑞 𝐷 −2. b usually varies from 2.9 to 3.4 for craters smaller 208 than 4 km in diameter [Gault, 1970]. If equilibrium occurs at a crater density of about four 209 percent of geometry saturation, then ceq is equal to 0.046 [Melosh, 1989]. heq is the 210 maximum likely thickness, and can be calculated as 211 (4) ℎ𝑒𝑞 = 𝐷𝑒𝑞 ⁄4 212 where Deq is the equilibrium diameter that can be determined directly from the cumulative 213 size-frequency distribution of craters. In equation (3), it is assumed that craters formed in 214 loose regolith. According to Shoemaker et al. [1970], craters formed in loose regolith are 215 about six times larger than craters formed in solid bedrock. As a result, overturn time in 216 loose regolith is about one sixth as that in solid bedrock. 217 218 219 Number of times the regolith has been overturned over the surface age, tsur, is 3𝜋𝑏𝑐𝑒𝑞 𝑑 −(𝑏−2) 𝑁𝑜𝑣𝑟 ≈ 2(𝑏−2) (ℎ ) 𝑒𝑞 (5) 220 Fig. S10a shows the counted craters for a region surrounding the CE-3 crater that is beyond 221 the continuous ejecta, and Fig. S10b shows the counted craters for the CE-3 landing region. 222 Using the counted craters, equilibrium diameters over the geologic unit I22 (Fig. S10a) and 223 the CE-3 landing region (white box in Fig. S10b) are identified as 161 m and 3.85 m, 224 respectively. Fig. S11 shows turnover time and number of turnovers for the newly formed 225 regolith and the paleoregolith, where the ages of the newly formed regolith and the 226 paleoregolith are 100 Myr and 2.96 Gyr, respectively, and b=3.4. As can be seen, for the 227 newly formed regolith, the ejecta have been overturned about 315, 12.5, and 0.5 times at 228 depths of 0.01, 0.1 and 1 m. Within the first meter, number of turnovers for the 229 paleoregolith is about 186 times larger than that of the newly formed regolith at the same 230 depth. 231 232 References 233 Arcone, S. A., D. E. Lawson, A. J. Delaney, J. C. Strasser, and J. D. Strasser (1998), Ground- 234 penetrating radar reflection profiling of groundwater and bedrock in an area of 235 discontinuous permafrost, Geophysics, 63, 1573–1584. 236 237 238 239 240 241 Basilevsky, A. T. (1976), On the evolution rate of small craters, Proc. Lunar Sci. Conf., 7, 1005–1020. Daniels, D. (2004), Ground Penetrating Radar, The Institution of Electrical Engineers, London. Eliason, E., C. Isbell, E. Lee, T. Becker, L. Gaddis, A. McEwen, and M. Robinson (1999), The Clementine UVVIS Global Lunar Mosaic, Lunar and Planetary Institute, Houston. 242 Fa, W., T. Liu, M.-H. Zhu, and J. Haruyama (2014), Regolith thickness over Sinus Iridum: 243 Results from morphology and size-frequency distribution of small impact craters, J. 244 Geophys. Res., 119, 1914–1935. 245 246 247 248 249 250 251 252 253 254 255 256 Fang, G. Y., et al. (2014), The Lunar Penetrating Radar (LPR) onboard Chang’E-3 mission, Research in Astron. Astrophys., 14, 1607–1622. Gault, D. E. (1970), Saturation and equilibrium conditions for impact cratering on the lunar surface: Criteria and implications, Radio Sci., 5, 273–291. Gault, D. E., F. Hörz, D. E. Brownlee, and J. B. Hartung (1974), Mixing of the lunar regolith, Proc. Lunar Sci. Conf., 7, 2365–2386. Hiesinger, H., R. Jaumann, G. Neukum, and J. W. Head (2000), Ages of mare basalts on the lunar nearside, J. Geophys. Res., 105, 29239–29275. Melosh, H. J. (1989), Impact Cratering: A Geologic Process, Oxford University Press, New York. Pike, R. J. (1974), Depth/Diameter relations of fresh lunar craters: Revision from spacecraft data, Geophys. Res. Lett., 1, 291–294. 257 258 259 260 261 262 Quaide, W. L., and V. R. Oberbeck (1968), Thickness determinations of the lunar surface layer from lunar impact craters, J. Geophys. Res., 73, 5247–5270. Shoemaker, E. M., et al. (1970), Origin of the lunar regolith at Tranquility Base, Proc. Apollo 11 Lunar Sci. Conf., 3, 2399–2412. Wilcox, B. B., M. S. Robinson, P. C. Thomas, and B. R. Hawke (2005), Constraints on the depth and variability of the lunar regolith, Meteorit. Planet. Sci., 40, 695–710. 263 Zhang, H. B., et al. (2014), Performance evaluation of lunar penetrating radar onboard the 264 rover of CE-3 probe based on results from ground experiments, Research in Astron. 265 Astrophys., 14, 1633–1641. 266 267 268 269 Figure S1. The CE-3 landing region as shown in LROC images: (a) Mare Imbrium, (b) the 270 northern Mare Imbrium, and (c) the CE-3 landing region (M102285549LE). (d) The 271 Clementine false-color composite map (red: 750 nm/415 nm; green: 750 nm/950 nm; and 272 blue: 415 nm/750 nm) for the northern Mare Imbrium. 273 274 Figure S2. All the counted small fresh craters over a region (8.1 km × 12.2 km) in geologic 275 unit I22 surrounding the CE-3 landing site, where the red, green, and blue dots indicate 276 normal, flat-bottomed, and concentric craters, respectively. 277 278 Fig. S3. (a) Percentage of normal (red), flat-bottomed (green), and concentric (blue) craters 279 as a function of crater diameter for all the counted craters. Relative distributions of (b) 280 normal, (c) flat-bottomed, and (d) concentric craters as a function of crater diameter. 281 282 283 Figure S4. Cumulative distribution of regolith thickness (percentage of area with regolith 284 thickness smaller than a given value) estimated from normal (red) and concentric (blue) 285 craters. The horizontal error bars show the uncertainty in the regolith thickness, and the 286 vertical error bars indicate the uncertainty in the percentage of the area for a given regolith 287 thickness, which is calculated based on the one standard deviation (1σ) of the counted 288 crater numbers with a Poisson distribution. 289 290 Figure S5. Three types of small craters with different rock populations: (a) a distinctly 291 blocky crater (type A), (b) a crater with a few blocks (type B), and (c) a rock-free crater 292 (type C). The white arrows indicate the location of surface rocks. 293 294 295 Figure S6. A LROC NAC image (mosaicked from M1144922100LE and M1144922100RE) 296 showing the counted rocky and rock-free craters over the region surrounding the CE-3 297 crater. Red dots represent rocky craters, green dots indicate craters with a few rocks, and 298 blue dots are rock-free craters. 299 300 Figure S7. The distribution of blocky types A, B, and C craters as a function of diameter for 301 the region surrounding the CE-3 crater. The threshold diameter between type A and B 302 craters and type C craters is 80-90 m. 303 304 305 Figure S8. (a) Layers in the reworked zone as in the LPR raw data after horizontal band 306 removal and band-pass filtering. (b) The ejecta layer in the LPR image after compensation 307 for geometrical spreading and dielectric attenuation, and the cyan line indicates the base of 308 the ejecta. The white boxes show four typical hyperbolas produced by buried rocks. (c) The 309 paleoregolith as seen in the LPR image processed after range migration, and the red arrows 310 indicate four reflectors within the paleoregolith. 311 312 Figure S9. (a-c) LPR images showing typical hyperbolas produced by buried rocks (white 313 boxes in Fig. S8b). The hyperbolas are indicated by the black arrows. (b) The traces of GPR 314 echoes from a buried rock with different dielectric permittivity of the lunar subsurface 315 material. The depth of buried rock is 1.5 m, and the height of the antenna is 0.273 m. 316 317 Figure S10. The counted craters over (a) a region surrounding the CE-3 crater that is 318 beyond the continuous ejecta, and (b) the CE-3 landing region. 319 320 321 Figure S11. (a) Regolith turnover time as a function of depth for the newly formed regolith 322 (the black line) and paleoregolith (the red line). (b) Numbers of turnover for the newly 323 formed regolith (the black line) and the paleoregolith (the red line).