Characterizing Rock Abundance at ExoMars Landing Site ... · Characterizing Rock Abundance at...
1
Characterizing Rock Abundance at ExoMars Landing Site Candidates E. Sefton-Nash1 ([email protected]), J. C. Bridges2, J. L. Vago1, L. Kissick3,4, F. Butcher5, P. Donnelly2, J. D. Piercy2, D. Louizeau6, L. Lorenzoni1, P. M. Grindrod3,4, and M. Balme5. The European Space Agency’s ExoMars programme comprises the Trace Gas Orbiter and Rover/Surface Platform missions. At the 3 rd ExoMars Rover Landing Site Work- shop (October 2015, ESTEC, Noordwijk, The Netherlands) 4 candidate sites were pre- sented and assessed based on thier scientific merit and compliance with engineer- ing and environmental constraints [1]. We here present preliminary work to gather representative float rock abundances for each candidate site. Float rocks are individual fragments not associated with contin- uous outcrop. Introduction Manual rock count method REFERENCES [1] LSSWG, Vago, J. L. and Rodionov, D. (2015) ExoMars 3rd Landing Site Selection Workshop, Final Report, ESA Ref.: EXM-RM-REP-ESA-00005. [2] ESA ExoMars Landing Site Working Group (2013), ‘ExoMars 2018 Landing Site Selection User’s Manual’, ESA Ref.: EXM-SCI-LSS-ESA/IKI-003. [3] McEwen, A. S. et al. (2007), Mars Reconnaissance Orbiter’s High Resolution Imaging Science Experiment (HiRISE), J. Geophys. Res., 112 (E05S02). [4] Golombek, M. and Rapp, D. (1997), Size-frequency distributions of rocks on Mars and Earth analog sites: Implications for future landed missions, J. Geophys. Res. 102 (E2) p. 4117–4129. [5] Golombek, M. P. et al. (2008), Size-frequency distributions of rocks on the northern plains of Mars with special reference to Phoenix landing surfaces, J. Geophys. Res. 113, E00A09. [6] Golombek, M. P. et al. (1997), Selection of the Mars Pathfinder landing site, J. Geophys. Res. 102 (E2), p. 3967–3988. [7] Christensen, P. R. (1986), The Spatial Distribution of Rocks on Mars, Icarus 68, p. 217–238. [8] Binder A. B. et al. (1977),The Geology of the Viking Lander 1 Site, J. Geophys. Res. 82 (28) p. 4439 – 4451. [9] Moore, H. J. and Keller, J. M. (1990), Surface material maps of Viking Landing Sites on Mars, Reports of Plan. Geology & Geophys. Program, NASA Tech Memo. TM 4210, p. 533–535. [10] Audouard, J. et al. (2014), Mars surface thermal inertia and heterogeneities from OMEGA/MEX, Icarus 233, p. 194–213. [11] Tao, Y. and Muller, J.-P. (2016), A novel method for surface exploration: Super-resolution restoration of Mars repeat-pass orbital imagery, Plan. & Space Sci. 121, p. 103 - 114. 6. Laboratoire de Géologie de Lyon, Université Claude Bernard Lyon 1, France. 1. European Space Agency, ESTEC 2200 AG Noordwijk, The Netherlands. 2. Space Research Centre, Depart- ment of Physics and Astronomy, Uni- versity of Leicester, UK. 3. Department of Earth and Plane- tary Sciences, Birkbeck, University of London, UK. 5. Department of Physical Sciences, The Open University, Milton Keynes, UK. CENTRE FOR PLANETARY SCIENCES CPS 4. UCL/Birkbeck Centre for Plane- tary Sciences, University College London, UK. Rock abundance model Results Discussion/Conclusions Rock abundance derived using anisothermality in IR datasets requires assumptions re- garding the SFD, and the number of thermally distinct components that contribute to the signal [7]. Use of thermal RA therefore requires constraint of thermal contributions from differ- ent features and terrain types. Thermal RA may be an upper limit for F from float rocks, because thermal contributions from float rocks are only a part of the signal, with both rough and smooth outcrop also contributing [8, 9]. The extent to which outcrop poses a hazard/obstacle to the lander and rover depends on the slope distribution. 500 x 500 m count areas were defined in HiRISE map-projected JP2s. Locations were selected to adequately sample geologic diversity and maximize spatial coverage within landing ellipses. Repeat counts 46°W 46°W 47°W 47°W 12°N 12°N 10 Km 24° W 24° W 25° W 25° W 18° N 18° N 10 Km 17°W 17°W 18°W 18°W 19°W 19°W 23°N 23°N 22°N 22°N 20 Km 11°W 11°W 12°W 12°W 8°N 8°N 10 Km Aram Dorsum Hypanis Mawrth Valles Oxia Planum Coverage by rock count areas in selected 2018 landing ellipses (green). Count areas are outlined in red and are accompanied by percentage areal coverage by fea- tures identified as float rocks. Background is THEMIS nighttime IR controlled mosaics. HiRISE images used for counting are overlaid. 0.007 0.024 0.005 0.022 0.536 0.287 0.107 0.015 0.168 0.723 0.087 0.039 0.380 0.000 0.296 0.004 0.201 0.054 0.000 0.066 0.005 0.005 0.149 0.014 0.000 0.117 0.570 0.052 0.915 0.040 0.862 0.222 1.330 0.256 0.031 0.110 0.017 0.020 0.102 0.095 0.117 0.000 0.000 0.000 0.000 0.047 0.000 0.000 0.008 0.012 0.050 0.027 0.021 0.006 0.109 0.027 0.630 0.000 0.630 0.027 0.003 0.045 0.110 0.024 0.052 0.009 0.005 Solar incidence Excess area Manual count Rock Shadow Fraction of surface covered by rocks of diameter ≥ D Counter 1 Counter 2 Counter 3 Human counters counted features that they identified as float rocks that cast shadows in the solar illumina- tion direction by drawing circles that completely en- close features. Areal coverage fraction, F, was calculated for each count area. F is an upper limit on actual areal coverage by resolv- able features, because 1. Excess area is included in the calculation and 2. Intersection of circles is possible for close features. Counts repeated in the same area by different counters illustrate differences in con- sistency of identification, e.g. for ESP_039747_1880 in Aram Dorsum: Discrepancies between counters occured most frequently in three cases: 1. In areas of shadow casting rocky outcrop, where isolated float rocks are difficult to distinguish from outcrop. 2. At diameters close to the resolution limit, and at large diamaters, where outcrop can be confused with partially buried boulders. 3. In areas where local topography (e.g. wrinkle ridges) imposes changes in solar il- lumination angle, and therefore shadow length, of features. Humans identify more features at lower solar illumination angles because shadow length is increased. Solar inci- dence angle can be fixed for a flat scene due to time of day, but may also vary across a scene with topography. HiRISE resolution affects human capability to identify features at D < 1.5 m, but possibly up to 2 m. We observe large disparity between RA that is manually-derived (almost always < 1%) and thermally-derived (often several %) [9]. This is expected because thermal RA does not exclude contributions from small, unresolved rocks and flat (non-shadow casting) high thermal inertia materials. Site Count Areas Total Area Unique Counts Features Aram Dorsum 23 5.75 27 13923 Hypanis Vallis 16 4.00 19 20386 Mawrth Vallis 13 3.25 9 2812 Oxia Planum 16 4.00 19 2122 TOTAL 68 17.00 74 39243 Rock abundance (RA) is often quantified as cumulative areal fraction (CFA), here represented as F, of a surface covered with rocks of diameter ≥ D. The canonical rock distribution model [4] was established by fitting an exponential curve to the CFA occupied by rocks at the Viking landing sites. The small rock population (D ≤ 1 m) observed in the immediate vicinity of the near- and far-field image mosaics was found to fit well to the relationship: where, and k is the rock abundance factor. k can be retrieved for a given rock size distribution by minimizing residuals in a fit (we employ a least squares fit). Typically the model is used in conjunction with thermally-derived RA, orbital im- agery, geological inferences and available in-situ results to infer rock SFDs [5, 6], to constrain the abundance of small rocks that are unresolved in orbital images. Caveats The model suffers from a lack of validation at large rock diameters due to the small area visible from landed craft. Common metrics of rock abundance such as CFA do not constrain rock shape or state of burial. Statistics for manually counted features. Cumulative fractional area (CFA) for each site compared to model runs, labeled by thier total percentage areal coverage. Kernel density function of 39243 features identified as rocks at all sites. The effects of reso- lution limit are most evident at D < 1.5 m but may have an effect on SFDs up to D ~ 2 m. Thermally-derived rock abundance Observed CFAs appear not to match well with any single model curve. It remains to be demonstrated that the family of curves employed in [4] can represent rock SFDs at sites with geologic histories dissimilar to those that formed the Viking landing sites. Despite caveats regarding human consistency, variable illumination conditions, and classification, rock SFDs for float rocks may still be used for relative ranking of hazard- ous rock abundance at each site, because isolated float rocks are some of the most easily distinguishable and uniqiuely hazardous of surface features. The use of high resolution thermal data [10], super-resolution imaging techniques [11] and detailed geological inferences may be key in ongoing assessment of ExoMars Rover candidate landing sites. Rock abundance is directly relevant to lander safety (the Landing Module’s clearance is 35 cm [2]) and rover traversability. MRO HiRISE [3] images allow identification of features down to ~1 m in size, but ambiguity increases below ~2 m.
Transcript of Characterizing Rock Abundance at ExoMars Landing Site ... · Characterizing Rock Abundance at...
Characterizing Rock Abundance at ExoMars Landing Site CandidatesE. Sefton-Nash1 ([email protected]), J. C. Bridges2, J. L. Vago1, L. Kissick3,4, F. Butcher5, P. Donnelly2, J. D. Piercy2, D. Louizeau6, L. Lorenzoni1, P. M. Grindrod3,4, and M. Balme5.
The European Space Agency’s ExoMars programme comprises the Trace Gas Orbiter and Rover/Surface Platform missions. At the 3rd ExoMars Rover Landing Site Work-shop (October 2015, ESTEC, Noordwijk, The Netherlands) 4 candidate sites were pre-sented and assessed based on thier scienti�c merit and compliance with engineer-ing and environmental constraints [1].
We here present preliminary work to gather representative �oat rock abundances for each candidate site. Float rocks are individual fragments not associated with contin-uous outcrop.
Introduction
Manual rock count method
REFERENCES[1] LSSWG, Vago, J. L. and Rodionov, D. (2015) ExoMars 3rd Landing Site Selection Workshop, Final Report, ESA Ref.: EXM-RM-REP-ESA-00005.[2] ESA ExoMars Landing Site Working Group (2013), ‘ExoMars 2018 Landing Site Selection User’s Manual’, ESA Ref.: EXM-SCI-LSS-ESA/IKI-003.[3] McEwen, A. S. et al. (2007), Mars Reconnaissance Orbiter’s High Resolution Imaging Science Experiment (HiRISE), J. Geophys. Res., 112 (E05S02).[4] Golombek, M. and Rapp, D. (1997), Size-frequency distributions of rocks on Mars and Earth analog sites: Implications for future landed missions, J. Geophys. Res. 102 (E2) p. 4117–4129.[5] Golombek, M. P. et al. (2008), Size-frequency distributions of rocks on the northern plains of Mars with special reference to Phoenix landing surfaces, J. Geophys. Res. 113, E00A09. [6] Golombek, M. P. et al. (1997), Selection of the Mars Pathfinder landing site, J. Geophys. Res. 102 (E2), p. 3967–3988.[7] Christensen, P. R. (1986), The Spatial Distribution of Rocks on Mars, Icarus 68, p. 217–238. [8] Binder A. B. et al. (1977),The Geology of the Viking Lander 1 Site, J. Geophys. Res. 82 (28) p. 4439 – 4451.[9] Moore, H. J. and Keller, J. M. (1990), Surface material maps of Viking Landing Sites on Mars, Reports of Plan. Geology & Geophys. Program, NASA Tech Memo. TM 4210, p. 533–535. [10] Audouard, J. et al. (2014), Mars surface thermal inertia and heterogeneities from OMEGA/MEX, Icarus 233, p. 194–213.[11] Tao, Y. and Muller, J.-P. (2016), A novel method for surface exploration: Super-resolution restoration of Mars repeat-pass orbital imagery, Plan. & Space Sci. 121, p. 103 - 114.
6. Laboratoire de Géologie de Lyon, Université Claude Bernard Lyon 1, France.
1. European Space Agency, ESTEC 2200 AG Noordwijk, The Netherlands.
2. Space Research Centre, Depart-ment of Physics and Astronomy, Uni-versity of Leicester, UK.
3. Department of Earth and Plane-tary Sciences, Birkbeck, University of London, UK.
5. Department of Physical Sciences, The Open University, Milton Keynes, UK.
C E N T R E F O RP L A N E T A R YS C I E N C E S
CPS
4. UCL/Birkbeck Centre for Plane-tary Sciences, University College London, UK.
Rock abundance model
Results
Discussion/Conclusions
Rock abundance derived using anisothermality in IR datasets requires assumptions re-garding the SFD, and the number of thermally distinct components that contribute to the signal [7].
Use of thermal RA therefore requires constraint of thermal contributions from di�er-ent features and terrain types. Thermal RA may be an upper limit for F from �oat rocks, because thermal contributions from �oat rocks are only a part of the signal, with both rough and smooth outcrop also contributing [8, 9]. The extent to which outcrop poses a hazard/obstacle to the lander and rover depends on the slope distribution.
500 x 500 m count areas were de�ned in HiRISE map-projected JP2s. Locations were selected to adequately sample geologic diversity and maximize spatial coverage within landing ellipses.
Repeat counts
46°W
46°W
47°W
47°W
12°N 12°N
10Km
24° W
24° W
25° W
25° W
18° N 18° N
10Km
17°W
17°W
18°W
18°W
19°W
19°W
23°N 23°N
22°N 22°N
20Km
11°W
11°W
12°W
12°W
8°N 8°N
10Km
Aram Dorsum Hypanis
Mawrth Valles Oxia Planum
Coverage by rock count areas in selected 2018 landing ellipses (green). Count areas are outlined in red and are accompanied by percentage areal coverage by fea-tures identi�ed as �oat rocks. Background is THEMIS nighttime IR controlled mosaics. HiRISE images used for counting are overlaid.
0.0070.024
0.005 0.022
0.536
0.287
0.1070.015
0.168
0.7230.087
0.0390.380
0.0000.296
0.004
0.2010.054
0.000
0.0660.005
0.005
0.149 0.014
0.000
0.117
0.570
0.052 0.9150.040
0.8620.222
1.3300.256
0.031
0.1100.017
0.020
0.102
0.095
0.117
0.0000.000
0.000
0.000
0.0470.000
0.0000.008
0.012
0.050
0.027
0.021
0.0060.109
0.027
0.630 0.0000.630
0.027
0.0030.045 0.110
0.024 0.052
0.009
0.005
Solar incidence
Excess area
Manual count
Rock
Shadow
Frac
tion
of s
urfa
ce c
over
ed b
y ro
cks
of d
iam
eter
≥ D
Counter 1 Counter 2 Counter 3
Human counters counted features that they identi�ed as �oat rocks that cast shadows in the solar illumina-tion direction by drawing circles that completely en-close features.
Areal coverage fraction, F, was calculated for each count area.
F is an upper limit on actual areal coverage by resolv-able features, because 1. Excess area is included in the calculation and 2. Intersection of circles is possible for close features.
Counts repeated in the same area by di�erent counters illustrate di�erences in con-sistency of identi�cation, e.g. for ESP_039747_1880 in Aram Dorsum:
Discrepancies between counters occured most frequently in three cases:
1. In areas of shadow casting rocky outcrop, where isolated �oat rocks are di�cult to distinguish from outcrop.
2. At diameters close to the resolution limit, and at large diamaters, where outcrop can be confused with partially buried boulders.
3. In areas where local topography (e.g. wrinkle ridges) imposes changes in solar il-lumination angle, and therefore shadow length, of features.
Humans identify more features at lower solar illumination angles because shadow length is increased. Solar inci-dence angle can be �xed for a �at scene due to time of day, but may also vary across a scene with topography.
HiRISE resolution a�ects human capability to identify features at D < 1.5 m, but possibly up to 2 m.
We observe large disparity between RA that is manually-derived (almost always < 1%) and thermally-derived (often several %) [9]. This is expected because thermal RA does not exclude contributions from small, unresolved rocks and �at (non-shadow casting) high thermal inertia materials.
Site Count Areas Total Area Unique Counts FeaturesAram Dorsum 23 5.75 27 13923Hypanis Vallis 16 4.00 19 20386Mawrth Vallis 13 3.25 9 2812Oxia Planum 16 4.00 19 2122
TOTAL 68 17.00 74 39243
Rock abundance (RA) is often quanti�ed as cumulative areal fraction (CFA), here represented as F, of a surface covered with rocks of diameter ≥ D.
The canonical rock distribution model [4] was established by �tting an exponential curve to the CFA occupied by rocks at the Viking landing sites.
The small rock population (D ≤ 1 m) observed in the immediate vicinity of the near- and far-�eld image mosaics was found to �t well to the relationship:
where, and k is the rock abundance factor.
k can be retrieved for a given rock size distribution by minimizing residuals in a �t (we employ a least squares �t).
Typically the model is used in conjunction with thermally-derived RA, orbital im-agery, geological inferences and available in-situ results to infer rock SFDs [5, 6], to constrain the abundance of small rocks that are unresolved in orbital images.
Caveats
The model su�ers from a lack of validation at large rock diameters due to the small area visible from landed craft.
Common metrics of rock abundance such as CFA do not constrain rock shape or state of burial.
Statistics for manually counted features.
Cumulative fractional area (CFA) for each site compared to model runs, labeled by thier total percentage areal coverage.
Kernel density function of 39243 features identi�ed as rocks at all sites. The e�ects of reso-lution limit are most evident at D < 1.5 m but may have an e�ect on SFDs up to D ~ 2 m.
Thermally-derived rock abundance
Observed CFAs appear not to match well with any single model curve. It remains to be demonstrated that the family of curves employed in [4] can represent rock SFDs at sites with geologic histories dissimilar to those that formed the Viking landing sites.
Despite caveats regarding human consistency, variable illumination conditions, and classi�cation, rock SFDs for �oat rocks may still be used for relative ranking of hazard-ous rock abundance at each site, because isolated �oat rocks are some of the most easily distinguishable and uniqiuely hazardous of surface features.
The use of high resolution thermal data [10], super-resolution imaging techniques [11] and detailed geological inferences may be key in ongoing assessment of ExoMars Rover candidate landing sites.
Rock abundance is directly relevant to lander safety (the Landing Module’s clearance is 35 cm [2]) and rover traversability. MRO HiRISE [3] images allow identi�cation of features down to ~1 m in size, but ambiguity increases below ~2 m.
