1. Technology Capability Development for Identification and Interpretation of Martian Craters and Climate History Technology Capability Development for Identification and Interpretation of Martian Craters and Climate History P.I.: Clark R. Chapman (15) CoI’s: William J. Merline (15) Steve W. Dellenback (10) Michael P. Rigney (10) Michael J. Magee (10) Collaborator: Prof. James Head III (Dept. Geology, Brown Univ.) P.I.: Clark R. Chapman (15) CoI’s: William J. Merline (15) Steve W. Dellenback (10) Michael P. Rigney (10) Michael J. Magee (10) Collaborator: Prof. James Head III (Dept. Geology, Brown Univ.) Proposal submitted 11 March 2002 to: Southwest Research Initiative for Mars (SwIM)

2. Main Features of Proposal Strongly interdivisional collaboration to develop latent talents to be competitive in NASA’s future Mars exploration programs Focus on science issues (geological history of water) central to NASA interests in Mars Strong technological component (Artificial Intelligence and Data Mining techniques) Involvement with prominent Collaborator Modest cost (<$95K) sufficient to bring us up to speed and make us competitive

3. Science Background: Crater Degradation and Water on Mars First evidence for “rainfall” on Mars was from “river” valley networks (run-off vs. sapping) Craters provide baseline initial conditions to assess subsequent modification of topography Variation in crater degradation classes (fresh to very degraded) as function of crater diameter led to hypothesis of “obliteration episode” on Mars, contemporaneous with valley networks Martian craters show much greater variety than on the Moon; voluminous data not yet studied Sequence of degraded to fresh craters

4. A wide variety of surface modication processes... Many kinds of processes, many different signatures Lacustrine, oceanic Volcanic Aeolian (dunes, storms) Tectonic Glacial Rivers and streams Subterranean flow Creep Isostatic adjustment Superimposed cratering Etc., etc. Wind Volcanism No modifica- tion at all!

5. Degrees of Terrain Softening None Some A Lot

6. Mars Crater Data Bases Original analysis for craters >8 km diam. from Mariner 9 images only (critical sizes: 10 - 50 km ) Vast additional imaging sets, with much higher resolution, better coverage: Viking Orbiter imaging MOC imaging (wide and narrow angle camera), MGS New THEMIS images (vis and IR) from Mars Odyssey Potential future missions Data cry out for cataloging, analysis of morphology classes…but tedious effort has inhibited progress, demanding a new approach

7. Background on Capabilities in Div. 15 and Div. 10 Analysis of Martian cratering statistics by Chapman in 1970s…new data need analysis Previously developed A.I. crater recognition algorithms, applied to simple lunar craters, by Div. 15 (Merline/Chapman/et al.), collaborating with JPL…enhanced techniques required for much more complex Mars craters Expertise (although in non-planetary science applications) in feature recognition and classification technologies by Div. 10…can now apply technical experience to Mars

8. Dr. Chapman’s 1970s Research on Mars Cratering Developed hypothesis of erosional “episode” Research finished be- fore availabil- ity of Viking Orb. images Credibility... but need to “get up to speed” on current data, issues Annual Revs. Earth Planet Sci. (1977) Icarus (1974)

9. Boulder Office/JPL Past Work on A.I. Identification of Lunar Craters Singular Value Decomposition and “template” approaches About 80% reliability for simple cratered terrains; but we desire >90% reliability for more complex Martian terrains Output of template approach to analysis of simple lunar scene. Identified craters are color-coded (yellow = most reliable). Blue circles (slightly offset to upper left) are human identifications and sizings.

10. Simple Cratered Surfaces… and then there is Mars! asteroid Gaspra planet Mercury the Moon Mars

11. Division 10 Expertise in A.I. And Data Mining Expertise in “expert systems” A.I. techniques to approximating human perception/decision- making processes Expertise in Data Mining, which can be applied to Mars crater data base to gain insights Past applications of circle-enhancing Hough transforms to identify wheels, tools, fiducial marks, rivets, etc. Illustration of Decoupled Circular Hough Transform method

12. Goals, Objectives, General Approach Combine relevant, as-yet-unconsolidated skills and experience in Divs. 10 & 15 to address new research opportunities concerning Mars Leverage capabilities to begin addressing fundamental questions concerning Martian geological history, role and location of water Two arenas for development: Further develop crater identification/classification in Martian context Evaluate and augment current lunar algorithms for Mars Parallel use of Hough circular transforms, contrast enhancement Evaluate how to interpret Mars crater forms in terms of processes (in collaboration with Prof. James Head)

13. Tasks and Expected Accomplishments Develop new, improved crater detection/ID technology to address wealth of current/future Mars imaging data Div. 10 develops filters (e.g. contrast enhance) to pre- process Mars images for analysis by Div. 15 algorithms Div. 10 develops, in parallel, alternative (Hough transform) methods to test on Martian images Divs. 10 & 15 collaborate on developing morphological classification criteria that are practical, geologically useful Div. 10 studies “next steps” in classifying, data mining Develop interpretive methods in 3 study areas Div. 15 & Prof. Head select areas, develop morph. indices Link indices to previous fresh-to-degraded crater studies Preliminary interpretations of geological history in 3 areas

14. An Example: Contrast Enhancement In the case of the lunar work previously done, craters are deep, bowl-shaped depressions Martian features are degraded, shallow and may have confusing surface shadings Raw image (upper left) shows subdued features, which become much more prominent after contrast enhancement (lower left) Other pre-processing filters may improve algorithm success

15. Opportunities; Benefits to the Institute Current NASA Data Analysis/Research programs: we’ll have tools, credibility to propose Mars Data Analysis Program (MDAP): prospects to study Viking, MGS/MOC, and Mars Odyssey/THEMIS images New Mars Fundamental Research Program Enchanced prospects to be selected for future Mars missions (specifics undergoing review, but NASA commitment to Mars is clear) Mars Express, Mars Exploration Rovers (Participating Scientist and/or follow-on research prospects) Mars Reconnaissance Orbiter, sample return missions Specific landing site selection opportunities Non-Mars prospects: e.g. Earth remote-sensing

16. Program Plan and Schedule 1 May 02 1 Aug 02 1 Nov 02 1 Feb 03 30 Apr 03 1. Crater ID/Classification Explore interfaces (Divs. 10&15) Preprocessing images to enhance existing algorithms Parallel (Hough) methods; development of classification criteria, data mining Analysis of “next steps” in classification, data mining 2. Interpretive Methodologies Explore interfaces (Divs. 10&15) Selection of 3 Mars study locales Develop morphological indices, link with prior studies Preliminary geological interpretation of 3 study areas; scientific publication

17. Personnel; Project Organization Overall project lead: Dr. Clark R. Chapman (15) Task 1 (crater identification technology) Dr. William J. Merline (15; assisted by Mr. Brian Enke) Dr. Michael P. Rigney (10) Dr. Michael J. Magee (10) Dr. Steve W. Dellenback (10; lead on “next steps” task) Task 2 (scientific interpretation methodology) Dr. Clark R. Chapman (15) Prof. James Head (collaborator, Brown University) Dr. William J. Merline (15) Other: trips to facilitate collaboration; no new equipment. Total budget: $92,544

18. Past IR&D Work by P.I. and CoI’s Dr. Chapman: 1 previous IR, $93K, 1999-2001; “white paper” received prominent, international discussion and use; facilitated small grant continuation; other contacts being pursued Dr. Merline: 1 previous QL, $28K, 9/98 - 1/99; contributed to much subsequent NASA/NSF funding of asteroid satellite searches Div. 10 CoI’s: 2 previous QL’s, 2 previous IR’s totalling ~$356K, range from being currently underway to leading to patents, aerospace/ government/industry opportunities

19. Conclusion Cost-effective way to develop latent skills, enhance inter-Divisional collaboration, be in good position for major roles in NASA’s chief Solar System program: Mars exploration Strong technology component married with strong science component -- ideal for NASA Prepares us to address some of the most compelling issues in planetary science (role of water on Mars) using state-of-the-art Artificial Intelligence techniques

20. The End: Back-up slides to follow

21. Modelling how time-variable erosion affects crater morphologies Total fresh slight moderate heavily …degraded Obliteration time history (Chapman, 1974)

22. Signature of “episode” in morphologic statistics (a) sequence f,s,m,h indicates incomplete- ness due to resolution (b) Mars data (Jones) (c) Sequence h,m,s,f indicates obliteration episode: smaller craters are most affected, largest ones least affected Among intermediate sized craters (tens of km diameter), smaller ones (~10 km) are most heavily degraded, largest ones (>30 km) only modestly degraded or nearly fresh.

23. Absolute age of obliteration Early Mariner 9 interpretations had obliteration tied to the declining early cratering flux. Depending on calibration of absolute ages, the obliteration could have happened toward the end of the decline (a), or considerably later (b). But the important conclusion is that it was decoupled from the end of the early bombardment.

24. LHB on Mars? One Mars meteorite (and only one: ALH84001) is very old and has an Ar-Ar age of ~3.9 Ga: statistics of ONE (Ash et al., 1996) Meteorite degassing ages are very “spread out” compared with lunar LHB and somewhat spread out compared with lunar rocks Evidence is dissimilar! Different impact histories or Different selection biases LHB Lunar rock de- gassing ages Mars  Kring & Cohen 2002