The Impact Winds of Mars

Home , Oudemans (crater), Resen (crater)

Stephanie Nicole Quintana Bouchey

B.S., Colorado School of Mines, 2011 Sc.M., Brown University, 2013

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Earth, Environmental and Planetary Sciences at Brown University

Providence, Rhode Island

May 2017

This dissertation by Stephanie N. Q. Bouchey is accepted in its present form by the

Department of Earth, Environmental and Planetary Sciences as satisfying the dissertation requirements for the degree of Doctor of Philosophy.

Date Peter H. Schultz, Brown University Advisor

Recommended to the Graduate Council

Date John F. Mustard, Brown University Reader

Date Amanda H. Lynch, Brown University Reader

Date Karen M. Fischer, Brown University Reader

Date Wesley A. Watters, Wellesley College Reader

Approved by the Graduate Council

Date Andrew G. Campbell, Brown University Dean of the Graduate School

iii

STEPHANIE NICOLE QUINTANA BOUCHEY Curriculum Vitae

Department of Earth, Environmental, and Home: Planetary Sciences Brown University 212 3rd Street, Apt. 4A 324 Brook Street, Box 1846 Troy, NY 12180 USA Providence, RI 02912 USA P: +1.720.425.1845 P: +1.401.863.3549 E: [email protected] E: [email protected] http://snqb.strikingly.com/

EDUCATION

Brown University, Providence, RI | PhD, Geological Sciences Advisor: Peter Schultz 05/2017 Thesis: The Impact-Winds of Mars

Brown University, Providence, RI | Sc.M, Geological Sciences Advisor: Peter Schultz Thesis: Methods to determine impact melting and vaporization using the 05/2013 shock physics analysis package, CTH

Colorado School of Mines, Golden, CO | BS, Engineering with Mechanical Specialty 05/2011

Tohoku University, Sendai, Japan | Mechanical and Aerospace Engineering 2007-2008 Yoshida-Nagatani Space Robotics Laboratory

EXPERIENCE Internships Sandia National Laboratories, Albuquerque, NM Supervisor: David Crawford  Performed verification projects with the shock physics code, CTH though 1D, 2D, and 3D studies of impact melt and vapor Summers determination 2012-2013  Tested newly developed Brittle Damage with Localized Thermal Softening (BDL) strength and fracture model for geologic materials

Lunar and Planetary Institute, Houston, TX Supervisor: David Kring  Evaluated and proposed possible robotic and human landing sites on Summer the Moon based upon science objectives outlined in the National 2011 Research Council’s Scientific Context for Exploration of the Moon

Colorado School of Mines, Golden, CO Supervisor: Christopher Dreyer  Developed a new lunar regolith simulant, CSM-CL series Summer  Reverse-engineered and redesigned a sorting trommel for regolith 2010 excavation

NASA Kennedy Space Center, Cape Canaveral, FL Supervisor: Paul Hintze  Studied regolith simulants for a lunar launch/landing pad design project Summer  Melted and sintered regolith simulant and studied results with a 2009 scanning electron microscope and energy dispersive x-ray spectroscopy

NASA Marshall Space Flight Center, Huntsville, AL Supervisors: Michael LaPointe and Barbara Cohen 01-05/  Performed risk mitigation research for the International Lunar 2009 Network Mission and analyzed seismology and heat flow experiment deployment methods

Teaching Summer@Brown Co-Instructor, Brown University, Providence, RI  Prepared and taught Habitable Worlds: Possible Places for Life in Summers the Solar System and Beyond, a week-long college preparatory 2013-2016 course for high school students about exoplanets and habitability

Geological Society of America’s GeoCorps America Program, U.S. Forest Service, Meeker, CO Supervisor: Victoria Houser, Blanco Ranger District  Designed and led public education and outreach about cave and Summer karst research, bats, and White-Nose Syndrome 2016  Developed educational materials: posters, short news articles, public presentations, and high school science lessons

Volunteer Science Teacher, 2nd Grade, Vartan Gregorian Elementary School, Providence, RI  Designed and taught sciences lessons and designed worksheets in 2013-2015 accordance with Rhode Island State Science Standards and Next Generation Science Standards

Rhode Island Space Grant Consortium Fellow, Brown University, Providence, RI  Guest lecturer at the Lincoln School (9th grade physics classes) 2012-2013 about the Mars Science Laboratory (Curiosity Rover)  Tutor at Nathan Bishop Middle School for Science Olympiad v

Teaching Assistant, Brown University, Providence, RI  TA for Geology 81: Planetary Geology Spring  Attended lectures and discussion sessions, assisted in in-class 2012 demonstrations, assisted students when needed

PROJECTS

Effects of Impact-Generated Winds on Mars Advisor: Peter Schultz  Use computer modeling with the CTH shock physics analysis Current package and laboratory tests at the Ames Vertical Gun Range to explore regional and global impact-induced morphology

Verification and Validation of Impact Melt/Vapor Determination in CTH Advisor: David Crawford  Utilized material properties in the CTH models, performed 3D 06/2012- studies, verified and validated results with previous studies and 06/2016 laboratory experiments

NASA Reduced Gravity Flight Testbed Senior Design Project Advisor: Paul van Susante 08/2010-  Reverse engineered a ground-based experiment to determine forces 05/2011 on an excavation tool for a reduced gravity flight of the testbed

Wheel Design for Planetary Rover Advisor: Kazuya Yoshida 11/2007-  Designed, built, and implemented a force-sensing wheel for a 07/2008 prototype planetary rover

Lunar Excavation Rover Development, Lunar Ventures Advisor: Masami Nakagawa 11/2006-  Worked as part of a team that designed and built a prototype lunar 08/2007 excavation rover for the 8th Continent Lunar Ventures business plan competition

SKILLS

Computer Skills Proficient with: CTH; MATLAB; JMARS; Adobe PhotoShop; Inkscape; KaleidaGraph; Microsoft Word, Excel, PowerPoint; Windows XP, 7, and 10 operating systems. Familiar with: ArcGIS; ENVI; SolidWorks; LabVIEW; C++; Linux operating system

Applied Skills Proficient at: Data analysis; development and execution of experiments at the NASA Ames Vertical Gun Range; technical writing and presentations; archival research; proposal writing; writing for popular audiences; university teaching; teaching non-traditional students Familiar with: NASA mission design; robotics research and design; JEOL Scanning Electron Microscope operation; lunar regolith simulant research, analysis, and design

PUBLICATIONS

Book Chapter Crites, S., A. Przepiorka, S. Quintana, C. Santiago, T. Trabucchi (2012), Science concept 7: The Moon is a natural laboratory for regolith processes and weathering on anhydrous airless bodies. In A Global Lunar Landing Site Study to Provide the Scientific Context for Exploration of the Moon, D.A. Kring and D.D. Durda (eds.), LPI Contribution No. 1694, Lunar and Planetary Institute, Houston, TX, pp. 413-475.

Refereed Publications Schultz, P.H., S.N. Quintana (2017), Impact-Generated Winds on Mars, Icarus 292, pp. 86-101. Quintana, S.N., D.A. Crawford, P.H. Schultz (2015), Analysis of impact melt and vapor production in CTH for planetary applications, Procedia Engineering 103, pp. 499-506. Hintze, P. E. and S. Quintana (2013), Building a Lunar or Martian Launch Pad with in situ Materials: Recent Laboratory and Field Studies, Journal of Aerospace Engineering 26(1), pp. 134-142.

Selected Conference Abstracts Quintana, S.N., and P.H. Schultz, Model Results for Impact-Winds on Mars, LPSC 48, Abstract no. 1123. Quintana, S.N., P.H. Schultz, S.S. Horowitz (2016), New Experiments in Martian Impact Vapor-Induced Wind Streak Analysis, LPSC 47, Abstract no. 1553. Quintana, S.N., and P.H. Schultz (2016), A Global Distribution of Impact-Wind Streak Craters on Mars, LPSC 47, Abstract no. 1548. Quintana, S.N., and P.H. Schultz (2015), Using Laboratory Experiments and Computational Modeling to Explain Impact-Related Winds on Mars, Bridging the Gap III (Germany), Abstract no. 1040. Quintana, S.N., P.H. Schultz, S.S. Horowitz (2015), Experimental Results Supporting an Impact-Related Blast Wind Formation Mechanism for Some Wind Streaks on Mars, LPSC 46, Abstract no. 2469. Quintana, S.N., P.H. Schultz, D.A. Crawford (2015), Target Strength as an Important Consideration for Low-Speed Impacts, LPSC 46, Abstract no. 2727. vii

Quintana, S.N., P.H. Schultz (2014), The Formation of Crater-Related Blast Wind Streaks on Mars, LPSC 45, Abstract no. 1971. Quintana, S.N., and P.H. Schultz (2014), Impact Melt on Asteroids: New Insights from One-Dimensional Simulations, Hayabusa 2014: 2nd Symposium on Solar System Materials (Japan). Quintana, S.N., P.H. Schultz (2013), Modeling Impact Blast Winds on Mars: The Formation of Permanent Wind Streaks, GSA Annual Meeting, Paper No. 200-9. Quintana, S., D. Crawford, P.H. Schultz (2013), Verification of Impact Melt and Vapor Determination Methods in CTH, LPSC 44, Abstract no. 1733. Schultz, P.H., S. Quintana (2013), Impact Blast Wind Scouring on Mars, LPSC 44, Abstract no. 2697. Quintana, S., D. Crawford, P.H. Schultz (2012), Verification of Impact Melt and Vapor Determination Methods in CTH, GSA Annual Meeting, Paper No. 202-12. Quintana S., S. Crites, A. Przepiórka, C. Santiago, T. Trabucchi, D. A. Kring (2012), Moscoviense Basin: A Landing Site to Study Science Goals Associated with Lunar Regolith Processes and Space Weathering, LPSC 43, Abstract no. 1215. Crites S. S. Quintana, A. Przepiórka, C. Santiago, T. Trabucchi, D. A. Kring (2012), Lunar Landing Sites that will Enhance our Understanding of Regolith Modification Processes, LPSC 43, Abstract no. 1086.

Publications for Popular Audiences Bouchey, S., O. Patick, V. Houser (2017), A New Cave Interpretation Program at Spring Cave, White River National Forest, NSS News 75 (4), pp. 22-23. Bouchey, S., O. Patick, V. Houser (2016), A New Cave Interpretation Program at Spring Cave, White River National Forest, Beneath the Forest, 9 (2), pp. 14-17.

INVITED TALKS

EAPS Planetary Lunch Colloquium Series, Massachusetts Institute of Technology, MA | Exploring Impact-Generated Winds on Mars 04/2017 Research Seminar, Johns-Hopkins Applied Physics Laboratory, MD | Exploring Impact-Generated Winds on Mars though Laboratory 10/2016 Experiments and Computational Modeling Meeker Public Library Seminar, Meeker, CO | Impact Cratering on Mars and the Formation of Peculiar Wind Streaks 08/2016 Meeker Public Library Seminar, Meeker, CO | Caves, Bats, and White- Nose Syndrome 07/2016 Frontiers of Astronomy, Wheaton College, MA | Exploring Planetary Impact Craters: Small Scale Experiments to Supercomputer Models 02/2016

viii

Research Seminar, U.S. Army Research Laboratory, Aberdeen Proving 01/2016 Ground, MD | Impact Melt and Vapor Production in CTH for Planetary Applications Introductory Astronomy, Wheaton College, MA | The Quest for Water on Mars and Beyond 11/2015 Earth Talks Seminar Series, Winona State University, MN | Rosetta, Comets, and Impacts. Oh My! 02/2015 Introductory Astronomy, Wheaton College, MA | Rosetta, Comets, and 11/2014 Impacts. Oh My!

HONORS AND AWARDS

Lunar and Planetary Institute Career Development Award 2017 Ford Foundation Fellowship, Honorable Mention 2016 GeoClub Award for Service to the Department, Brown University DEEPS 2016 Elected to the Brown University Chapter of Sigma Xi honor society 2014 Archambault Award for Teaching Excellence, 2nd Prize 2014 Colorado Engineering Council Certificate of Merit 2011

GRANTS AND FELLOWSHIPS National Science Foundation Graduate Research Fellowship Program 2013-2016 Fellowship (tuition + stipend) Paul G. Benedum Graduate Travel and Research Fund for Geological 2015 Sciences Award (international travel award) Barringer Crater Company Travel Grant (international travel award) 2015 Hitachi Travel Award (international travel award) 2014 NASA Rhode Island Space Grant Consortium Fellowship, (tuition + 2012-2013 stipend) Geological Society of America Northeast Section Student Travel Grant 2012, 2013

PROFESSIONAL DEVELOPMENT

Professional DEEPS-GWiSE Professional Development Seminar (Networking and the 10/2016 Earth Science Women’s Network), Brown University DEEPS Professional Development Seminar (Work/Life Balance), Brown 02/2015 University DEEPS Professional Development Seminar (Figure Generation), Brown 11/2014 University Job Searching with Social Media Workshop, Brown University 01/2014 ArcGIS for Planetary Applications (Independent Study), Brown 2012 University CTH Training Class, Sandia National Laboratories, Albuquerque, NM 11/2011

Teaching Diversity and Inclusion Professional Development Lunch Series, Brown 2016-2017 University Teaching at Teaching-Intensive Institutions Workshop, University of 10/2016 Massachusetts Boston Earth Educators’ Rendezvous, Preparing for an Academic Career 07/2016 Workshop, University of Wisconsin-Madison Sheridan Center for Teaching and Learning Certificate II (Course 2014-2015 Design), Brown University Teaching at Teaching-Intensive Institutions Workshop, Marlborough, MA 10/2014 Sheridan Center for Teaching and Learning Certificate I (Reflective 2013-2014 Teaching), Brown University Various workshops on teaching, teaching diverse audiences, and InTeGrate modules, Brown University Sheridan Center and Carlton 2013-2016 College

SERVICE

Brown University Department of Earth, Environmental, and Planetary 2016 Sciences, Organized the first annual department-wide Graduate Student Conference Brown University Department of Geological Sciences, incoming graduate 2013- student mentor present Planetary Geology Division, Dwornik Award Judge (for undergraduate 2014, 2016 presentations) Brown University Sheridan Center for Teaching and Learning, graduate 2015-2016 student liaison Brown University, Graduate Women in Science and Engineering, 2013-2016 founding member and coordinator Cambridge University Press, book proposal review 2015 Brown University School of Professional Studies, Summer@Brown 2015 instructor orientation panelist Brown University Department of Geological Sciences, Graduate Women 2013-2015 in Science and Engineering, department representative Brown University Department of Geological Sciences GeoClub, treasurer 2012-2013

PROFESSIONAL MEMBERSHIPS

Association for Women Geoscientists since 2013 National Speleological Society since 2013 Geological Society of America since 2012 American Institute of Aeronautics and Astronautics since 2011

Acknowledgements

I’ve often commented, especially to prospective students, that the grad students make the department. The relationships I’ve formed, even with just a fraction of the wonderful people who have been through Brown during my time here, make up for the difficulties of research and all the times grad school beats you down. This is also true in a broader sense in the network of colleagues, mentors, and friends I’ve made throughout my graduate career and who have helped me become the scientist I am today. And, of course, my friends and family have been an integral part of getting me through this dissertation, emotionally. I could never properly express how much the people listed below have influenced me, encouraged me, and supported me. I will try, but know that there is a lot of emotion behind this, too, that I can’t express in words.

I’ll begin first and foremost with my advisor, Pete. Pete’s enthusiasm for impact cratering and his ability to see something new and interesting every time we’re looking at a Martian landscape (or even just sitting in the Planetary Data Center) drew me to want to work with him. Of course, the Ames Vertical Gun Range was a perk, too. But really,

Pete’s enthusiasm is what started it all. His intuition about cratering processes is astounding. It seems that there are few people who get genuinely giddy about their job, and Pete at the AVGR is one of them. It’s not a job for Pete; it’s his life (well, half of it anyway!).

I’ve come to find that academia might not be the place for me. This has been a bumpy, frustrating six years. So I want to thank Pete for guiding me through my PhD. I loved doing experiments, so thank you for four awesome trips to the gun. Thank you for encouraging me to learn modeling. Combined with the experiments, I’m now a more xi well-rounded scientist than I would have been had I chosen to study somewhere other than Brown. And thanks also for giving me the freedom to pursue summer internships and to work remotely in order to live with my husband.

David Crawford has been something like my second advisor throughout graduate school, and I want to give him special thanks. After a short introductory class in CTH,

Dave invited me to work with him at Sandia National Laboratories two summers in a row, and allowed me to continue working on interesting projects as a No-Fee Consultant.

Thank you so much for this opportunity, which I believe led to the job I will have when I graduate. A huge chunk of this dissertation would not have been possible without your help. Thank you for your extreme patience with my ineptitudes in everything from using

CTH, to thermodynamics, to simple math. Thank you (and Jenn) for inviting me into your home (twice! And one of those times was soon after baby Isaac arrived, no less) so that we could work together one-on-one. While most of that work is not represented here, it gave me an exponentially better understanding of CTH, and I hope we are able to continue working on these problems.

I haven’t worked as closely with other faculty as some other graduate students do.

Nonetheless, my time in graduate school was inspired by many of them through classes, meetings, and department events. I want to specifically thank Amy Barr, Karen Fischer,

Don Forsyth, Meredith Hastings, Amanda Lynch, and Jack Mustard. Amy, thank you for being a fantastic teacher, for encouraging me, for making me feel like a colleague, and for being an awesome mentor and role model. Honestly, without you, I’m not sure if I would have stuck with it. And Meredith, thank you for being a great mentor and for encouraging my efforts and involvement with GWiSE and the Graduate Student

xii

Conference. Amanda, thank you for taking time out of your busy schedule to meet with me and give me advice. Jack, Karen, and Don, thank you for serving on my committee and meeting with me to talk about my research, asking me questions, and helping me clarify my work.

Other than faculty, I have a long list of staff to whom I would like to express my appreciation. First of all, all the Lincoln Field administrative assistants: Jess Murihead,

Melissa Shein, and Karen Leap-Canis. I also want to acknowledge Nancy Ciminelli,

Gloria Correra, Anne Côté, Pat Davey, Melissa De Augustinis, Nancy Fjeldheim, Peter

Neivert, and Lisa Sheehan. On the technical side, a big thanks Joseph Boesenberg, Lynn

Carlson, and last but certainly not least, Bill Collins (thanks for many years of smiles and laughs with GeoChron). I also want to thank the Lincoln Field custodian, Jimmy and the groundskeeper, Robert for various fun and spontaneous conversations.

I want to thank the whole “Pete Family,” who went before me. With special recognition, I want to call out Jenn Anderson, Kelly Wrobel, Seiji Sugita, and Olivier

Barnouin for sharing your research, network, and sanity with me. And separately, I want to thank Terik Daly, Megan Bruck Syal, and Angela Stickle, who overlapped with me, advised me, taught me, and shared in the experience of being a Pete-student along with me. Terik, I particularly want to thank you for being the best officemate I could ask for, for always being willing to lend me an ear (or a shoulder), and for sharing Habitable

Worlds with me. I have greatly enjoyed teaching with you and learning from you.

I also want to acknowledge the AVGR crew for helping me collect four spectacular sets of data: Don Bowling, Charles Cornelison, Alfredo Perez, Adam Parish,

xiii and Jon-Pierre Wiens. Thanks also to Seth Horowitz and China Blue Wong for your help at the gun and fun dinners at your home afterwards.

And now I come to my grad school friends. I can be a shy person, and I am grateful that you saw past that and got me to come out of my shell. You all mean so much to me. I’ve interacted with everyone listed below in different, and many times overlapping, contexts from Ren Faires, field trips, and froyo trips; to teaching, rock climbing, and concerts… You all helped keep me sane. Thank you to Mike Bramble,

Leah Cheek, Sydney Clark, Tim Goudge, Rebecca Greenberger, Noah Hammond, Chris

Havlin, Erica Jawin, Taka Kanaya, Hillary O’Brien, Chelsea Parker, Kei Shimizu, J.R.

Skok, Vivian Sun, Jenny Whitten, Janette Wilson, Diane Woodruff, and Yinsui Zheng.

To Jillian Bohnker, Steven Ahn, and Ben Johnson, thanks for befriending me and continuing to hang out even when I stopped taking engineering classes. Thank you to my

LPI buddies, Sarah Crites, Dave Blair, and Carrie Roberts for a fun, but intense, summer of lunar landing site studies and many discussions afterwards. And thank you Taylor and

Rachel Dotson, Karin Patzke, Laura Rabinow, and Colin and Yurie Garvey, my friends from RPI for accepting and including me without question even though I wasn’t part of your school or department.

Mary Peterson, Emily Hopper, my other giraffe Lauren Jozwiak, and Tess

Caswell, I can’t express what your friendship has meant to me. You are like my family.

You’ve helped me grow; you’ve challenged me, and have been my support when I needed it. Best of all, you kept me laughing and reminded me that there’s a life outside of grad school.

xiv

Support from my pre-grad school friends and mentors has also been essential.

Thank you to Jackie Barnes, Jake and Maria Gogue (and Elena and Joaquin, too!), Mark

Lu and Minh Thy Le, and Céclie Bopp for being such wonderful friends. Jake and

Maria, thank you especially for the many many times you’ve let Mike and me practically live at your house. To my mentors, Mr. and Mrs. Channell, Angel Abbud-Madrid,

Masami Nakagawa, Barbara Cohen, and Dr. Hintze, thank you for setting me on this path.

Finally, I’d like to acknowledge my family. Mom and Dad, thank you for the support and encouragement that got me here, that made me competitive and stubborn enough to stick through an engineering degree and then a PhD. Thank you for supporting my decision to come to grad school, even though you didn’t really know what that entailed or meant, or what kind of career I could possibly get from studying impact cratering. And last but definitely not least, thank you to Mike, the love of my life. Words cannot do it justice. You are my world. There is no way that I would be who I am today without you.

You’ve pushed me to be the best I can be and you encourage me endlessly. Thank you for marrying me, for taking care of me (always, but particularly these last few months), and building me up when life gets overwhelming. I am looking forward to starting the next chapter of our lives together.

Preface

1. A Brief History

Impact cratering is one of the few geological processes that is ubiquitous throughout the solid bodies of the Solar System, from the rocky terrestrial planets to the icy moons of the giant planets, and even asteroids. Yet the impact origin of rimmed circular depressions was not always clear. Initial observations focused on the Moon.

Galileo recognized these pervasive landforms simply as spots and recorded them in sketches in the early 1600s. Hooke later explored the same features and, unlike Galileo, proposed an origin for them. He considered celestial impacts, but dismissed it because he could not conceive of a source for the impactors, and rather settled on an origin by volcanism (Green, 1965).

The discovery of Ceres in 1801 and the link between comets and meteors (meteor showers) later that century, however, represented a viable source for impactors (Schultz,

1998). Yet even with a catalog of meteor events and minor planets, a conflict of origins for lunar craters persisted. Conflicts between disciplines, added to the effects of authority and perception at the time, led to a long-standing contention of crater origins, which was dominated for some time by volcanism (Schultz, 1998). Simply, volcanoes could be observed and studied, whereas no impact crater had yet been identified as such on Earth.

The conflict, particularly between astronomers and geologists, was described in an amusing passage by W. M. Davis in his biography of G. K. Gilbert (e.g., Kopal, 2012, p.

293):

…It has been remarked that the majority of astronomers explain the craters on the Moon by volcanic eruption – that is, by an essentially geological process – while a considerable number of geologists are inclined to explain them by the impact of bodies falling upon the Moon – that is, by an essentially astronomical process. xvi

This suggests that each group of scientists finds the craters so difficult to explain by processes with which they are professionally familiar that they prefer to take recourse to a process belonging to another field that than their own, with which they are probably imperfectly acquainted and with which they therefore feel freer to take liberties. (Davis, 1926).

While G. K. Gilbert was not the first to propose an impact origin for the craters on the Moon, he did develop a critical shift in thought and analysis when he proposed an impact origin in 1892 (Gilbert, 1893). Later, high explosives (and nuclear detonations even later) provided an important missing link to understanding the physical process of impact cratering, which Ives noted in 1919 (Ives, 1919). A few years later, the discovery of a crater near Odessa, TX finally linked the formation to a meteorite for the first time

(Barringer Jr, 1928).

2. Atmospheres and Impact Cratering

Even though the scars of new and ancient celestial collisions are found on nearly every solid body in the Solar System, the effects of impact cratering on different bodies varies as much as the bodies themselves. Target characteristics shape impact craters and affect everything from heat production (e.g., O’Keefe and Ahrens, 1977; Schultz et al.,

1998; Housen and Holsapple, 2003) to final crater shape (e.g., Gault et al., 1974;

Holsapple, 1987). For example, large, basin-sized impacts in ice are not the same as those in rock. Impacts on the Moon and Mars are affected by the differing gravity of each body. Similarly, impact speed plays a role when gravity is similar (for example, between Mars and Mercury) (e.g., Gault et al., 1975).

Impacts on airless bodies contrast those on planets with an atmosphere. In this case, the differences are not necessarily in the shock physics occurring in the target or the morphology of the crater itself, but in the effects of and on the environment. An

xvii atmosphere lends a great deal of complication to ejecta flow dynamics (Schultz and Gault,

1979; Schultz, 1992a; Barnouin-Jha and Schultz, 1996, 1998, Barnouin-Jha et al., 1999a,

1999b) and vapor expansion (Schultz, 1992b; Sugita and Schultz, 2002). And the processes that occur depend on the density and pressure of the atmosphere.

Mars provides an excellent laboratory for natural cratering experiments. It has the same gravity as Mercury, where an atmosphere is absent. Despite extensive erosion early in its history, the Martian landscape preserves impact craters dating back 3 Ga to 4 Ga, in contrast to only a few million years on Earth. Yet the presence of a thin atmosphere allows for the exploration of both impact processes and atmospheric effects without introducing extreme conditions, e.g., on Venus. In fact, the low atmospheric pressure of

Mars results in a unique process that is the focus of this dissertation.

3. Motivation

The objective of this study is to describe the formation of enigmatic streak-like features found around some, but not all, craters on Mars. These features are best observed through the Thermal Emission Imaging System (THEMIS) nighttime infrared images, where they appear as thermally bright, long, straight, and often double-tailed streaks extending from preexisting topography, such as small impact craters. The features often occur alongside a radially alternating bright and dark pattern in the thermal infrared. Both the streaks and the bright/dark pattern radiate from a single feature in a given area: a well-preserved impact crater.

The double tails could easily be mistaken as simple wind patterns, common across

Mars. Their association with the formation of an impact crater and the intense scouring leading to their unique thermal signature leads us to call them impact-wind streaks. The

xviii parent craters from which they originate are termed impact-wind streak craters. Impact- wind streak craters are rare: only 12 craters across the surface of Mars have the clear bright streaks. This paucity leads to questions about their formation and uniqueness.

The evolution of this study began in 1992 when Schultz predicted extreme surface effects from early-time impact processes (Schultz, 1992a) and the observation of the role of impact vaporization under low atmospheric pressures (Schultz, 1996). Wrobel et al.

(2006) later modeled the intense scouring found up to eight crater radii away from high- latitude craters and applied this model to the formation of pedestal craters. At that time, the authors indicated that a volatile-derived impact-vapor plume could have various alteration effects on the surface, including preconditioning, armoring, and scouring.

Although Jones et al. (2011) noted strange thermal features in the vicinity of Hale crater,

Schultz and Wrobel (2012) were the first to study these features in detail and described the intense surface effects by impact-vapor driven winds. Finally, (Schultz and Quintana,

2017) introduced impact-wind streaks (as they are defined here) and constrained possible modes of formation through a case study of Santa Fe crater.

4. Tools

A variety of tools aids in the study of impact cratering processes. Remote and direct observations of craters provide morphological, structural, chemical, and even environmental evidence of the impact process. But observations, whether remote or direct, are necessarily post-impact, and it is better for our health that we do not witness a large-scale cratering event in person. We are left simply with a remnant or scar of the event that may well lead to a great many more mysteries, rather than solutions – hence the basis for this thesis.

xix

Laboratory experiments are an invaluable tool because they provide a small scale representation of the large-scale processes occurring on the planets. Moreover, experiments allow for manipulation and testing of a wide parameter space. We can record the process with high-speed cameras so that we can play it back and study it, frame by frame. Ingenious researchers have cut targets in half (even particulate ones) to watch the impact in quarter-space and observe within the forming cavity and target material (e.g., Barnouin-Jha and Schultz, 1996; Schultz et al., 2005). Although laboratory experiments are limited and require careful consideration to design and interpret, they permit us to isolate processes and variables that then can focus and constrain models.

A final approach is modeling – creating a computational simulation of an impact event. Models allow for the manipulation and isolation of even more variables beyond the reach of experiments, such as gravity and impact speed. Furthermore, they span the spatial spectrum. Models can replicate small-scale impact experiments or large-scale planet-planet collisions and therefore allow for empirical connections between experimental data and real world observations. Within computational constraints, any number of models may be performed for a particular study, with varying realism.

However, it is important to note that a model is just that: a model and not reality. It is based on our best understanding of physical processes and the mathematics used to describe those processes. It can be over-manipulated. Therefore, as with experiments, care must be taken when designing a model and interpreting its results.

5. Dissertation Summary

Whenever possible, the combination of observation, experiments, and modeling is an excellent approach to impact cratering problems. Each aspect builds from the limitations of the others and allows for the exploration of a particular impact cratering process more fully. The study of impact-wind streaks is ripe for this kind of multi- pronged approach.

5.1 Chapter 1

Schultz and Wrobel (2012) presented a hypothesis for impact-wind streak formation based on observations of the crater Hale. Later, Schultz and Quintana (2017) further developed the hypothesis through observation and morphological analysis that led to the model of winds induced by impact vapor expanding into an atmosphere. But are these winds generated by the expanding vapor or the passage of the atmospheric shock wave? Chapter 1 tests these two hypotheses.

The NASA Ames Vertical Gun Range (AVGR) is a unique facility that uses a two-stage hydrogen light gas gun to launch projectiles in a small-scale recreation of impact events. While the name implies a fixed position, the AVGR is actually mounted on an A-frame that can be raised and lowered in order to fire projectiles at angles varying from 15°-90°. Such a capability keeps the gravity vector normal to the target surface that allows for the study of particulate or water targets, in addition to solid materials. Many other hypervelocity impact facilities utilize horizontal guns, which necessitate the use of solid target materials and head-on impacts (unless the target itself is tilted). The versatility of the AVGR makes it a truly remarkable facility.

Experiments performed at the AVGR introduced a variety of tracers and instruments in order to document three interrelated processes: (1) a surface roughening

xxi spreading outward from the impact point, (2) an expanding vapor plume, and (3) outward winds made visible by dusty pipe cleaners. For these experiments, a Pyrex projectile impacted into an easily vaporized powdered dolomite target at different angles under different atmospheric pressures (and densities). The clear connection between the surface roughening, vapor expansion, and outward winds demonstrated that the controlling process is an expanding vapor plume that interacts with the atmosphere to generate winds.

5.2 Chapter 2

Aside from the obvious difference in scale, the major limitations of the AVGR for this work include impactor speed and composition, as well as target materials. The

AVGR is limited to firing projectiles at speeds less than 7 km s-1 (Gault and Wedekind,

1978), a speed that will not generate significant vaporization of silicate targets. Because of this limitation, we used a powdered dolomite target in Chapter 1, which induced target vaporization even at low impact speeds (Schultz, 1996; Schultz and Eberhardy, 2015).

Nylon, a material that absorbs water and has a low vaporization point, was later used for tests of projectile vaporization in Chapter 4. While the results of the experiments performed with these materials are applicable to the impact winds hypothesis for Mars, the particular materials are unrealistic.

Therefore, computational models allow for the testing of planetary scale impacts with relevant materials and impact speeds. Chapter 2 outlines a study that uses the CTH shock physics family of codes in order to model hypervelocity impacts. CTH was developed at Sandia National Laboratories, and is suited for multimaterial, multidimensional shock physics applications (McGlaun et al., 1990; Hertel et al., 1993).

At their core, shock physics codes like CTH solve shock physics equations, called the

xxii

Rankine-Hugoniot equations for conservation of mass, momentum, and energy over a nearly discontinuous shockwave. In order to solve these equations completely, an equation of state is necessary. An equation of state describes the thermodynamic properties of a material, such as its pressure and internal energy, over many temperatures and densities. Additional constitutive models further refine the material properties to include factors such as strength, porosity, and damage for simulating more realistic materials.

Whereas Chapter 1 assessed whether impact vapor could generate impact winds,

Chapter 2 assesses the vapor formation process, i.e., how impact vapor is produced on

Mars. This chapter describes a suite of three-dimensional simulations (models) in CTH that explore the conditions necessary to generate high vaporization and intense surface winds upon impact. Models investigate the effects of target and impactor properties, such as composition, impact speed, and impact angle. High-speed impactors (> 12 km s-

1) and ice-rich materials (targets or impactors) are more likely create conditions necessary to form impact-wind streaks as observed on Mars. High-speed dunite (asteroid) impactors traveling at speeds above 20 km s-1 yield sufficient vapor but are unlikely at

Mars. While near-surface ice may occur on Mars, it would be restricted to high latitudes, in contrast to the observed distribution of impact-wind streak craters. The model results reveal that comet impactors (and perhaps some cases of thick near-surface ice) are most likely to produce impact-wind streaks.

5.3 Chapter 3

Chapter 3 turns to Mars in order to further focus on the source of the impact-wind streaks. A global survey located only 12 impact-wind streak craters over the planet,

xxiii between 30° North and South latitudes. Besides these 12 impact-wind streak craters, an additional 35 craters, termed radial thermal streak craters, have a similar radial alternating light and dark pattern in the THEMIS nighttime infrared images. In these cases, however, no clear, bright wind streak features are associated with any preexisting craters or other topographic obstacles.

An understanding of the global distribution of the two subsets of craters is necessary in order to better compare impact-wind streak and radial thermal streak craters.

A paucity of craters exists in Arabia Terra, Amazonis Planitia, the Tharsis Rise, Medusa

Fossae, and the high latitudes. These areas are all heavily mantled (Schultz and Lutz,

1988) and have high amounts of dust cover. As a result, they have been termed stealth regions due to their low radar reflectivity (Muhleman and Butler, 1991; Edgett, 2002;

Karunatillake et al., 2009). Because of the low thermal contrast provided by thick dust mantling, it is unsurprising that impact-wind streak craters are either not found in these areas or are expressed differently (e.g., as long-run out ejecta flows around high-latitude craters, (Wrobel et al., 2006). Conversely, areas with high thermal contrast tend to display more impact-wind streak craters. A series of case studies explored seven impact- wind streak craters and did not find any target characteristic that could account for the impact-wind streaks other than a supply of thermally bright material. The streaks may form regardless of terrain or geologic unit, which indicates that surface characteristics

(such as buried water-ice) cannot be the sole source of impact vaporization. Instead, ice

(comet) impactors may be the best explanation. This result reveals clues about the impact wind formation and extent, the cometary impact flux on Mars, as well as other processes, such as polar mantling due to obliquity variations.

xxiv

5.4 Chapter 4

Finally, Chapter 4 provides a summary of the previous work in the dissertation, presents recent supporting experiments from the AVGR, and applies the work to a case study of the craters Jijiga and Mojave. This chapter begins by summarizing the observational, experimental, and modeling work from Chapters 1-3. Because the conclusion of these previous chapters is that a cometary impact is a likely source for the impact vaporization and resulting vapor-winds on Mars, the next section delves into more detail about the cometary origin with additional experiments from the AVGR.

Laboratory experiments in this supporting study incorporated a different impactor and target setup than that described in Chapter 1. Instead of a Pyrex projectile striking a powdered dolomite target, a nylon impactor (simulating a comet) struck a particulate target made of either powdered dolomite or sand. A sand target suppressed the target- derived vapor such the impactor could be the only source of vapor. The resulting dramatic, self-luminescent vapor plume drove winds that were somewhat slower than those reported in Chapter 1, but they would still be considered severe tornadic winds at larger scales on the Earth. Additionally, the nylon impactor-derived winds enveloped the surface and resulted in longer surface interaction. If such a result for nylon projectiles is also true for cometary impacts on Mars, then comets may allow for longer-lasting winds at the surface and therefore more surface modification. The same result was found in computational models.

A final section of this chapter outlines the effects of impact winds in the regions of the Jijiga and Mojave craters where we use topography to isolate the controlling processes even further. This undertaking required the combined use of observations,

xxv laboratory experiments, and numerical modeling. The Jijiga impact occurred on a relict plateau (island) between two outflow channels and displays radial wind streaks crossing the channel floor over 500 m below the crater. The Mojave impact occurred on a channel floor, yet wind streaks cross not only a nearby plateau but also on a channel beyond the plateau. Wind streaks even occur on the floor of a large crater that intersects the plateau.

Laboratory results reveal that impact vapor expands both upward and downward into a depression, as required in these two cases. CTH models then mimicked the topography of Jijiga crater and further indicated that a volatile-rich impactor (or surface ice) could lead to the observed development of winds crossing landscapes with large topographic variations. Such terrains do not prevent impact-wind streak formation, but would preclude other models of wind streak formation, such as a basal flow of ejecta (e.g.,

Boyce and Mouginis-Mark, 2006; Boyce et al., 2015).

xxvi

References

Barnouin-Jha, O.S., Schultz, P.H., 1996. Ejecta entrainment by impact-generated ring vortices: Theory and experiments. Journal of Geophysical Research: Planets 101, 21099–21115. doi:10.1029/96JE01949.

Barnouin-Jha, O.S., Schultz, P.H., 1998. Lobateness of impact ejecta deposits from atmospheric interactions. Journal of Geophysical Research 103, 25–739. doi:10.1029/98JE02025.

Barnouin-Jha, O.S., Schultz, P.H., Lever, J.H., 1999a. Investigating the interactions between an atmosphere and an ejecta curtain: 1. Wind tunnel tests. Journal of Geophysical Research: Planets 104, 27105–27115. doi:10.1029/1999JE001026.

Barnouin-Jha, O.S., Schultz, P.H., Lever, J.H., 1999b. Investigating the interactions between an atmosphere and an ejecta curtain: 2. Numerical experiments. Journal of Geophysical Research: Planets 104, 27117–27131. doi:10.1029/1999JE001027.

Barringer Jr, D.M., 1928. A new meteor crater. Proceedings of the Academy of Natural Sciences of Philadelphia 307–311.

Boyce, J.M., Mouginis-Mark, P.J., 2006. Martian craters viewed by the Thermal Emission Imaging System instrument: Double-layered ejecta craters. Journal of Geophysical Research: Planets 111. doi:10.1029/2005JE002638.

Boyce, J.M., Wilson, L., Barlow, N.G., 2015. Origin of the outer layer of martian low- aspect ratio layered ejecta craters. Icarus 245, 263–272. doi:10.1016/j.icarus.2014.07.032.

Davis, W.M., 1926. Biographical Memoir Grove Karl Gilbert, 1843-1918. National Academy of Sciences Memoirs v. XXI, 303.

Edgett, K.S., 2002. Low-albedo surfaces and eolian sediment: Mars Orbiter Camera views of western Arabia Terra craters and wind streaks. Journal of Geophysical Research: Planets 107, 5–1–5–22. doi:10.1029/2001JE001587.

Gault, D.E., Guest, J.E., Murray, J.B., Dzurisin, D., Malin, M.C., 1975. Some comparisons of impact craters on Mercury and the Moon. Journal of Geophysical Research 80, 2444–2460. doi:10.1029/JB080i017p02444.

Gault, D.E., Quaide, W.L., Oberbeck, V.R., 1974. Impact cratering mechanics and structures, in: A Primer in Lunar Geology. Presented at the A Primer in Lunar Geology, pp. 177–189. xxvii

Gault, D.E., Wedekind, J.A., 1978. Experimental studies of oblique impact, in: Lunar and Planetary Science Conference Proceedings. pp. 3843–3875.

Gilbert, G.K., 1893. The Moon’s Face: A Study of the Origin of Its Features. Philosophical Society of Washington.

Green, J., 1965. Hookes and Spurrs in Selenology. Annals of the New York Academy of Sciences 123, 373–402. doi:10.1111/j.1749-6632.1965.tb20376.x.

Hertel, E.S., Jr., Bell, R.L., Elrick, M.G., Farnsworth, A.V., Kerley, G.I., Mcglaun, J.M., Petney, S.V., Silling, S.A., Taylor, P.A., Yarrington, L., 1993. CTH: A Software Family for Multi-Dimensional Shock Physics Analysis, in: Proceedings of the 19th International Symposium on Shock Waves. pp. 377–382.

Holsapple, K.A., 1987. The scaling of impact phenomena. International Journal of Impact Engineering 5, 343–355. doi:10.1016/0734-743X(87)90051-0.

Housen, K.R., Holsapple, K.A., 2003. Impact cratering on porous asteroids. Icarus 163, 102–119. doi:10.1016/S0019-1035(03)00024-1.

Ives, H.E., 1919. Some Large-Scale Experiments Imitating the Craters of the Moon. The Astrophysical Journal 50, 245. doi:10.1086/142503.

Jones, A.P., McEwen, A.S., Tornabene, L.L., Baker, V.R., Melosh, H.J., Berman, D.C., 2011. A geomorphic analysis of Hale crater, Mars: The effects of impact into ice- rich crust. Icarus 211, 259–272. doi:10.1016/j.icarus.2010.10.014.

Karunatillake, S., Wray, J.J., Squyres, S.W., Taylor, G.J., Gasnault, O., McLennan, S.M., Boynton, W., El Maarry, M.R., Dohm, J.M., 2009. Chemically striking regions on Mars and Stealth revisited. Journal of Geophysical Research: Planets 114, E12001. doi:10.1029/2008JE003303.

Kopal, Z., 2012. The Moon. Springer Science & Business Media.

McGlaun, J.M., Thompson, S.L., Elrick, M.G., 1990. CTH: A three-dimensional shock wave physics code. International Journal of Impact Engineering 10, 351–360. doi:10.1016/0734-743X(90)90071-3.

Muhleman, D.O., Butler, B.J., 1991. Radar images of Mars. Science 253, 1508. doi:10.1126/science.253.5027.1508.

xxviii

O’Keefe, J.D., Ahrens, T.J., 1977. Impact-induced energy partitioning, melting, and vaporization on terrestrial planets. in: Lunar and Planetary Science Conference Proceedings, pp. 3357–3374.

Schultz, P.H., 1996. Effect of impact angle on vaporization. Journal of Geophysical Research: Planets 101, 21117–21136. doi:10.1029/96JE02266.

Schultz, P.H., 1998. Shooting the Moon: Understanding the History of Lunar Impact Theories. Earth Sciences History 17, 92–110.

Schultz, P.H., 1992a. Atmospheric effects on ejecta emplacement. Journal of Geophysical Research: Planets 97, 11623–11662. doi:10.1029/92JE00613.

Schultz, P.H., 1992b. Atmospheric effects on ejecta emplacement and crater formation on Venus from Magellan. Journal of Geophysical Research: Planets 97, 16183– 16248. doi:10.1029/92JE01508.

Schultz, P.H., Eberhardy, C.A., 2015. Spectral probing of impact-generated vapor in laboratory experiments. Icarus 248, 448–462. doi:10.1016/j.icarus.2014.10.041.

Schultz, P.H., Ernst, C.M., Anderson, J.L., 2005. Expectations for crater size and photometric evolution from the Deep Impact collision, in: Deep Impact Mission: Looking Beneath the Surface of a Cometary Nucleus. Springer, pp. 207–239.

Schultz, P.H., Gault, D.E., 1979. Atmospheric effects on Martian ejecta emplacement. Journal of Geophysical Research: Solid Earth 84, 7669–7687. doi:10.1029/JB084iB13p07669.

Schultz, P.H., Lutz, A.B., 1988. Polar wandering of Mars. Icarus 73, 91–141. doi:10.1016/0019-1035(88)90087-5.

Schultz, P.H., Quintana, S.N., 2017. Impact-generated winds on Mars. Icarus 292, 86– 101. doi:10.1016/j.icarus.2017.03.029.

Schultz, P.H., Wrobel, K.E., 2012. The oblique impact Hale and its consequences on Mars. Journal of Geophysical Research 117, E04001. doi:10.1029/2011JE003843.

Schultz, P.H., Zarate, M., Hames, W., Camilión, C., King, J., 1998. A 3.3-Ma Impact in Argentina and Possible Consequences. Science 282, 2061–2063. doi:10.1126/science.282.5396.2061.

Sugita, S., Schultz, P.H., 2002. Initiation of Run-Out Flows on Venus by Oblique Impacts. Icarus 155, 265–284. doi:10.1006/icar.2001.6731. xxix

Wrobel, K., Schultz, P., Crawford, D., 2006. An atmospheric blast/thermal model for the formation of high-latitude pedestal craters. Meteoritics & Planetary Science 41, 1539–1550. doi:10.1111/j.1945-5100.2006.tb00434.x.

xxx

TABLE OF CONTENTS

Title Page ...... i Copyright Page...... ii Signature Page...... iii Curriculum Vitae ...... iv Acknowledgements ...... xi Preface ...... xvi Table of Contents ...... xxxi

CHAPTER ONE: Experimental Constraints on Impact-Induced Winds ...... 1 Abstract ...... 2 1. Introduction ...... 3 2. Wind Streaks on Mars ...... 4 3. Impact-Wind Streaks ...... 4 4. Experimental Setup and Methods ...... 6 4.1 The AVGR ...... 6 4.2 Detection Devices ...... 8 5. Results ...... 9 5.1 Surface Roughening ...... 10 5.2 Winds ...... 13 5.3 Wake Effects ...... 15 5.4 Summary ...... 15 6. Discussion ...... 17 6.1 Experimental Limitations ...... 17 6.2 A Comparison with Terrestrial Explosions ...... 18 7. Conclusion ...... 20 Acknowledgements ...... 22 References ...... 23 Tables ...... 29 Figure Captions ...... 35 xxxi

Figures...... 41

CHAPTER TWO: Numerical Models of Impact-Induced Winds on Mars ...... 53 Abstract ...... 54 1. Introduction ...... 55 2. Computational Models ...... 56 2.1 Vapor Determination in CTH ...... 59 3. Methods ...... 60 3.1 Planetary-Scale Simulations ...... 60 3.2 Data Acquisition and Resolution...... 63 4. Results ...... 64 4.1 Vaporization ...... 65 4.2 Wind Speeds ...... 66 4.3 Effects of Topographic Obstacle ...... 69 4.4 Effects of a Distant Obstacle ...... 70 4.5 Shockwave ...... 70 4.6 Summary ...... 71 5. Discussion ...... 71 5.1 Refining the Analysis ...... 71 5.2 Model Limitations ...... 73 6. Conclusion ...... 74 Acknowledgements ...... 76 References ...... 77 Tables ...... 83 Figure Captions ...... 97 Figures...... 99

CHAPTER THREE: Distribution, Observations, and Implications of Impact-Wind Streaks on Mars ...... 104 Abstract ...... 105 1. Introduction ...... 106 xxxii

2. Background ...... 107 2.1 Thermal Emission Imaging System (THEMIS) ...... 107 2.2 Impact-Wind Streaks ...... 109 3. Global Distribution of Impact-Wind Streaks...... 110 4. Case Studies ...... 113 4.1 Xainza Crater ...... 113 4.2 Pál Crater ...... 114 4.3 Mojave Crater ...... 116 4.4 Kotka Crater ...... 118 4.5 Two Unnamed Craters in or near Isidis Planitia ...... 119 4.6 Prao Crater ...... 120 5. Discussion: Comparing Radial Thermal Streak Craters and Impact-Wind Streak Craters ...... 122 5.1 Radial Thermal Streak Crater Characteristics ...... 122 5.2 Crater Ages ...... 124 5.3 Impact-Generated Winds ...... 125 6. Wind Streak Origin ...... 126 7. Discussion: Applications ...... 129 7.1 Constraints on Impactor Size ...... 129 7.2 Constraints on Geological Processes ...... 130 8. Conclusion ...... 132 Acknowledgements ...... 133 References ...... 134 Tables ...... 142 Figure Captions ...... 146 Figures...... 152

CHAPTER FOUR: Cometary Impacts and Impact-Wind Streaks on Mars ...... 168 Abstract ...... 169 1. Introduction ...... 170 2. Background ...... 171 xxxiii

2.1 Vapor-Driven Winds ...... 171 2.2 Impact Vaporization ...... 172 2.3 Impact-Wind Streak Craters ...... 173 3. Cometary Origin of Impact-Generated Winds ...... 174 3.1 Methods ...... 174 3.2 Results ...... 175 4. Topographic Case Studies ...... 179 4.1 Methods ...... 181 4.2 Results ...... 182 5. Discussion ...... 184 5.1 Comet Impacts ...... 184 5.2 Topographic Effects ...... 185 6. Conclusion ...... 186 Acknowledgements ...... 188 References ...... 190 Tables ...... 193 Figure Captions ...... 203 Figures...... 207

xxxiv

CHAPTER ONE:

Experimental Constraints on Impact-Induced Winds

Stephanie N. Quintana1

and

Peter H. Schultz1, Seth S. Horowitz2

1Department of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI 02912

2The Engine Institute, Inc. Warwick, RI 02888

Abstract

A new class of wind streaks on Mars uniquely associated with impact craters are most clearly visualized in nighttime thermal infrared imaging. These thermally bright streaks radiate from some well-preserved impact craters and are related to the impact process.

Through laboratory experiments performed at the NASA Ames Vertical Gun Range, we test the hypothesis that these streaks are formed from either the winds within an air-blast or winds set up by expanding impact vapor interacting with the atmosphere. The experiments utilize a variety of tracers and instruments to document three interrelated processes occurring in the impact of a Pyrex projectile into an easily vaporized powdered dolomite target: (1) a surface roughening spreading outward from the impact point, (2) an expanding vapor plume, and (3) outward winds made visible by dusty pipe cleaners. The clear connection between the surface roughening, vapor expansion, and outward winds implicate an expanding vapor interacting with the atmosphere as the controlling process.

1. Introduction

Wind streaks on Mars were identified first from the Mariner 9 mission. Sagan et al. (1972) termed these albedo patterns ‘variable features,’ which speaks to their transience on the surface of Mars. Greeley et al. (1974, 1978), among others, modeled the process of wind streak formation through wind tunnel experiments in order to determine threshold wind speeds for the mobilization of particles on Mars. They also tested how to distinguish erosional and depositional features revealed in Mariner 9 images. Later work by Thomas et al. (1981) classified these variable features into wind- streak categories based upon their albedo contrast and location with respect to particular topographic obstacles. From these and later studies (e.g., Greeley and Iversen, 1985;

Pelkey et al., 2001), it became clear that these wind streaks were related to global wind patterns and likely global dust storms, which could change the length and orientation of the streaks over time.

The Mars Odyssey’s Thermal Emission Imaging System (THEMIS) revealed a new category of wind streaks (Schultz and Wrobel, 2012; Schultz and Quintana, 2013;

Quintana and Schultz, 2014; Quintana et al., 2015, 2016; Quintana and Schultz, 2016).

These streaks are visible most clearly in nighttime infrared images and extend from topographic highs, such as crater rims. More importantly, they radiate from an impact crater; hence, they are related to the crater-forming process.

This contribution investigates the process of formation for this new category of wind streaks through a series of laboratory experiments conducted at the NASA Ames

Vertical Gun Range in Mountain View, CA. Two potential formation processes include:

1) an atmospheric air-blast similar to an explosion-generated air-blast on Earth and 2) outward winds created by impact-generated vapor coupled to the atmosphere.

2. Wind Streaks on Mars

Thomas et al. (1981) classified wind streaks on Mars based on their appearance with respect to the surrounding albedo and source regions. In general, the authors identified three basic categories in which most of the wind streaks on Mars fall: (a) bright streaks without a clear source deposit, (b) dark streaks without a source deposit, and (c) dark streaks with a source. Most of the wind streaks in these categories are related to topographic relief such as craters. Other categories are unrelated to our current study, including splotches and related streaks, dune shadow streaks, and frost streaks (e.g. Sagan et al., 1972; Arvidson, 1974; Veverka et al., 1974; Greeley et al., 1974; Thomas et al.,

1981; Veverka et al., 1981; Greeley and Iversen, 1985). Despite the connection between streaks and craters in the Thomas et al. (1981) study, those streaks are not related to the cratering process. Thus, we term this new classification of wind streaks impact crater- generated wind streaks, which we abbreviate here as impact-wind streaks. While impact- wind streaks are most easily found in nighttime thermal images, they are distinct from the more typical thermal signatures of wind features classified by Pelkey et al. (2001).

3. Impact-Wind Streaks

Impact-wind streaks are bright features in nighttime infrared images that extend from topographic highs (e.g., preexisting crater rims or high ridges). Schultz and

Quintana (2017) abbreviated the thermally bright features seen in nighttime infrared NT-

B features, whereas thermally dark streaks were labeled NT-D features. NT-B features often occur as single or nearly parallel double wind streak tails (Figure 1). Wind-streak

4 tails form in laboratory wind tunnel experiments (Iversen et al., 1973; Greeley et al.,

1974; Iversen et al., 1976; Greeley et al., 1978; Greeley and Iversen, 1985), although the tails curve together to meet at a point not far from the obstacle in these low-wind speed cases. The nearly parallel nature of the thermally streaks on Mars indicates that they were formed by very strong, sustained winds. The higher thermal inertia of the streaks further implies they are composed of materials with larger particle sizes compared to their surroundings, such as sand or gravel, cemented grains, exposed substrate or a mixture of these types of materials (Woodside and Messmer, 1961; Wechsler and Glaser,

1965; Fountain and West, 1970; Kieffer et al., 1973; Christensen, 1982; Presley and

Christensen, 1997a, 1997b, 1997c, Fergason et al., 2006a, 2006b).

In a case study of the 20-km diameter Santa Fe crater in Chryse Planitia, Schultz and Quintana (2017) found that the NT-B impact-wind streaks correlate with irregular higher relief ridges and low-relief yardangs with exposed blocky materials radiating from

Santa Fe. In contrast, smoother surfaces or pitted surfaces filled with dunes are characteristic of the NT-D zones radiating from Santa Fe. These dark zones are more diffuse than the distinct NT-B streaks, but they are often found in close proximity to the streaks in a radial pattern.

It is important to note that the NT-B impact-wind wind streaks defined here are distinct from bright radial ejecta rays, some of which look similar in the THEMIS nighttime infrared (Gregg, 2015). The presence of bright ejecta deposits in an area is indicative of high relative thermal contrast, which may be an important factor in impact- wind streak development. However, the NT-B impact-wind streaks appear outside of the continuous ejecta deposits (Schultz and Quintana, 2017) and are not related to the bright

5 ejecta deposits or secondary crater chains. Conversely, the dark ‘blast zones’ identified around small, fresh craters on Mars (e.g. Malin et al., 2006; Ivanov et al., 2008, 2009,

2010; Daubar et al., 2013; Dundas et al., 2014; Daubar et al., 2016) may be smaller scale expressions of a similar process.

This contribution uses impact experiments to distinguish between potential formation processes for the impact-wind streaks. Schultz and Quintana (2017) list two atmospheric processes that could generate winds consistent with the NT-B streaks: (a) atmospheric blast winds, which are initiated by the contact-coupled atmospheric shockwave, and (b) winds driven by impact-generated vapor expanding into the atmosphere. Therefore, we address both of these possibilities in the following laboratory experiments.

4. Experimental Setup and Methods

4.1 The AVGR

Laboratory experiments were conducted at the NASA Ames Vertical Gun Range

(AVGR) in Mountain View, CA. This unique facility uses a two-stage hydrogen light gas gun to fire projectiles with a variety of sizes, compositions, and velocities up to nearly 7 km s-1. The gun is mounted on an A-frame that allows the barrel to be positioned at angles between 15° and 90° with respect to the horizontal target, which is housed in a large (2.5-m diameter, 3 m high) target chamber. This chamber can accommodate many target setups and atmospheric or near-vacuum environments. Figure

2 is a diagram of the AVGR and a simplified camera setup used in this study. Because the impact angle can vary, the AVGR facility can be used to study the effects of oblique impacts in particulate and liquid targets (Gault and Wedekind, 1978).

Laboratory experiments are not necessarily meant to be a direct simulation or comparison for planetary-scale impacts; rather, they are most efficiently used to isolate and study particular processes in more detail. For this study, a target material that enhances impact-vapor at the relatively low impact speeds attainable at the AVGR was necessary in order to test the hypothesis that intense impact-related winds can be initiated from impact-vapor expansion into an atmosphere, in contrast to blast winds generated at first contact. Ivanov and Deutsch (2002) describe how carbonates like dolomite will decompose from the residual temperatures after pressure release, rather than the pressures attained during the impact itself. Indeed, some studies show that dolomites are stable under shock pressures as high as 60 GPa (Agrinier et al., 2001; Ivanov and Deutsch,

2002), but decomposition may occur at temperatures around 975 K (McCauley and

Johnson, 1991; Kök and Smykatz-Kloss, 2001). Dolomite decomposes through a single- stage process in inert atmospheres (Kök and Smykatz-Kloss, 2001). Previous studies

(e.g. Schultz and Gault, 1990; Schultz, 1996; Sugita et al., 1998; Bruck Syal and Schultz,

2014; Schultz and Eberhardy, 2015; Quintana et al., 2015) make use of this property and document the generation of vapor from powdered dolomite at laboratory impact velocities where measured temperatures in the vapor plume initially exceed 5000 K

(Sugita et al., 1998; Bruck Syal and Schultz, 2014). Therefore, compacted, powdered dolomite provided an ideal target for assessing the effects of vaporization (Schultz, 1996;

Sugita et al., 1998; Bruck Syal and Schultz, 2014).

The large AVGR target chamber also allows the testing of different atmospheric conditions. Most experiments presented here used a pressure of approximately 33 mbar

(25 Torr). Different gases (air, argon, and helium) held at the same pressure allow for

7 exploration of the effects of atmospheric coupling during vapor expansion by changing atmospheric density.

4.2 Detection Devices

High-speed cameras record the impact process in great detail through windows in the impact chamber at the AVGR. This study used up to eight cameras with frame rates from 1900 to 125,000 frames per second in both color and black and white (Phantom® high speed cameras and Shimadzu® imaging cameras). At least two color Phantom cameras recorded processes normal to the impact trajectory, while additional color and black and white cameras at different positions focused on different areas of the target and its surroundings, including at least one view from above. The ultra-high-speed Shimadzu cameras recorded the earliest time processes, including the downrange vapor expansion and high-speed ejecta evolution. Each camera provided a unique perspective of the impact process and, when combined, provided insights into the expanding vapor plume and wind generation process.

In addition to the high-speed cameras, a series of instruments directly measured passing shockwaves and changes in pressure, vapor expansion, and wind development.

Two types of PCB Piezotronic pressure sensors recorded pressure after impact: up to four free-field ICP® blast pressure pencil probes and up to five high-frequency ICP® pressure sensors. Customized aerodynamic sensor holders minimized turbulent interactions with the sensors and optimized data collection under variable environmental conditions. The sensor holders allowed stacking the sensors up to three high or placing them side-by-side.

A schematic of the ideal sensor setup is illustrated in Figure 3. In conjunction with the pressure sensors, contact microphones and pressure zone microphones recorded

8 independent time measurements for atmospheric phenomena. Finally, three geophones attached to the side of the target container with their centers 4.5 cm below the target surface recorded the timing of seismic waves within the target.

Different aspects of the impact process can be visualized through the use of tracers, summarized in Table 1. Millimeter-sized Styrofoam balls sprinkled on the surface of the target (Experimental Setups 1 and 2) provided tracers for the effects of expanding vapor and atmospheric wind. Other experiments used a flat calibration plate

(decoupled from the target container) that extended the target surface and tracer placement well beyond the target container (Experimental Setups 3 and 4). Vertical pipe cleaners dusted with fine (< 20 μm) powdered dolomite also provided visual indicators of both the arrival time and the speed of passing winds, e.g., vapor-generated wind development (Experimental Setup 5). Finally, some conditions better captured (visually,

Experimental Setup 6) or suppressed (physically, Experimental Setup 7) vapor development. Pressure sensors, microphones, and geophones included in some experiments from Experimental Setup s 3-7 provided other methods to capture impact processes.

A series of experimental runs allowed for a systematic exploration of vapor and wind generation. Each run (Table 2) used a 0.635 cm-diameter Pyrex projectile impacting the powdered dolomite target at angles between 30° and 90° from the horizontal. The target of powdered dolomite filled a 59 cm-diameter container placed inside a well within the impact chamber (Figure 4) provides a schematic of the target- chamber setup).

5. Results

Each experimental technique (in addition to the use of instruments and tracers) aided in the separation of different atmospheric responses. An observed surface roughening provided clues into impact-vapor expansion behavior. Mobilization of dolomite powder dusted on pipe cleaners indicated onset and speeds of winds. These two processes allowed for the simultaneous testing of both impact-wind streak generation hypotheses: (a) blast winds associated with an atmospheric shockwave and (b) winds initiated by impact-vapor expansion into an atmosphere.

5.1 Surface Roughening

Directly after impact, a surface darkening began to spread out asymmetrically from the impact point: expanding faster downrange than uprange at a given time. With target chamber lights on, high-speed cameras were positioned approximately 40 cm above the target surface such that light scattered off of the surface of the target and revealed the expanding darkened zone (Figure 5). Positioning of the lights in the chamber ensured that this surface darkening was not due to a shadow (e.g., caused by the growing ejecta curtain) but was instead a physical change to the uppermost structure of the target, i.e., a decrease in forward reflection as the result of surface roughening. The zone of surface roughening expanded at supersonic speeds, with the downrange disturbance moving faster than the uprange disturbance for all angles except 90°.

Typically, 90° angles yielded an initial surface disturbance traveling 520 m s-1, slowing to

480 m s-1 by the time the disturbance reached the edge of the target surface.

Comparatively, for shots at 45° in air, initial downrange surface disturbance speeds ranged from 1400-1990 m s-1 and uprange speeds ranged from 590-1040 m s-1.

Analysis of the high-speed images revealed that the expansion speed of roughening

10 decayed with distance and increasing impact angle (Figure 6). By the time that the disturbance reached the edge of the target surface, it had slowed to between 870-1370 m s-1 downrange and 590-770 m s-1 for the 45 ° case. Figure 7 provides representative downrange and uprange surface roughening speeds for impacts at 45° under differing ambient atmospheres (at 33 mbar pressure). Surface roughening speed was nearly equivalent in air and argon, but was much faster in helium due to the lower density of the gas. The time it took for the surface roughening to reach from the point of impact to the edge of the target container determined an ‘overall speed,’ which provided a comparison with measurements from the detectors placed around the target.

Several strategies investigated the cause of the surface roughening: an expanding shockwave (either in the target or in the atmosphere) or vapor-atmosphere interactions.

Two separate tests conclusively ruled out a shockwave in the target, termed the ground- coupled shock. First, geophones placed on the outside of the target container along two axes (downrange-uprange and lateral) recorded the initial impulse of the ground-coupled shockwave (seismic wave), as well as the later ringing of the shockwave within the target.

The first signal occurred well after the surface roughening had passed the edge of the container, as shown in Figure 8. Consequently, the ground-coupled shockwave could not be responsible for the observed roughening. Second, two near-vacuum experiments

(approximately 0.4 Torr of air for both a 30° and a 45° impact) provided further evidence that the surface roughening was not a ground-coupled effect. No surface roughening was apparent for these experiments, thereby demonstrating that the roughening was instead an atmospheric phenomenon.

The high-speed cameras provide telling evidence for vapor-atmosphere interactions. The high-speed camera frames (Figure 9) reveal that the surface roughening advances with the leading edge of the vapor plume. In this image, the top three quarters of the image has been stretched in order to show the vapor plume. The actual extension of the leading edge of the plume is invisible near the surface due to poor contrast with the background. Experiments with the chamber lights off demonstrated that the vapor plume did, in fact, reach the surface.

In the second test, pressure sensors under a low-density atmosphere of helium

(rather than air), revealed clear separation between the atmospheric shockwave (from first contact of the projectile and target) and the expanding vapor plume. Free-field blast pressure pencil probes were placed in two locations in order to record the passage of this shockwave: one approximately 50 cm downrange from the impact point at an offset of

~40° (in order to protect the sensors); the other 50 cm directly uprange of the impact point. This positioning produced the cleanest initial signal and avoided reflections and reverberations of blast and shockwaves off of the chamber walls.

Because the plume boundary fades and becomes difficult to measure, the correlated surface roughening provides a possible proxy. At large distances from the impact point, a vapor plume eventually decelerates exponentially due to atmospheric drag.

An exponential fit to the observed speeds at earlier times allows for an estimation of the speed farther out. In a helium atmosphere, the exponential fit for earlier data is

푣 = 606푒−0.044푑, (1) where v is the estimated vapor plume speed and d is the distance from the impact point

(Figure 10). We used this equation (equation 1) to estimate that the vapor plume was

12 traveling at a speed of 3400 m s-1 when it reached the distance of the pressure sensors.

The pressure sensors, on the other hand, detected the shockwave traveling at ~1370 m s-1

(near the speed of sound in helium). Uprange, however, the shockwave outpaced the vapor plume. The overall speed of the shockwave was ~1200 m s-1, while the vapor plume traveled at 450 m s-1 at the same distance. The shockwave clearly separated from the vapor plume (the source for the surface roughening). Therefore, the surface roughening could not have been caused by an atmospheric shockwave generated from the first contact; instead, it must have been caused by the expanding vapor plume, which evolved slightly later.

5.2 Winds

Analysis of high-speed images yields the most reliable data about wind speeds.

Pipe cleaners dusted with dolomite placed within and outside of the target container served as simple but effective markers for wind initiation, development, and speed.

Initial experiments used pipe cleaners placed approximately 5 cm inside the target edge

(toward the center), while other experiments used a line of pipe cleaners placed in 5 cm increments (two pipe cleaners within the target container and up to four pipe cleaners outside the bucket). Figure 11a and b is an example of a high-speed frame that captured the mobilized pipe cleaner dust (also in Figure 9). Within a single high-speed frame after the passage of the vapor plume (within 35 μs), dust began detaching from the pipe cleaners, directed away from the impact point. This mobilization is interpreted as wind entrainment of the fine dust. Onset of the observed winds depended on impact angle, beginning downrange at earlier times for more oblique impact angles.

Frame-by-frame analysis of the pipe cleaner dust trails provides an estimate of wind speeds (summarized in Table 3). Wind speeds from oblique impacts exceeded those generated by higher angle impacts due to the added component of downrange motion to the expanding vapor. Pipe cleaners placed uprange and lateral to the trajectory for higher impact angles not only exhibited lower speeds but less entrainment. For example, dust trails for some instances of 60° and 90° impact angles were faint or even imperceptible. In such cases, the vapor plume was observed to expand nearly hemispherically above and around the impact point, which resulted in winds passing at a higher level on the pipe cleaners. This evolution is attributed to temporary containment and upward direction of expanding vapor (e.g., Schultz, 1996; Schultz and Eberhardy,

2015). Outside of the target container, the vapor plume from high-angle impact angles expanded above the maximum height of pipe cleaners and resulted in an apparent absence of winds.

Wind reversal (reveled by inward-directed dust) occurred at a later time (e.g. approximately 350 μs downrange and 400 μs uprange). Figure 11c and d captures this process about 420 μs later than the frame shown in Figure 11a and b. Reverse winds were generally much slower than the preceding outward-directed winds (demonstrated in

Table 3) and showed no strong asymmetry downrange or uprange of the impact point. As with the initial outward-directed winds, lower impact angles generated faster reverse winds. Reverse winds also exhibited much less variation in speeds, relative to outward- directed winds.

As the vapor expanded outwards, the trailing winds interacted with small, millimeter-sized Styrofoam balls sprinkled on and around the target. The balls recorded

14 boundary layer effects and the development of turbulence within the impact chamber.

Experiments testing the response of the Styrofoam balls revealed that they mobilized immediately (within 70 μs) after the passage of the surface roughening (vapor plume).

The balls lifted from the target as they moved away from the impact point, and many eventually curved back toward the impact point. The curved trace of these Styrofoam balls resembled the pattern of wind reversal captured by the dusty pipe cleaners.

5.3 Wake Effects

The wake, caused by the passage of the projectile through the atmosphere before impact, had little to no effect on the uprange surface roughening, vapor, or winds except at lower impact angles. At 30°, a later surface disturbance uprange was detected in high- speed images positioned above the target (looking down on the impact). The uprange surface disturbance began with lineations or grooves in the target material. Later,

Styrofoam ball tracers in the area were pushed aside perpendicularly to the projectile trajectory. These occurrences are attributed to the wake, as it forms a cylindrical shock front that expands outward around the projectile trajectory.

This wake-related surface disturbance began well after the surface roughening reached the edge of the target container. For example, in an impact at 5.6 km s-1 in 33 mbar of air (Run 160908), the surface roughening reached the edge of the target container at about 570 μs after impact. The uprange surface disturbance did not begin until 1140 μs.

Therefore, wake effects occurred late enough that the vapor plume (surface roughening) and wind calculations were not affected, even for low angle impacts.

5.4 Summary

In summary, laboratory experiments allowed for the separation of three related processes in response to a hypervelocity impact into an easily vaporized target material:

(1) a surface roughening spreading outward from the impact point, (2) an expanding vapor plume, and (3) winds blowing debris from dusty pipe cleaners aligned radial to the impact point. The surface roughening was an atmospheric process related to the expanding vapor plume. Various tests demonstrated that the surface roughening was neither a ground-coupled shockwave nor a contact-coupled atmospheric shockwave.

Rather, it was an atmospheric process caused by an expanding vapor plume. Pipe cleaners dusted with dolomite powder and millimeter-sized Styrofoam balls traced the effects of impact-generated winds. Because the initiation of the winds occurred directly after the passage of the vapor plume, we conclude that impact-generated vapor moving through the ambient atmosphere initiated fast-moving winds. Even under the low atmospheric pressures used in the impact chamber, the winds were strong enough to entrain and mobilize fine dust attached to the pipe cleaners. Furthermore, the winds reached and interacted with Styrofoam ball tracers sprinkled on and around the target.

Figure 12 diagrams the sequence of events during an impact that leads to impact wind development, which was observed in the laboratory experiments described above.

The projectile creates a wake shockwave as it travels through the atmosphere. Upon impact, a shockwave expands through the air above the target and in the ground. Due to the porosity of dolomite, the ground-coupled shockwave within the target is slower than the air-coupled shockwave. Jetting occurs at first contact of the projectile and target, but it is not shown in the diagram because the amount of vapor in the early-time jet is very low compared to the other vapor plumes in the diagram. An additional initial, hot and

16 fast-moving vapor plume expands downrange and upward. Later, a growing vapor plume that had been partly contained within the growing crater cavity expands roughly hemispherically around the impact point. The plume overpowers the uprange wake shock (from the projectile passage through the atmosphere) for all but the shallowest impact angles (<30°), and may overtake the air-coupled shockwave in low-density atmospheres like helium. The passage of this vapor plume causes the surface roughening and initiates impact winds.

6. Discussion

6.1 Experimental Limitations

Comparing vapor plume speeds and pressure sensor signal arrival times

(shockwave arrival times) necessitated the use of an overall speed (averaged from the time of impact to a given position) in order to compare the two measurements.

Consequently, comparison of the vapor-plume and surface-roughening arrival times near

(and beyond) the edge of the target container at the distance of the pressure sensors (50 cm) required an alternative approach. Vapor plume (surface roughening) speed at early times was measured directly from frame-by-frame analysis; this speed was recorded for every frame in which the plume or surface roughening was visible. An exponential fit to this early-time data then yielded the vapor-plume speed at greater distances.

Similarly, the pressure sensors only recorded the arrival time of the shockwave at a particular point (50 cm from the impact point). The shockwave speed calculated, therefore, is the overall speed from the point of impact to the pressure sensor. Although shockwave speed also decays with time, the pressure sensors were not arranged in such a

17 way to record this decay. This deficiency, however, is not critical for this study, because the comparison of overall shockwave speed and vapor plume speed sufficed.

Errors in the presented data arise mostly from observational uncertainty. Vapor plume, surface roughening, and dust stream speeds were determined through high-speed camera frame analysis. For surface roughening measurements, four measurements of the surface roughening edge allow for determining a representative measurement error for impacts under air, argon, and helium (see Figure 7). The impact point and subsequent timing have an error up to the frame rate interval (52 μs for runs beginning 1409XX and

35 μs for runs beginning 1609XX). Furthermore, the leading edges of each of these observations were often diffuse. For wind speed analysis, a double image of the vertical pipe cleaners resulted from multiple reflections in the window and safety plastic and may have interfered with speed measurements. Edge-determination errors in the analyses are presented in Table 4 for wind speed measurements uprange, downrange, and lateral to the impact trajectory.

6.2 A Comparison with Terrestrial Explosions

The effects of nuclear explosions—and to an extent in high explosive chemical detonations—have long been used to understand the cratering process (e.g. Ahrens and

O’Keefe, 1977; Schultz and Gault, 1979; Schmidt and Housen, 1987; Holsapple, 1987).

However, such explosions initiate a nearly instantaneous release of the heated gas (from radiation or chemical reactions) that generates high pressures and a strong atmospheric shockwave. The resulting supersonic blast travels outward across the surface and decays with distance. Due to the momentum of the air behind the shockwave, outward-directed winds continue even under the negative overpressure phase (the time at which the

18 pressure behind the shockwave is lower than the ambient atmospheric pressure).

Immediately afterwards, however, the winds decrease enough that the negative pressure behind the shockwave draws the winds in reverse. Later, the rising fireball (from the heated atmosphere) draws winds inward (Glasstone and Dolan, 1977).

On Mars, the lower atmospheric pressure relative to Earth allows separation of different components of early-time shock effects. The expanding impact vapor overwhelms the initial shockwave and engulfs the surrounding atmosphere. In this case, the expanding gas should generate and control impact winds. The downrange motion of the plume and amount of expanding vapor distinguish this process from the coupled atmospheric shock generated by expanding atmospheric hot gas from a nuclear or chemical explosion.

Similarly, the low atmospheric pressure used in the experiments here also separates the shockwave from winds effects. As documented in previous experiments

(e.g., Schultz, 1996; Schultz and Eberhardy, 2015), vapor is also briefly contained within the growing impact cavity (as opposed to the instantaneous coupling of the shockwave and vapor in a point-source explosion), which further delays impact winds from an initial shock generated at first contact. The winds in the experiments, therefore, must be due to atmospheric displacement from impact vapor expansion.

In terrestrial explosion craters, wind reversal occurs following the passage of an air blast and the later rising heated atmosphere. Wind reversal in the experiments, however, is a two-part process due to the lower atmospheric pressures used and the scale of the experiments. Immediately after an oblique impact, hot vapor travels rapidly downrange due to the initial impactor momentum and directed shock (red plume in

Figure 12), as observed in experiments (e.g., Schultz, 1992a, 1996), models (O’Keefe and

Ahrens, 1986), and on the planets (Schultz, 1992a, 1992b; Schultz and Gault, 1990).

This downrange-moving plume also creates a partial vacuum that draws in the surrounding air. A second vapor plume (grey plume in Figure 12) expands hemispherically around the crater but is centered downrange. Passage of this plume correlates with the observed outward-directed winds. Because of this plume, later winds are directed inward (just downrange of the forming crater) near the target surface and upward as time progresses. At still later times, dust from the pipe cleaners near the ejecta curtain becomes entrained in vortices generated by the upward-moving stream of ejecta

(Schultz and Gault, 1979, 1982; Schultz, 1992a; Barnouin-Jha and Schultz, 1996, 1998,

Barnouin-Jha et al., 1999a, 1999b).

7. Conclusion

Impact experiments provide insights for the model of impact-wind streak formation by separating various impact processes. Sensors and cameras within and outside the impact chamber of the NASA AVGR reveal that the impact winds are not associated with an atmospheric shockwave generated at first contact between the projectile and target. Instead these winds result from expanding vapor, a separate process that is uniquely visible around impacts on Mars due to its tenuous atmosphere. Filling the chamber with different gases allowed for assessing the effects of atmospheric density on vapor expansion. Through these observations, we recognized a series of separable processes:

1) A surface roughening spreads asymmetrically away from the impact point at

supersonic speeds for both the atmosphere and the target. Image analysis

demonstrated that the surface roughening was directly related to the expanding

vapor plume.

2) A ground-coupled shockwave (detected by geophones placed on the target

container) travels slower than the surface roughening. Experiments under a

vacuum showed no evidence for surface roughening (but did detect the later

ground shock). Hence, surface roughening (1) was due to an atmospheric, rather

than target, process.

3) Use of a low-density gas (helium) as the ambient atmosphere clearly separated the

effects of an air-coupled shockwave (initiated upon first contact of the projectile

and target) from the expanding vapor plume.

4) Winds, as visualized by Styrofoam balls and dust blowing off of dusty pipe

cleaners placed around the target, occurred immediately after the passage of the

vapor plume.

5) For 30° impacts, the projectile wake initiated motion of the Styrofoam ball tracers

uprange at late times (>1140 μs after impact), but associated winds were highly

localized.

These results provide the basis for more detailed 3D computational models that will allow the investigation of the process at larger scales. Based on both experiments at the AVGR and preliminary computational modeling in CTH, we propose that impact- wind streaks found around some craters on Mars result from impact-vapor expansion, which sets up powerful winds by displacing the atmospheric column. The extent and morphology of impact-wind streaks on Mars may help to constrain the physical processes that formed them, including the nature of the Martian surface and impacting body.

Acknowledgements

We thank the AVGR crew (Don Bowling, Charles Cornelison, Alfredo Perez,

Adam Parish, and Jon-Pierre Wiens) and China Blue for their hard work and diligence during every experimental run we performed. We would not have been able to attain this data without them. We also acknowledge and thank the North Eastern Planetary Data

Center for its 3D printing capabilities (fabrication of the pressure sensor holders used in these experiments) and for the use of images of Mars. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under

Grant (DGE-1058262), the Mars Fundamental Research Program Grant (NNX13AG43G), and a Graduate Research Fellowship from the NASA Rhode Island Space Grant

Consortium (NNX10AI95H). Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation, NASA, or the NASA

Rhode Island Space Grant Consortium.

References

Agrinier, P., Deutsch, A., Schärer, U., Martinez, I., 2001. Fast back-reactions of shock- released CO2 from carbonates: an experimental approach. Geochimica et Cosmochimica Acta 65, 2615–2632. doi:10.1016/S0016-7037(01)00617-2.

Ahrens, T.J., O’Keefe, J.D., 1977. Equations of state and impact-induced shock-wave attenuation on the moon, in: Impact and Explosion Cratering: Planetary and Terrestrial Implications. pp. 639–656.

Arvidson, R.E., 1974. Wind-blown streaks, splotches, and associated craters on Mars: Statistical analysis of Mariner 9 photographs. Icarus 21, 12–27. doi:10.1016/0019-1035(74)90086-4.

Barnouin-Jha, O.S., Schultz, P.H., 1998. Lobateness of impact ejecta deposits from atmospheric interactions. Journal of Geophysical Research 103, 25–739. doi:10.1029/98JE02025.

Bruck Syal, M., Schultz, P.H., 2014. Spatially-Resolved Spectroscopy of Impact- Generated Vapor Plumes. Lunar and Planetary Science Conference 45, Abstract 2760.

Christensen, P.R., 1982. Martian dust mantling and surface composition: Interpretation of thermophysical properties. Journal of Geophysical Research: Solid Earth 87, 9985–9998. doi:10.1029/JB087iB12p09985.

Daubar, I.J., Dundas, C.M., Byrne, S., Geissler, P., Bart, G.D., McEwen, A.S., Russell, P.S., Chojnacki, M., Golombek, M.P., 2016. Changes in blast zone albedo patterns around new martian impact craters. Icarus 267, 86–105. doi:10.1016/j.icarus.2015.11.032.

Daubar, I.J., McEwen, A.S., Byrne, S., Kennedy, M.R., Ivanov, B., 2013. The current martian cratering rate. Icarus 225, 506–516. doi:10.1016/j.icarus.2013.04.009.

Dundas, C.M., Byrne, S., McEwen, A.S., Mellon, M.T., Kennedy, M.R., Daubar, I.J., Saper, L., 2014. HiRISE observations of new impact craters exposing Martian ground ice. Journal of Geophysical Research: Planets 119, 109–127. doi:10.1002/2013JE004482.

Fergason, R.L., Christensen, P.R., Bell, J.F., Golombek, M.P., Herkenhoff, K.E., Kieffer, H.H., 2006a. Physical properties of the Mars Exploration Rover landing sites as inferred from Mini-TES–derived thermal inertia. Journal of Geophysical Research: Planets 111, E02S21. doi:10.1029/2005JE002583.

Fergason, R.L., Christensen, P.R., Kieffer, H.H., 2006b. High-resolution thermal inertia derived from the Thermal Emission Imaging System (THEMIS): Thermal model and applications. Journal of Geophysical Research: Planets 111, n/a–n/a. doi:10.1029/2006JE002735.

Fountain, J.A., West, E.A., 1970. Thermal conductivity of particulate basalt as a function of density in simulated lunar and Martian environments. Journal of Geophysical Research 75, 4063–4069. doi:10.1029/JB075i020p04063.

Gault, D.E., Wedekind, J.A., 1978. Experimental studies of oblique impact, in: Lunar and Planetary Science Conference Proceedings. pp. 3843–3875.

Glasstone, S., Dolan, P.J., 1977. The Effects of Nuclear Weapons. US Department of Defense and US Department of Energy.

Greeley, R., Iversen, J.D., 1985. Wind as a geological process on Earth, Mars, Venus and Titan. Cambridge University Press, New York.

Greeley, R., Papson, R., Veverka, J., 1978. Crater streaks in the Chryse Planitia region of Mars: Early Viking results. Icarus 34, 556–567. doi:10.1016/0019- 1035(78)90045-3.

Greeley, R., Udovich, N., Iversen, J.D., White, B., Pollack, J.B., 1974. Wind tunnel simulations of light and dark streaks on Mars. Science 183, 847–849. doi:10.1126/science.183.4127.847.

Gregg, T.K.P., 2015. Large (>1 km) Rayed Craters in Hesperia Planum, Mars: What’s the Ejecta Trying to Say? Lunar and Planetary Science Conference 46, Abstract 2442.

Holsapple, K.A., 1987. The scaling of impact phenomena. International Journal of Impact Engineering 5, 343–355. doi:10.1016/0734-743X(87)90051-0.

Ivanov, B.A., Deutsch, A., 2002. The phase diagram of CaCO3 in relation to shock compression and decomposition. Physics of the Earth and Planetary Interiors 129, 131–143. doi:10.1016/S0031-9201(01)00268-0.

Ivanov, B.A., Melosh, H.J., McEwen, A.S., HiRISE Team, 2008. Small impact crater clusters in high resolution HiRISE images, Lunar and Planetary Science Conference 39, Abstract 1221.

Ivanov, B.A., Melosh, H.J., McEwen, A.S., HiRISE Team, 2009. Small impact crater clusters in high resolution HiRISE images-II, Lunar and Planetary Science Conference 40. Abstract 1410.

Ivanov, B.A., Melosh, H.J., McEwen, A.S., HiRISE Team, 2010. New Small Impact Craters in High Resolution HiRISE Images - III. Lunar and Planetary Science Conference 41, Abstract 2020.

Iversen, J.D., Greeley, R., White, B.R., Pollack, J.B., 1976. The effect of vertical distortion in the modeling of sedimentation phenomena: Martian crater wake streaks. Journal of Geophysical Research 81, 4846–4856. doi:10.1029/JB081i026p04846.

Iversen, J.D., White, B.R., Greeley, R., Pollack, J.B., 1973. Simulations of Martian Eolian Phenomena in the Atmospheric Wind Tunnel. NASA Special Publication 36, 191–213.

Kieffer, H.H., Chase, S.C., Miner, E., Münch, G., Neugebauer, G., 1973. Preliminary report on infrared radiometric measurements from the Mariner 9 spacecraft. Journal of Geophysical Research 78, 4291–4312. doi:10.1029/JB078i020p04291.

Kök, M.V., Smykatz-Kloss, W., 2001. Thermal characterization of dolomites. Journal of Thermal Analysis and Calorimetry 64, 1271–1275. doi:10.1023/A:1011521802817.

Malin, M.C., Edgett, K.S., Posiolova, L.V., McColley, S.M., Dobrea, E.Z.N., 2006. Present-day impact cratering rate and contemporary gully activity on Mars. Science 314, 1573–1577. doi:10.1126/science.1135156.

McCauley, R.A., Johnson, L.A., 1991. Decrepitation and thermal decomposition of dolomite. Thermochimica Acta 185, 271–282. doi:10.1016/0040-6031(91)80049- O.

O’Keefe, J.D., Ahrens, T.J., 1986. Oblique impact: a process for obtaining meteorite samples from other planets. Science 234, 346–349. doi:10.1126/science.234.4774.346.

Pelkey, S.M., Jakosky, B.M., Mellon, M.T., 2001. Thermal inertia of crater-related wind streaks on Mars. Journal of Geophysical Research: Planets 106, 23909–23920. doi:10.1029/2000JE001433.

Presley, M.A., Christensen, P.R., 1997a. Thermal conductivity measurements of particulate materials 1. A review. Journal of Geophysical Research: Planets 102, 6535–6549. doi:10.1029/96JE03302.

Presley, M.A., Christensen, P.R., 1997b. Thermal conductivity measurements of particulate materials 2. Results. Journal of Geophysical Research: Planets 102, 6551–6566. doi:10.1029/96JE03303.

Presley, M.A., Christensen, P.R., 1997c. The effect of bulk density and particle size sorting on the thermal conductivity of particulate materials under Martian atmospheric pressures. Journal of Geophysical Research: Planets 102, 9221–9229. doi:10.1029/97JE00271.

Quintana, S.N., Schultz, P.H., 2014. The Formation of Crater-Related Blast Wind Streaks on Mars. Lunar and Planetary Science Conference 45, Abstract 1971.

Quintana, S.N., Schultz, P.H., 2016. A Global Distribution of Impact-Wind Streak Craters on Mars. Lunar and Planetary Science Conference 47, Abstract 1548.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2015. Experimental Results Supporting an Impact-Related Blast Wind Formation Mechanism for Some Wind Streaks on Mars. Lunar and Planetary Science Conference 46, Abstract 2469.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2016. New Experiments in Martian Impact Vapor-Induced Wind Streak Analysis. Lunar and Planetary Science Conference 47, Abstract 1553.

Sagan, C., Veverka, J., Fox, P., Dubisch, R., Lederberg, J., Levinthal, E., Quam, L., Tucker, R., Pollack, J.B., Smith, B.A., 1972. Variable features on Mars: Preliminary mariner 9 television results. Icarus 17, 346–372. doi:10.1016/0019- 1035(72)90005-X.

Schmidt, R.M., Housen, K.R., 1987. Some recent advances in the scaling of impact and explosion cratering. International Journal of Impact Engineering 5, 543–560. doi:10.1016/0734-743X(87)90069-8.

Schultz, P.H., 1992a. Atmospheric effects on ejecta emplacement. Journal of Geophysical Research: Planets 97, 11623–11662. doi:10.1029/92JE00613.

Schultz, P.H., 1992b. Atmospheric effects on ejecta emplacement and crater formation on Venus from Magellan. Journal of Geophysical Research: Planets 97, 16183– 16248. doi:10.1029/92JE01508.

Schultz, P.H., 1996. Effect of impact angle on vaporization. Journal of Geophysical Research: Planets 101, 21117–21136. doi:10.1029/96JE02266.

Schultz, P.H., Eberhardy, C.A., 2015. Spectral probing of impact-generated vapor in laboratory experiments. Icarus 248, 448–462. doi:10.1016/j.icarus.2014.10.041.

Schultz, P.H., Gault, D.E., 1979. Atmospheric effects on Martian ejecta emplacement. Journal of Geophysical Research: Solid Earth 84, 7669–7687. doi:10.1029/JB084iB13p07669.

Schultz, P.H., Gault, D.E., 1982. Impact ejecta dynamics in an atmosphere: Experimental results and extrapolations. Geological Society of America Special Papers 190, 153–174. doi:10.1130/SPE190-p153.

Schultz, P.H., Gault, D.E., 1990. Prolonged global catastrophes from oblique impacts. Geological Society of America Special Papers 247, 239–262. doi:10.1130/SPE247-p239.

Schultz, P.H., Quintana, S., 2013. Impact Blast Wind Scouring on Mars. Lunar and Planetary Science Conference 44, Abstract 2697.

Schultz, P.H., Quintana, S.N., 2017. Impact-generated winds on Mars. Icarus 292, 86– 101. doi:10.1016/j.icarus.2017.03.029.

Schultz, P.H., Wrobel, K.E., 2012. The oblique impact Hale and its consequences on Mars. Journal of Geophysical Research 117, E04001. doi:10.1029/2011JE003843.

Sugita, S., Schultz, P.H., Adams, M.A., 1998. Spectroscopic measurements of vapor clouds due to oblique impacts. Journal of Geophysical Research: Planets 103, 19427–19441. doi:10.1029/98JE02026.

Thomas, P., Veverka, J., Lee, S., Bloom, A., 1981. Classification of wind streaks on Mars. Icarus 45, 124–153. doi:10.1016/0019-1035(81)90010-5.

Veverka, J., Gierasch, P., Thomas, P., 1981. Wind streaks on Mars: Meteorological control of occurrence and mode of formation. Icarus 45, 154–166. doi:10.1016/0019-1035(81)90011-7.

Veverka, J., Sagan, C., Quam, L., Tucker, R., Eross, B., 1974. Variable features on Mars III: Comparison of Mariner 1969 and Mariner 1971 photography. Icarus 21, 317– 368. doi:10.1016/0019-1035(74)90046-3.

Wechsler, A.E., Glaser, P.E., 1965. Pressure effects on postulated lunar materials. Icarus 4, 335–352. doi:10.1016/0019-1035(65)90038-2.

Woodside, W., Messmer, J.H., 1961. Thermal conductivity of porous media. I. Unconsolidated sands. Journal of Applied Physics 32, 1688–1699. doi:10.1063/1.1728419.

Tables

Table 1 - Summary of Experimental Setups

Setup Description 1 Styrofoam ball tracers sprinkled on target surface Styrofoam ball tracers on aluminum pedestals (above) or small shelves 2 (below) target 3 Calibration plate extends target surface Tracers (Styrofoam balls or dolomite) outside the target, on the 4 calibration plate 5 Pipe cleaners dusted with dolomite powder Darkened chamber (no sun lamps) in order to capture self-luminescent 6 vapor 7 Vapor-suppressing target materials

Table 2 - List of AVGR Experiments

Velocity Angle Pressure Shot (km/s) (deg) (Torr) Gas Scenario Notes 130502 4.94 30 0.34 Air 1 - 130503 4.87 30 25.47 Argon 1 - 130505 4.71 30 10.28 Argon 1 - 130506 5.33 45 25.32 Argon 1 - 130507 5.46 45 0.41 Air 1 - 130508 5.12 90 25.36 Argon 1 - 130509 4.96 90 25.35 Argon 1 - 130510 5.05 90 0.37 Air 1 - 130511 5.09 30 0.39 Air 1 - 130512 4.85 60 25.43 Argon 1 - 130513 5.22 60 25.28 Argon 2 - 130514 5 60 0.45 Air 2 - 130515 5.39 45 25.29 Argon 1 - 130516 4.9 45 25.39 Argon 1 - 130517 4.66 45 25.3 Argon 1 - 140901 5.39 45 24.25 Air 3 - 140902 5.31 45 24.68 Air 3 - 140903 5.49 45 25.18 Air 4 - 140904 5.49 45 25.24 Argon 4-5 - 140905 5.68 30 24.83 Air 4-5 - 140906 5.21 90 25.27 Air 4-5 - 140907 5.09 60 24.95 Air 4-5 - 140908 5.32 45 25.59 Air 4-5 Cluster Impact 140909 5.04 45 25.03 Air 4-5 - 140910 5.76 45 25 Air 3 - 140911 5.47 45 25.5 Air 3 - 140912 5.48 30 25 Air 3 - 140913 5.37 60 24.94 Air 3 - 140914 5.79 45 24.78 Air 3 - 140915 5.68 30 25.34 Air 3 - 150803 6.04 45 25 Air 6 - 150804 5.24 45 25.1 Air 6 - 150805 5.28 45 25.21 Argon 6 - 150806 5.34 45 25.27 Helium 6 - 150807 5.37 45 25.32 Helium 6 - 150808 5.89 45 25.37 Helium 6 - 150809 5.33 60 25.3 Helium 6 - Note: Table 2 continues on next page.

Table 2 continued

Velocity Angle Pressure Shot (km/s) (deg) (Torr) Gas Scenario Notes 150810 5.66 30 25.28 Helium 6 - 150811 5.06 30 25.63 Argon 6 - 150812 5.47 60 25.43 Argon 6 - 150813 5.33 90 25.29 Argon 6 - 150814 5.38 90 25.3 Helium 6 - 3 mm layer of 150815 5.41 45 25.11 Argon 6, 7 pumice over dolomite target 13 mm layer of 150816 5.28 45 25.19 Argon 6, 7 pumice over dolomite target 160901 4.88 45 24.32 Air 5 - 160902 5.2 45 24.54 Air 5 - 160905 5.25 45 25.2 He 5 - 160906 5.33 45 25.23 Ar 5 - 160907 4.91 60 25.2 Ar 5 - 160908 5.02 30 25.14 Ar 4, 5 - 160917 5.49 45 49.85 Air 4, 5 -

110

(m/s)

Winds

Reverse

100

140

(m/s)

Winds

Outward

LATERAL

42.8

36.2

29.8

24.6

19.3

29.4

23.1

18.4

13.1

(cm)

Point

From

Impact

Distance

150

100

110

130

110

100

170

100

110

(m/s)

Winds

Reverse

200

220

230

200

190

250

180

100

150

230

150

(m/s)

Winds

Outward

UPRANGE

40.6

34.7

28.6

23.9

19.0

28.9

22.9

18.8

14.3

25.7

22.6

21.2

26.0

25.0

22.6

26.4

26.0

(cm)

Point

From

Impact

Distance

100

110

140

130

160

120

110

140

100

(m/s)

Winds

Reverse

140

160

210

200

210

200

190

160

190

280

120

110

170

160

240

190

200

(m/s)

Winds

Outward

DOWNRANGE

41.6

35.4

29.2

23.6

18.7

47.9

41.7

35.5

29.3

24.2

19.0

24.1

25.5

23.1

23.7

24.2

22.9

22.6

23.4

27.3

(cm)

Point

From

Impact

Distance

5.2

4.88

5.04

5.32

5.09

5.21

5.68

5.49

Speed

(km/s)

Impact

Angle

Wind speeds as determined by frame differencing of dust pipe cleaner dust streams pipeof differencing by frame speedsdetermined Wind as

Shot

160902

160901

140909

140907

140906

140905

140903

140904

140908

Table 3 continued onpage. next

Table 3 Table

(m/s)

Winds

Reverse

(m/s)

Winds

Outward

LATERAL

32.4

26.7

21.1

31.8

26.1

20.4

26.3

20.5

(cm)

Point

From

Impact

Distance

230

190

160

100

(m/s)

Winds

Reverse

150

170

110

180

160

230

310

(m/s)

Winds

Outward

UPRANGE

24.0

19.1

14.7

35.6

29.9

24.2

26.8

22.1

17.8

32.6

27.2

22.3

18.2

(cm)

Point

From

Impact

Distance

100

110

(m/s)

Winds

Reverse

120

180

150

130

110

170

190

200

190

210

130

120

160

180

250

(m/s)

Winds

Outward

DOWNRANGE

50.4

44.4

38.5

32.9

26.4

21.2

49.7

44.0

37.1

31.1

25.4

20.2

35.3

29.8

23.6

41.7

35.4

29.6

23.7

18.6

(cm)

Point

From

Impact

Distance

5.49

5.02

4.91

5.33

Speed

(km/s)

Impact

Angle

Shot

Cluster impact (projectile was disrupted before collision with target) with collision before disrupted was (projectile impact Cluster

Ambient atmosphere was approximately 67 mbar (all other shots at 33 mbar) at shots (all other 67 mbar was approximately atmosphere Ambient

Impact into an argon atmosphere argon into an Impact

160917

160908

160907

160906

a Table 3 continuedTable

Table 4 - Total errors associated with dust streamer edge-determination in high-speed camera frames (in m s-1)

Runs Runs beginning beginning 1409XX 1609XX Downrange 60 60 Uprange 70 60 Lateral N/A 50

Figure Captions

Figure 1 - THEMIS nighttime infrared mosaic of the area around ~70 km diameter Pál crater (108°E,31°S), Mars. The bright streaks (arrows) appearing as tails from preexisting craters (craters that predate the Pál impact) can be traced back to crater Pál.

Such wind streaks are visible primarily in thermal infrared images, though they can be seen in high resolution visible images, as well. Impact-wind streaks are not present around every crater on Mars.

Figure 2 - Diagram of the AVGR test chamber setup used in the experiments described herein. (1) is the test chamber, a cylindrical chamber ~2.5 m in diameter, ~2.5 m high.

The target container (2) is placed within a well, and a calibration plate (3) is placed over the well in order to extend the target surface. A small gap separates the calibration plate from the target surface. Ports at 15° (4), 30°, 45°, 60°, and 90° allow for testing oblique impacts into particulate and water targets, in addition to solid ones. A camera may also be mounted on the 15° port to record along the impact trajectory. A large viewing window (5) allowed up to four cameras to record lateral to the impact. This study used up to eight cameras (6) positioned at the viewing window, as well as other ports and viewing windows around and on top of the chamber.

Figure 3 - Schematic of the most frequently used sensor setup within the impact chamber at the AVGR. The projectile enters the chamber from the top in this figure. The majority of the high-speed cameras record the process from a side (lateral) viewing window

(although two other cameras, positioned in higher viewing windows also provided context and calibration for the impact). The target is surrounded by a calibration plate,

35 on which the various sensors and microphones are attached via magnetic bases. The rings on the calibration plate (in grey, above) are separated by approximately 15 cm.

Note: schematic is not to scale.

Figure 4 - Target chamber diagram (a) and tracer setup (b). Within the chamber, the target is prepared in the center of the platform (a1) within a well (a2) approximately 90 cm in diameter, such that the target surface is level with the platform. The target container (a3) is 59 cm in diameter, and may hold particulate or liquid target material (a4).

Some experiments for this contribution utilized a calibration plate (b1) that nearly abutted the target container. A small gap was left around the edge of the container in order to prevent ground-coupled shockwaves from transferring to the plate. Styrofoam balls (b2, millimeter in size), sprinkled around the edge of the target and sometimes on the target itself helped to track the winds and the development of a turbulent boundary layer near the surface. Pipe cleaners dusted with powdered dolomite (b3) allowed tracking the development of winds, as well as determining the wind speeds. Finally, powdered dolomite sprinkled on the calibration plate (b4) helped to visually track surface effects

(e.g., roughening, see Figure 5).

Figure 5 - High speed cameras captured a roughening of the target surface, which appeared as a subtle darkening of the target that traveled supersonically and asymmetrically away from the impact point. Due to the camera angle and the position of the lights within the chamber, this roughening is not a shadow, but rather an impact- related effect. The sequence of images on the left step through time as the surface roughening expands both uprange and downrange. The images on the right are the same

36 as those on the left except the surface roughening is traced in a red dashed line. In the top image, the vapor plume (cyan dashed line) can also be seen. This figure depicts Shot

140910, a 45° impact of a Pyrex projectile into powdered dolomite at ~5.8 km/s (impact from the right, white arrow). The ambient atmosphere was approximately 33 mbars of air.

Figure 6 - Overall surface roughening speeds as a function of impact angle for various impacts through an ambient atmosphere of 33 mbars of air. Here, the overall speed is defined as the time at which the surface roughening reaches the edge of the target container. A clear asymmetry exists between the downrange and uprange surface roughening speeds, likely due to the added downrange momentum of the impact. The roughening was diffuse at high impact angles.

Figure 7 - Representative surface roughening results and related error for impacts into air

(top row), argon (middle row) and helium (bottom row) at 45°. Downrange results are the panels on the left, and uprange results are the panels on the right. Errors results from measurement uncertainty in the determination of the edge of the surface roughening.

Four separate measurements of the surface roughening were made in order to determine the representative error indicated here.

Figure 8 - Comparison of the average downrange surface roughening speed, as calculated from high-speed camera images, with the overall downrange ground-coupled shockwave speed, as recorded by geophones placed on the outside of the target container,

4.5 cm below the target surface. The differences in speeds clearly demonstrate that the two phenomena are separate events.

Figure 9 - A frame taken at 250 μs after impact of a Pyrex projectile into a powdered dolomite target at 5.2 km/s depicting how the surface roughening and vapor expansion are coupled. The surface roughening is shown in the bottom quarter of the image, outlined in a black dashed line. The vapor plume, shown in the stretched top three quarters of the image is outlined in a white dashed line. The white dotted line indicates that the surface roughening aligns with the outer edge of the expanding vapor plume.

This frame is taken from Shot 160902. The double image of the vertical pipe cleaners

(excepting the first one) is the result of multiple reflections in the window and safety plastic revealed in the contrast stretched image at top.

Figure 10 - Exponential fit for surface roughening speeds for impacts at 45° under a helium atmosphere used to obtain equation 1.

Figure 11 - The development of winds is visualized by dusty pipe cleaners placed near the edge of the target. Within 35 μs after the passage of the vapor plume (surface roughening), winds begin to blow the dust off of the pipe cleaners. Initially, the winds blow outward (a), away from the crater center. Later, the winds reverse direction (c) and blow back toward a point just downrange of the growing crater. This reversal is a response to the partial vacuum left by an early, hot, fast-moving vapor plume that is unrelated to the plume that mobilizes the pipe cleaner dust outward. The partial vacuum of this early vapor plume mobilizes the dust from both the downrange and uprange pipe cleaners. (b) and (d) are enhanced and enlarged views of the boxes shown in (a) and (c), respectively, with arrows to highlight the dust being blown off of the pipe cleaners.

Wind speed estimates were made from images like these by reconstructing the geometry

38 of the pipe cleaners, camera positions, and impact point. This figure is from Shot 140903, a 45° impact at about 5.5 km/s through an ambient atmosphere of approximately 33 mbar air. (a) and (b) are ~580 μs after impact. (c) and (d) are ~1000 μs after impact.

Figure 12 - Impact wind development process diagram for an impact into a low-density atmosphere as determined by experiments. The diagram is not to scale, but is rather intended to show observed trends. (a) A wake shock (WS, blue dashed line) expands cylindrically around the impactor as it travels through the atmosphere at hypervelocities.

Directly after impact, a shockwave propagates through the atmosphere (air-coupled shockwave, red dashed line) and the target (ground-coupled shockwave, white dashed line). The air-coupled shockwave is denoted AS, whereas the ground-coupled shockwave is denoted GS. A hot (luminescent) plume expands rapidly downrange and upward (red region). Impact direction is indicated with an arrow. The black line downrange represents a dusty pipe cleaner within the target (as in the experiments). (b)

At slightly later times (~250 μs after impact in the experiments in air), the air-coupled shockwave quickly outpaces the ground-coupled shockwave. Initially, the vapor is contained within the transient crater, which results in hemispherical expansion above the target. Some asymmetry exists, as the vapor travels faster downrange than uprange. The vapor meets and may overtake the air-coupled shockwave downrange, but in low-density atmospheres (helium), the shockwave outpaced the vapor plume uprange. The vapor plume overpowers the wake shock. Passage of this vapor plume initiates impact winds that remove dust from pipe cleaners both downrange and uprange, as indicated by the black dotted lines extending from the pipe cleaner. (c) Half of the formed crater and ballistic ejecta deposit at much later time (~115 ms in air). Flow separation induces 39 vortices at the top edge of the ejecta curtain, while a basal vortex may scour the ejecta facies and cause extensive runout of ejecta (Schultz, 1992; Barnouin-Jha and Schultz,

1996, 1999a, 1999b). These vortices are unrelated to the earlier impact winds, but rather come much later (after the ejecta is deposited).

Figures

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

CHAPTER TWO:

Numerical Models of Impact-Induced Winds on Mars

Stephanie N. Quintana1

and

Peter H. Schultz1

1Department of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI 02912

Abstract

Thermally bright impact-wind streaks, as revealed in Thermal Emission Imaging System

(THEMIS) images radiate from certain craters on Mars. These streaks are interpreted to form from impact vapor expansion into the atmosphere of Mars, which initiates strong winds. When these winds interact with topographic obstacles (e.g., preexisting craters), they generate surface-scouring vortices travelling downrange. A suite of three- dimensional simulations (models) in the shock physics package, CTH, explores the conditions necessary to generate high vaporization and intense winds upon impact.

Models investigate the effects of target and impactor properties, such as composition, impact speed, and impact angle. High-speed impactors (> 12 km s-1) and ice-rich targets or impactors are more likely create conditions necessary to form impact-wind streaks.

Dunite (asteroid) impactors traveling at speeds above 20 km s-1 are unlikely at Mars, and surface-ice layers would be restricted to high latitudes in contrast to most observed occurrences of impact-wind streak craters. The model results reveal that comet impactors

(and perhaps transient thick ice-layered terrains) are most likely to produce impact-wind streaks. In high-vapor generating cases, the vapor plume and resulting winds envelop the surface, but a topographic obstacle is needed for vortex development downwind.

1. Introduction

Wind streaks are common features on Mars. Many of the most easily identifiable examples extend from craters and are often found in groups with similar directions in a given area. Wind streaks such as these were originally coined ‘variable features’ (Sagan et al., 1972), and some are transient in nature. Some examples have changed shape, orientation, or have even disappeared over human timescales. These types of wind streaks respond to and change with global wind patterns and global dust storms (Greeley et al., 1974; Greeley et al., 1974; Greeley and Iversen, 1985; Pelkey et al., 2001).

The focus here, however, is on a subset of crater-related streaks revealed by the

Mars Odyssey’s Thermal Emission Imaging System (THEMIS) (e.g., Schultz and Wrobel,

2012; Schultz and Quintana, 2013; Quintana and Schultz, 2014; Quintana et al., 2015c,

2016; Quintana and Schultz, 2016; Schultz and Quintana, 2017). These streaks are thermally bright in the THEMIS nighttime infrared images, and they radiate away from certain Amazonian and Hesperian age craters on Mars (e.g., Schultz and Quintana, 2013;

Schultz and Wrobel, 2012). Unlike the variable features described above, the streaks here are formed as part of the impact process itself. For convenience, we label these thermally bright wind streaks impact-wind streaks.

Detailed examination of visible surface processes associated with Santa Fe crater revealed that thermally bright streaks are the thermal manifestation of surface scouring by strong wind-related vortices (Schultz and Quintana, 2017). This study examined observational constraints on crater-related streak formation and found that impact-wind streaks precede secondary cratering or ejecta emplacement. Because of their long, straight tails, that study also concluded that the winds must be strong and sustained due to

55 interactions between the impact-generated vapor and the Martian atmosphere. The paucity of impact-wind streak craters on Mars requires that some sort of impact-vapor enhancement is necessary to produce the impact-wind streaks; otherwise, all well- preserved craters should display this pattern. Enhanced vaporization may be a consequence of target properties (composition), or impactor properties, such as composition, impact speed, or impact angle.

Another study (Quintana et al., 2015c, 2016, Chapter 1) assessed experiments performed at the NASA Ames Vertical Gun Range (AVGR) in order to further test the hypothesis that expansion of impact vapor coupled to the ambient atmosphere could be responsible for the intense winds. The large impact chamber at the AVGR allowed sensors, tracers, and cameras to record the impact process in a series of hypervelocity impact experiments. While such experiments allow exploration of the underlying processes, they cannot fully replicate an impact on Mars; rather, they allow isolating variables that inform computational modeling, which is the focus of this contribution.

The objective here is to assess the conditions needed to develop high amounts of impact vaporization, strong winds, and surface-scouring vortices on Mars.

2. Computational Models

Computational models (simulations) are essential in the study of impact-generated winds because they give access to more realistic conditions than laboratory experiments.

For example, models allow for planetary-scale studies with appropriate geologic materials and impact speeds. What is more, planetary-scale impact models are ideal for testing suites of scenarios in order to isolate and explore the effects of particular variables.

This project uses the CTH (version 11.2) shock physics analysis package

(McGlaun et al., 1990; Hertel et al., 1993), which was developed at Sandia National

Laboratories. CTH is composed of a family of codes that are suited for the study of strong shock and high deformation processes like those in planetary impact cratering.

CTH allows for the analysis of multi-material systems in 1D, 2D, or 3D (McGlaun et al.,

1990; Hertel et al., 1993) through an Eulerian framework. In such a framework, a continuum variable is treated from a fixed reference frame and materials move through a fixed mesh.

3D models are best employed to study the impact-wind streak generation process on Mars. 2D models use axial symmetry to decrease the complexity of the calculation, but they can then only be used to model vertical (90°) impacts when in fact most natural impacts occur nearer to 45° (Gilbert, 1893; Shoemaker, 1962). Numerous examples from the laboratory (e.g., Schultz, 1996; Sugita et al., 1998; Sugita and Schultz, 1999; Schultz and Eberhardy, 2015) and in models (e.g., Ivanov and Artemieva, 2002; Pierazzo and

Melosh, 1999, 2000; Quintana et al., 2015a) demonstrate that impact-angle effects are also important for impact melt and vapor production. While 3D models are much more computationally expensive than 2D ones, CTH is parallelizable, which allows it to run more quickly on multiple processors. CTH also utilizes adaptive mesh refinement (AMR,

Crawford, 1999), which allows for variable resolution throughout the model. The user may set specific areas of high resolution without the need to refine the entire problem space. Thus, AMR decreases computation time without the loss of refinement where it is needed most.

Codes such as CTH require an equation of state (EoS) in order to describe the thermodynamic properties of materials within an impact calculation. Unless otherwise noted, this study primarily used the sophisticated analytical equation of state, ANEOS

(Thompson and Lauson, 1972; Thompson, 1990; Melosh, 2007) because of its reliability and stability, especially for geological materials of interest (e.g. dunite, basalt, and ice).

ANEOS calculates temperature and pressure simultaneously, making it an ideal EoS for this study. It is also widely used in the planetary impact community because it is the most accurate equation of state currently available (Collins and Melosh, 2014) and it is the primary equation of state used in another common impact cratering code, iSALE

(Amsden et al., 1980; Elbeshausen and Wünnemann, 2011).

Recent improvements to ANEOS have been made for some geologic materials

(Melosh, 2007), further enhancing its usefulness in this study. Melosh (2007) and others have since defined ‘molecular’ versions of some library materials, which allows for better treatment of vaporization. The ‘molecular’ material vaporizes into its appropriate molecular gas phase, instead of an atomic gas phase. Table 1 lists the equations of state used in this study, and Table 2 lists the appropriate ANEOS parameters for these materials taken from the library definitions for each material (unless otherwise noted) in version 11.2.

It is important to note that computational models are numerical representations of physical processes and, as such, have limitations. Previous work (Quintana et al., 2013;

Collins and Melosh, 2014; Quintana et al., 2015a, 2015b) discovered some inconsistencies with the treatment of melting and vaporization in ANEOS for hydrodynamic calculations that would also apply to more complex calculations such as

58 the ones here. For example, self-consistent values for peak pressure for melting

(calculated from experimentally-determined entropy of melting) differed between

Quintana et al. (2015a) and Pierazzo et al. (1997) by ~10% for materials used in this study. In order to combat these inconsistencies, this work tracked vaporization using a variety of methods with self-consistent criteria in order to give a range of estimates.

2.1 Vapor Determination in CTH

A subroutine with set criteria recorded the vaporized masses of each material in the calculation determined the amount of vapor generated in each model. This method used two commonly accessible variables for determining melting and vaporization in an impact: peak pressure and final release state temperature (e.g. Pierazzo et al., 1997; Barr and Citron, 2011; Quintana et al., 2013, 2015a, 2015b). The peak pressure method, similar to the critical entropy method (Kraus et al., 2011), uses the peak shock pressure that a material experiences (as a surrogate for entropy) as the criterion to determine whether that material has melted or vaporized. Similarly, the final release-state temperature method uses the temperature of the material after the shockwave has passed and the material has been released to an unshocked state. The subroutine employed here used a given peak pressure and release-state temperature as a cutoff and records the mass of any material above that set value.

Pierazzo et al. (1997, 2005) list the peak pressures for vaporization of dunite, ice, and basalt. From those values, ANEOS produced self-consistent values of the vaporization temperature. This step was necessary because Pierazzo et al. (1997, 2005) did not provide values for the vaporization temperatures, nor did the JANAF thermochemical tables (Chase et al., 1995), except for ice. For ice, the subroutine

59 recorded both the vaporization temperatures derived by Pierazzo et al. (1997) and the

JANAF value. Table 3 displays the critical values for peak pressure and final release- state temperature used in this study.

Collins and Melosh (2014) indicated that the shock pressure required for vaporization may be overestimated in ANEOS, leading to an underestimation for the vapor generated. Consequently, density provided an additional vapor-determination criterion to the more conventional pressure and temperature ones. Specifically, a subroutine recorded all materials with a density low enough to be considered vapor. This cutoff was taken to be 4x10-4 g cm-3. Because porous ice is an important component in many of the models, the cutoff density for vapor had to be sufficiently lower than that of the ice. The density subroutine provided an upper estimate for the amount of vapor generated in the impact.

3. Methods

3.1 Planetary-Scale Simulations

A series of 3D computational simulations with CTH stepped through variable space in order to test target composition, impactor composition, impact angle, and impact speed. This suite of models tested which conditions contribute to enhanced vaporization compared to a ‘nominal’ impact on Mars. That is, what is the underlying cause of the impact-wind streaks on Mars, and why is there a relative paucity of craters bearing such streaks?

For simulations relevant to Mars, the calculations assumed that the surface temperature is 213.5 K with a surface pressure is 6.5 mbar (650 Pa) of CO2 (density of

1.55e-5 g cm-3). The atmosphere was modeled as an ideal gas, rather than by the ANEOS

60 package. The basalt target material was modeled with the Brittle Damage with Localized

Thermal Softening (BDL) strength and damage model developed by Crawford and

Schultz (2013). This model is similar to the damage model of Collins et al. (2004), but improves upon it by including an estimate of the statistical crack spacing in the damaged material, shear heating within the cracks, and conductive heat loss from the cracks

(Crawford and Schultz, 2013).

We tested scenarios based on two major factors: target properties and impactor properties (see Table 4). In each scenario, a 1.5 km diameter impactor struck a semi- infinite slab that represents the Martian surface. Both target and impactor properties can drive vapor generation, but testing each of these factors separately allows for further clarification about characteristics that could lead to the thermal expressions on Mars.

Impact-wind streaks on Mars appear most predominantly downwind of sufficiently high topographic obstacles and are due to scouring by trailing horseshoe vortices. As winds sweep across the surface, a gradient in the fluid flow (boundary layer) develops near the surface and can protect the surface from the most intense winds

(Greeley et al., 1980; Greeley and Iversen, 1985). A topographic obstacle that extends through or disrupts the flow will establish vortices on the downwind side, which effectively bring the full intensity of the winds to the surface (Greeley et al., 1974;

Greeley and Iversen, 1985).

Consequently, each modeled scenario included a 500 m high cylindrical obstacle,

1 km in diameter, emplaced at 25 km downrange of the impact. The obstacles in these simulations represent some form of preexisting topography, such as a crater rim, that initiated surface-wind interactions. A more realistic crater model (i.e. the combination of

61 a high-relief obstacle surrounding a depression) would only enhance this process. In reality, impact-wind streaks are often found much farther than 25 km from the impact point. The primary ejecta deposit could cover expressions of the wind streaks for obstacles within about a crater radius, (e.g., Schultz and Quintana, 2017). The relatively close obstacle distance chosen here served to reduce the model computation time while allowing for longer observation before ejecta is deposited. For comparison, two longer simulations tested the effects of an obstacle at 60 km from the impact point.

While the code can simulate vortex development within the vapor plume in response to an obstacle, CTH can only provide a first order estimate of such vortices.

Specialized codes such as computational fluid dynamics (CFD) codes would be better suited for analyzing the precise vortex formation patterns. The objective of this work, however, is to track the general effect of an obstacle on the response of the impact- generated winds. The detail of a CFD code is unnecessary here but will be considered in the future.

Target-property models tested whether some aspect of the surface of Mars could be responsible for enhanced vaporization with respect to a ‘control’ case that represents nominal Mars. In the control case, a solid dunite sphere served as the model for an asteroid impactor striking the planet. The surface was simply modeled as a solid basalt slab half-space, which was then modified to include: a surface water ice layer, a subsurface water ice layer, or a wet tuff layer, all in varying thicknesses (Table 4, above).

Note that wet tuff was modeled with the SESAME equation of state, rather than ANEOS, because of its stability in the calculations.

As a comparison, impactor-property models tested whether the impactor speed and/or composition could play a role in enhancing vaporization upon striking Mars. For these cases, impact speeds ranged from 6 km s-1 to 40 km s-1. One series of models tested different projectile impact speeds on the surface water ice layer case and the wet-tuff case, in order to assess the combined effects of increased impactor speed on unique target properties. A second series of tests explored impactor composition by simulating a cometary impact on Mars with a ~60% porous impactor striking a solid basaltic target.

This porosity was chosen to roughly match an estimate of the density of Comet 9P

Tempel 1 (Schultz et al., 2007; Richardson et al., 2007) and Comet 67P/Churyumov-

Gerasimenko (Sierks et al., 2015). Because the comet is modeled solely as ice, the porosity required to attain a density of 0.4 g/cm3 is less than porosity estimates of Comet

67P/Churyumov-Gerasimenko, which is 70-80% (Sierks et al., 2015).

3.2 Data Acquisition and Resolution

Seventy-two fixed tracer particles, distributed above the surface at varying distances downrange, uprange, and orthogonal to the impact point (Figure 1), recorded the regional response to impact processes. Fixed tracers allow materials to flow through them and recorded material density, pressure, and velocity. Ten tracers were positioned between 100 m and 5 km of the surface, followed by one tracer each at 10 and 15 km from the surface. Sets of 12 tracers, as described above, were positioned at 15 km downrange of the impact, followed by 25 km downrange, 30 km downrange, 40 km downrange, 15 km uprange, and 15 km orthogonally.

CTH can plot model variables recorded through the calculation, such as pressure, temperature, velocity, and density on-the-fly (as the calculation is running) or through

63 post-processing. Plots here primarily focused on density and velocity. Density visualizations reveal vapor plume-surface interactions, whereas velocity plots are particularly useful for determining the effects of the downrange obstacle. X-velocity is defined as being along the impact trajectory, while y-velocity is orthogonal. Data plots, such as these, combined with the tracer data allow for testing various wind-streak assessment criteria.

The models here took advantage of CTH’s AMR capability, focusing resolution in user-defined areas. The highest refinement of 25 cells per projectile radius was achieved closest to the impact point. In order to clearly visualize the interaction of the expanding vapor and the surface, the surface was also highly refined. Cells in the vicinity (±5 km from the atmosphere-target boundary, 50 km downrange, and 20 km uprange) measured

122 m on a side. The number of 3D models needed in this study necessitated a compromise on resolution. Higher resolution may provide more accurate melting and vaporization estimates, but Pierazzo et al. (1997) states that a resolution of at least 20 cppr is enough to reduce any error in such estimates to about 10%. Tracers recorded data at a temporal resolution of 5 ms, while data for image processing was recorded every 0.1 s. Calculations ran to a model time of one minute.

4. Results

The models served four purposes: 1) determine the relative amounts of vapor generated by each scenario (Table 4); 2) calculate wind speeds for each scenario at select locations around the crater; 3) assess the effects of an obstacle downrange from the impact point; and 4) reveal the relationship between the air-coupled shockwave and vapor expansion.

4.1 Vaporization

In order to determine the amount of vapor generated in each model, a set of subroutines (described above) recorded the vaporized masses of each material in the calculation. The subroutine results allow for some generalizations to be made. Table 5 lists the mass of impact-vaporized material (determined using the subroutines described above) normalized by projectile mass. High vapor mass ratios (>0.8 projectile masses) occurred in models that included ice in either the projectile or target, when compared to low-speed control models. Impact speed, however, plays an important role. As might be expected, higher impact speeds (>12 km s-1) result in more vaporization regardless of target or impactor composition. No significant vapor formed for impacts at speeds <20 km s-1 for control and tuff-layer cases (Scenarios 1, 2, and 10, respectively).

The density-vapor criterion yielded much higher vapor masses than the two other criteria, which is more consistent with laboratory results (Schultz, 1996; Schultz and

Eberhardy, 2015). The fastest dunite (asteroid) and ice (comet) impactors generated the highest mass ratios of vapor (Scenarios 4 and 14, respectively). The next highest- producing scenarios (surface ice Scenarios 5-9) generated comparably much less vapor.

In some cases, the surface-ice targets yielded nearly two times less vapor than did a 40 km s-1 impact of either an ice (comet) or dunite (asteroid) impactor. Heated silicate target material may contribute to vaporization of the ice below in the thickly layered terrain case (Scenario 9) (Schultz and Eberhardy, 2015).

Given the results presented here, runs with the highest impact speeds and those with water ice in surface and substrate layers generated the most vapor. But ice impactors (comets) consistently generated the highest vaporization with each criterion

65 tested (peak pressure, final release-state temperature, and density). In addition, only at the highest speeds was the amount of vaporization generated for surface ice layers comparable to that generated from comets, even at low impact speeds.

4.2 Wind Speeds

The wind-speed analysis used a subset of tracers from the 72 total model tracers.

This subset included four near-surface tracers positioned 100 m above the surface at distances of 30 km downrange, 40 km downrange, 15 km orthogonal to the impact trajectory, and 15 km uprange. In addition to those four, the analysis also considered tracers positioned 590 m above the surface at the same distances. These tracers tracked the near surface wind development radially to the impact point. Tracers used to track higher altitude atmospheric response included those between 2.3 and 5 km in altitude at the same distances as above.

Wind speed records in the x-direction (y-direction for the orthogonal tracers) often cataloged a spike in wind speed. The pressure records also catalogued a spike that was assumed to be the air-coupled shockwave initiated at first contact between the impactor and target surface. If the velocity peak corresponded in time with the pressure peak (within 0.3 seconds at a distance of 40 km from the impact point), then the peak wind speed was attributed to the passing shockwave. In order to estimate the true peak wind speeds experienced at each tracer (away from the shockwave), the shockwave peak was ignored and the next highest peak was recorded. Additional data of interest included the minimum wind speed peak. For tracers placed downrange, the minimum wind speed peak was the peak speed of winds traveling back toward the crater, which corresponded to either reverse winds (Chapter 1) or vortex development (Quintana and Schultz, 2014).

Another recorded value was the average wind speed experienced at a given tracer location. This value was taken as an estimation of the overall sustainability and direction of the winds.

Table 6 lists the results of the models. Downrange, outward speeds are negative values (because they travel in the negative x-direction). Uprange, negative values indicate inward-directed winds. A maximum inward wind speed of zero indicates that winds were not directed inward. In such cases, vortices downrange of the obstacle may still have formed but resulted only in a reduction of the outward wind speed. In general, wind speeds exceeding a severe tornado on Earth (>70 m s-1, Hyndman and Hyndman,

2016) were considered exceptionally high speed because such winds may mobilize centimeter or larger sized material on Mars (Iversen et al., 1976).

A general overview of the resulting data in Table 6 indicates that the overall surface winds are fastest for the control cases (Scenarios 1-4). For example, the control impact at 20 km s-1 (Scenario 3) generated outward-blowing winds averaging 230 m s-1 over 48 s (excluding 12 s around the shockwave-induced peak) at a distance 40 km downrange of the impact point. Comparatively, at 20 km s-1 over the length of the calculation (60 s) an ice impactor (Scenario 13) yielded an average of 140 m s-1 of inward-blowing (reverse/vortex) winds at the same distance, whereas a thin surface ice layer (Scenario 8) had an average of 65 m s-1 of reverse/vortex winds. The peak wind speeds (not associated with the shockwave speeds) for a 20 km s-1 impact were 530 m s-1,

380 m s-1, and 12 m s-1 (all outward) for Scenarios 3, 13, and 8, respectively. In comparison, a faster-moving ice impactor with a speed of 40 km s-1 (Scenario 14)

67 produced consistently high peak (300 m s-1) and average (90 m s-1) wind speeds at 40 km from the impact point.

While silicate vapor generated from an asteroid impact yielded higher speeds in the tracer data, it is important to note the combined speed and density of the vapor.

Silicate vapor from either the control cases or the tuff-layer cases will naturally have higher speeds because silicate vapor exhibits much higher temperatures. Nevertheless, the silicate vapor is much less dense than the water vapor generated in a comet impact at the same speed and distance. Figure 2 compares the density of vapor produced in an asteroid and comet impact at 20 km s-1 (Scenarios 3 and 13, respectively) at a point just above the obstacle, 25 km downrange. In the asteroid case, the downrange shockwave actually decreases the atmospheric density to below the ambient density. In the comet case, however, the vapor increases atmospheric density at several points throughout the calculation, even up to ~4 times ambient density. This increased density near the surface would mobilize surface-scouring sands or even gravels.

Of the scenarios that considered target conditions, the thin ice layer on top of basalt (Scenarios 6-8) produced only moderate winds at the surface, even for high-speed impacts. The other scenarios in this category (5: a thick surface ice layer above basalt, 9: a thickly layered terrain of ice and porous basalt) produced surface winds above those categorized as ‘severe’ tornadic winds on Earth (> 60.8 m s-1) (Hyndman and Hyndman,

2016). Only the thickly layered terrain (Scenario 9) produced winds faster than the control case at the same speed.

These results suggest that a layer of ice even 100 m thick is insufficient to induce the intense, tornadic-speed winds required to produce the morphology of the thermally

68 bright wind streaks. Thickly layered terrains or a thick layer of surface ice may result in wind streaks for impacts of at least 20 km s-1, but any wind streaks would not remain identifiable once the thick ice disappeared. Instead, such conditions may be expressed by distal flows as found at high latitudes (e.g., Schultz, 1992; Wrobel et al., 2006) where thick near-surface ice may persist (e.g. Mellon et al., 2008; Smith et al., 2009).

4.3 Effects of Topographic Obstacle

In nearly all cases, the 500 m high topographic obstacle placed at 25 km downrange extended into the gradient in fluid flow between the surface and the expanding vapor and generated turbulence downwind. Vapor often enveloped the surface both uprange and downrange, but the obstacle was necessary to create turbulence. X- velocity data plots through time clearly record the development of vortices (Figures 3 and

4). High-speed vortices generally developed in response to the obstacle, and were particularly strong in cases that generated high amounts of impact-vapor. In order to verify the effects of the obstacle, an ice impactor simulation was run without the obstacle.

In this case, high-speed vapor reached the surface, but vortices did not develop.

Analysis of the velocity data images, in conjunction with the tracer velocity data, revealed that vortex development was the dominant process downrange after the initial passage of the vapor plume for nearly all high-vapor-producing scenarios. Reverse winds developed later from ice impactors striking at 12 km s-1 and 20 km s-1 (Scenarios 12 and

13, respectively) and clearly dominate for the 12 km s-1 case. At 20 km s-1, vortices develop first, then spread downrange, and may combine with reverse winds at late times.

At such times, these reverse winds would have random turbulence and would be unlikely to overpower the vortex-generated winds.

4.4 Effects of a Distant Obstacle

Three scenarios tested an obstacle placed a more realistic 60 km from the impact point: a comet impacting at 12 km s-1, a comet impacting at 20 km s-1, and an asteroid impacting into 500 m of ice overlaying basalt at 12 km s-1. In order to track the effects of the obstacle, these models ran to a simulation time of two minutes (rather than one minute for all other models). Tracers recorded density and velocity data 200 m from the surface at 45 km, 60 km, and 65 km downrange, 15 km uprange, and 15 km orthogonal to the impact trajectory.

Peak wind speeds near the surface exceeded 300 m s-1 for each case; however, sustained winds only matched expectations (outward-blowing) for the fastest comet impactor and the surface ice layer scenarios. For these two cases, sustained outward- blowing winds exceeded 90 m s-1 at distances of 45 km. Strong vortices clearly dominated wind development downrange of the obstacle in every case; no reverse winds were apparent. Additionally, winds moved outward until they reached the obstacle, further demonstrating the importance of the obstacle and the validity of testing the effects of the obstacle at closer range.

4.5 Shockwave

The air-coupled shockwave generated at the first contact between the impactor and the target is apparent most clearly in the tracer data, and in particular, in the pressure data logs. The peaks of all data recorded (pressure, velocity, density, and temperature) almost all correspond to the same time as the pressure peak, within about 0.3 s at 40 km downrange. Because of the consistency of these peaks, it is inferred that these peaks represented a common atmospheric shockwave.

The low density of the atmosphere on Mars is expected to cause a separation between the shockwave and the expanding vapor plume uprange, and therefore wind development (Quintana et al., 2016; Chapter 1). Downrange, the vapor may expand faster than the shockwave such that the shock is coupled to the vapor front. Uprange, however, the vapor plume should create a second peak or spike, particularly in density and velocity. Tracer data uprange (and sometimes orthogonally to the impactor trajectory) did record double peaks in all scenarios (similar to the example shown in

Figure 5). The uprange and orthogonal velocity data presented earlier ignored both peaks in order to present a conservative wind speed estimate. For comparison, Table 7 is the data without the removal of the second peak.

4.6 Summary

The models in this study explored the effects of both surface and impactor properties in order to determine which conditions would lead to a large vapor cloud that expands against the Martian atmosphere and would set up intense, long-lasting winds at the surface. Results from these models indicate that, out of the entire suite of models, those with 1) high-speed (>12 km s-1) impactors, 2) volatile impactors, 3) a thick (>500 m) layer of surface ice, and/or 4) a layered terrain with near-surface water ice meet the required conditions. The models confirm basic expectations that the impact-wind streaks on Mars are likely related to volatiles, whether in the surface or in the impactor.

However, impact speed is also a key factor in the development of impact vaporization and winds at the surface and may narrow realistic streak-forming scenarios.

5. Discussion

5.1 Refining the Analysis

The above results summary ruled out few scenarios from the initial suite of impacts. For example, models that included asteroids traveling at typical impact speeds at Mars, a thin (100 m) layer of ice on the surface, and a wet tuff layer at the surface cannot produce both high amounts of vaporization and fast-moving, outward-blowing surface winds. The list of sufficient models can be further reduced by considering their feasibility. Few areas on Mars have greater than 500 m of ice exposed at the surface as in the models. Of these areas, few exist in concurrent locations as any impact-wind streak craters (Quintana and Schultz, 2016), although both Pál and Hale occur near the boundary of mantled terrains related to maximum polar advance during obliquity cycles.

Furthermore, while the hot impact vapor might vaporize a portion of the surface ice as it expands, the vapor-driven winds would have to somehow scour the surface below the ice.

Once the cover ice is removed, however, evidence of strong winds would be missing or very faint. Such a scenario is inconsistent with observed occurrences on Mars.

The surface-ice cases were intended to be bounding scenarios for testing surface properties: if an impact into a thick layer of surface ice could not produce sufficient vaporization and winds, then thinner layers would also not result in the desired conditions.

Because the thick ice layer case did produce the desired result, we also tested a thinner ice case. But this scenario creates problems similar to its thicker ice layer counterpart: the ice would have to be removed in order to create the observed thermal contrast in the underlying surface. Nevertheless, the results are consistent with unusually long, decoupled run-out ejecta flows for certain craters at high latitudes (e.g., Schultz, 1992;

Wrobel et al., 2006).

Finally, while the control cases (those impacts of a dunite projectile into a solid basalt crust) developed the fastest wind speeds, the associated vapor density was lower than that of the ambient atmosphere. Density data plots indicated that the higher density associated with greater amounts of vapor from target or impactor volatilization may result in the ability to mobilize more significant surface material (Iversen et al., 1976).

Furthermore, asteroid impactors with speeds much greater than 20 km s-1 on Mars are highly unlikely (Steel, 1998; Ma et al., 2002). Thus, we can effectively rule out asteroid impactors traveling at 20-40 km s-1. Table 8 lists each scenario tested (as in Table 4) and the criteria needed to form impact-wind streaks. Matching criteria are indicated with a checkmark such that scenarios with more checkmarks are more likely to form impact- wind streaks.

5.2 Model Limitations

The specific scenarios tested in this study were limited by the time it took to run each model. Models tested specific contributions of different variable properties where appropriate, but some interesting aspects necessarily had to be omitted. For example, no models tested different target porosities explicitly, and only test one porosity value for ice impactors could be tested. Similarly, impact angle tests were limited, but introducing more angles would primarily result in more asymmetries until impact angles fall below about 40°. This work could be expanded through further 3D models in these areas.

Vaporization is a necessary aspect of the impact-wind streak model explained in

Schultz and Quintana (2017) and Chapter 1. The melting and vaporization produced in oblique impacts is expected to increase from laboratory experimental results (Schultz,

1996; Schultz and Eberhardy, 2015), yet has been shown instead to decrease in models

(e.g. Pierazzo and Melosh, 2000; Ivanov and Artemieva, 2002; Pierazzo et al., 2005;

Collins et al., 2005). The model here used peak pressure to determine amounts of melting and vaporization, but peak pressure decreases with impact angle (θ) as sin2θ, the vertical component of the pressure (Schultz, 1996). Thus, the sin2θ dependence of peak pressure on impact angle likely causes an artificial decrease in vaporization. The increase in real vaporization for oblique impacts in the laboratory and the expected increase in vaporization for planetary scale impacts is due to shear and frictional heating. Such forces are not perfectly incorporated in the ANEOS package, but the final release-state temperature method (Quintana et al., 2015a) and the density method presented here do appear to better estimate vapor determination.

6. Conclusion

Thermally bright impact-wind streaks on Mars are a result of the impact process itself, but their paucity across the Martian surface requires additional explanation.

Previous contributions concluded that these wind streaks most likely formed by vapor expansion and atmospheric coupling. As the impact vapor displaces the atmosphere of

Mars, it creates fast-moving winds that may reach the surface if they encounter a sufficiently high-relief obstacle. Obstacles result in the development of horseshoe vortices and cause intense scouring and surface modification, which is expressed as bright streaks in nighttime thermal infrared images. While impact-wind streak craters on

Mars are generally well preserved (Amazonian to late Hesperian in age), they are also a subset of this crater population. Hence, some unique process must be responsible. The present contribution considers different scenarios that might lead to high vaporization and

74 fast-moving wind development through the study of a suite of 3D models that consider both target and impactor properties. The following conclusions can be made:

1) High-speed impactors (>12 km s-1) generate enhanced vaporization compared to

typical impact speeds on Mars (6-12 km s-1). Such speeds most likely occur for

cometary impacts and are unlikely for asteroid impacts.

2) Volatiles (ice at or beneath the surface) substantially contribute to impact

vaporization. Layered terrains similar to those described in the model (500 m of

ice overlain by 500 m of porous basalt) may produce the vaporization and winds

necessary to generate impact-wind streaks. Thick surface ice layers lead to fast,

sustained wind speeds on the surface, but surface effects would not be preserved

once the ice disappeared.

3) Wet tuff is unable to produce the vaporization and the wind speeds expected to be

responsible for impact-wind streaks on Mars.

4) Volatile-rich impactors (especially those impacting at the expected high speeds)

generate substantial vapor and peak wind speeds. Peak wind speeds for all ice

impactors traveling above 12 km s-1 exceed those in terrestrial tornadoes.

5) Thin ice layered models may simulate high-latitude regions. In such cases, vapor-

wind effects would be expressed differently, e.g., long run-out ejecta flows

(Wrobel et al., 2006).

6) A topographic obstacle is required to initiate turbulence (vortices) downwind and

is necessary to the formation of the brightest thermal wind streaks on Mars.

7) Separation between the first-contact shockwave and the expanding vapor plume

for oblique impacts is recognized uprange and orthogonal to the impact but is

overwhelmed by the motion of the plume downrange.

Acknowledgements

We wish to thank David Crawford for his assistance in running the CTH models described here, as well as his helpful conversations about this project, including melt and vapor determination, resolution, AMR, and model setup. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under

Grant No. (DGE-1058262), the Mars Fundamental Research Program Grant No.

(NNX13AG43G), and the NASA Rhode Island Space Grant Consortium Graduate

Research Fellowship. Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation, NASA, or the NASA Rhode Island Space Grant

Consortium.

References

Amsden, A.A., Ruppel, H.M., Hire, C.W., 1980. SALE a simplified ALE computer program for fluid flow at all speeds, Los Alamos Scientific Lab (No. LA- 8095:101p). Los Alamos National Laboratories, Los Alamos, NM.

Barr, A.C., Citron, R.I., 2011. Scaling of melt production in hypervelocity impacts from high-resolution numerical simulations. Icarus 211, 913–916. doi:10.1016/j.icarus.2010.10.022.

Chase, M.W., Jr, Davies, C.A., Downey, J.R., Jr., Frurip, D.J., McDonald, R.A., Syverud, A.N., 1995. NIST-JANAF Thermochemical Tables, 3rd ed. Journal of Physical and Chemical Reference Data 3.

Collins, G.S., Melosh, H.J., 2014. Improvements to ANEOS for Multiple Phase Transitions. Lunar and Planetary Science Conference 45, Abstract 2664.

Collins, G.S., Melosh, H.J., Ivanov, B.A., 2004. Modeling damage and deformation in impact simulations. Meteoritics & Planetary Science 39, 217–231. doi:10.1111/j.1945-5100.2004.tb00337.x.

Collins, G.S., Melosh, H.J., Marcus, R.A., 2005. Earth Impact Effects Program: A Web- based computer program for calculating the regional environmental consequences of a meteoroid impact on Earth. Meteoritics & Planetary Science 40, 817–840. doi:10.1111/j.1945-5100.2005.tb00157.x.

Crawford, D., 1999. Adaptive Mesh Refinement in CTH (No. SAND99–1118C). Sandia National Laboratories, Albuquerque, NM.

Crawford, D.A., Schultz, P.H., 2013. A model of localized shear heating with implications for the morphology and paleomagnetism of complex craters, in: Large Meteorite Impacts and Planetary Evolution. Large Meteorite Impacts and Planetary Evolution, Sudbury, Canada, Abstract 3047.

Elbeshausen, D., Wünnemann, K., 2011. iSALE-3D: A three-dimensional, multi-material, multi-rheology hydrocode and its applications to large-scale geodynamic processes, in: Schäfer, F. (Ed.) ; Hiermaier, S. (Ed.): Proceedings of 11th Hypervelocity Impact Symposium (HVIS), Fraunhofer Verlag. Freiburg, Germany.

Gilbert, G.K., 1893. The Moon’s Face: A Study of the Origin of Its Features. Philosophical Society of Washington.

Greeley, R., Iversen, J.D., 1985. Wind as a geological process on Earth, Mars, Venus and Titan. Cambridge University Press, New York.

Greeley, R., Iversen, J.D., Pollack, J.B., Udovich, N., White, B., 1974. Wind Tunnel Studies of Martian Aeolian Processes. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences 341, 331–360. doi:10.1098/rspa.1974.0191.

Greeley, R., Leach, R., White, B., Iversen, J., Pollack, J., 1980. Threshold windspeeds for sand on Mars: Wind tunnel simulations. Geophysical Research Letters 7, 121–124. doi:10.1029/GL007i002p00121.

Greeley, R., Udovich, N., Iversen, J.D., White, B., Pollack, J.B., 1974. Wind tunnel simulations of light and dark streaks on Mars. Science 183, 847–849. doi:10.1126/science.183.4127.847.

Hyndman, D., Hyndman, D., 2016. Natural Hazards and Disasters. Cengage Learning.

Ivanov, B.A., Artemieva, N.A., 2002. Numerical modeling of the formation of large impact craters. Geological Society of America Special Papers 356, 619–630. doi:10.1130/0-8137-2356-6.619.

Iversen, J.D., Greeley, R., Pollack, J.B., 1976. Windblown dust on earth, Mars and Venus. Journal of the Atmospheric Sciences 33, 2425–2429. doi:10.1175/1520- 0469(1976)033<2425:WDOEMA>2.0.CO;2.

Kraus, R.G., Senft, L.E., Stewart, S.T., 2011. Impacts onto H2O ice: Scaling laws for melting, vaporization, excavation, and final crater size. Icarus 214, 724–738. doi:10.1016/j.icarus.2011.05.016.

Ma, Y., Williams, I.P., Ip, W.H., Chen, W., 2002. The velocity distribution of periodic comets and the meteor shower on Mars. Astronomy & Astrophysics 394, 311–316. doi:10.1051/0004-6361:20021099.

McGlaun, J.M., Thompson, S.L., Elrick, M.G., 1990. CTH: A three-dimensional shock wave physics code. International Journal of Impact Engineering 10, 351–360. doi:10.1016/0734-743X(90)90071-3.

Mellon, M.T., Boynton, W.V., Feldman, W.C., Arvidson, R.E., Titus, T.N., Bandfield, J.L., Putzig, N.E., Sizemore, H.G., 2008. A prelanding assessment of the ice table depth and ground ice characteristics in Martian permafrost at the Phoenix landing site. Journal of Geophysical Research: Planets 113, E00A25. doi:10.1029/2007JE003067.

Melosh, H.J., 2007. A hydrocode equation of state for SiO2. Meteoritics & Planetary Science 42, 2079–2098. doi:10.1111/j.1945-5100.2007.tb01009.x.

Pelkey, S.M., Jakosky, B.M., Mellon, M.T., 2001. Thermal inertia of crater-related wind streaks on Mars. Journal of Geophysical Research: Planets 106, 23909–23920. doi:10.1029/2000JE001433.

Pierazzo, E., Artemieva, N.A., Ivanov, B.A., 2005. Starting conditions for hydrothermal systems underneath Martian craters: Hydrocode modeling. Geological Society of America Special Papers 384, 443–457. doi:10.1130/0-8137-2384-1.443.

Pierazzo, E., Melosh, H.J., 1999. Hydrocode modeling of Chicxulub as an oblique impact event. Earth and Planetary Science Letters 165, 163–176. doi:10.1016/S0012- 821X(98)00263-5.

Pierazzo, E., Melosh, H.J., 2000. Melt Production in Oblique Impacts. Icarus 145, 252– 261. doi:10.1006/icar.1999.6332.

Pierazzo, E., Vickery, A.M., Melosh, H.J., 1997. A Reevaluation of Impact Melt Production. Icarus 127, 408–423. doi:10.1006/icar.1997.5713.

Quintana, S.N., Crawford, D.A., Schultz, P.H., 2013. Verification of Impact Melt and Vapor Determination Methods in CTH. Presented at the Lunar and Planetary Science Conference 44, Abstract 1733.

Quintana, S.N., Crawford, D.A., Schultz, P.H., 2015a. Analysis of Impact Melt and Vapor Production in CTH for Planetary Applications. Procedia Engineering, Proceedings of the 2015 Hypervelocity Impact Symposium (HVIS 2015) 103, 499–506. doi:10.1016/j.proeng.2015.04.065.

Quintana, S.N., Schultz, P.H., 2014. The Formation of Crater-Related Blast Wind Streaks on Mars. Presented at the Lunar and Planetary Science Conference 45, Abstract 1971.

Quintana, S.N., Schultz, P.H., 2016. A Global Distribution of Impact-Wind Streak Craters on Mars. Presented at the Lunar and Planetary Science Conference 47, Abstract 1548.

Quintana, S.N., Schultz, P.H., Crawford, D.A., 2015b. Target Strength as an Important Consideration for Low-Speed Impacts. Presented at the Lunar and Planetary Science Conference 46, Abstract 2727.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2015c. Experimental Results Supporting an Impact-Related Blast Wind Formation Mechanism for Some Wind Streaks on Mars. Presented at the Lunar and Planetary Science Conference 46, Abstract 2469.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2016. New Experiments in Martian Impact Vapor-Induced Wind Streak Analysis. Presented at the Lunar and Planetary Science Conference 47, Abstract 1553.

Richardson, J.E., Melosh, H.J., Lisse, C.M., Carcich, B., 2007. A ballistics analysis of the Deep Impact ejecta plume: Determining Comet Tempel 1’s gravity, mass, and density. Icarus, Deep Impact at Comet Tempel 1 191, 176–209. doi:10.1016/j.icarus.2007.08.033.

Schultz, P.H., 1992. Atmospheric effects on ejecta emplacement. Journal of Geophysical Research: Planets 97, 11623–11662. doi:10.1029/92JE00613.

Schultz, P.H., 1996. Effect of impact angle on vaporization. Journal of Geophysical Research: Planets 101, 21117–21136. doi:10.1029/96JE02266.

Schultz, P.H., Eberhardy, C.A., 2015. Spectral probing of impact-generated vapor in laboratory experiments. Icarus 248, 448–462. doi:10.1016/j.icarus.2014.10.041.

Schultz, P.H., Eberhardy, C.A., Ernst, C.M., A’Hearn, M.F., Sunshine, J.M., Lisse, C.M., 2007. The Deep Impact oblique impact cratering experiment. Icarus 191, 84–122. doi:10.1016/j.icarus.2007.06.031.

Schultz, P.H., Quintana, S., 2013. Impact Blast Wind Scouring on Mars. Lunar and Planetary Science Conference 44, Abstract 2697.

Schultz, P.H., Quintana, S.N., 2017. Impact-generated winds on Mars. Icarus 292, 86– 101. doi:10.1016/j.icarus.2017.03.029.

Schultz, P.H., Wrobel, K.E., 2012. The oblique impact Hale and its consequences on Mars. Journal of Geophysical Research 117, E04001. doi:10.1029/2011JE003843.

Shoemaker, E.M., 1962. Exploration of the Moon’s Surface. American Scientist 50, 99– 130.

Sierks, H., Barbieri, C., Lamy, P.L., Rodrigo, R., Koschny, D., Rickman, H., Keller, H.U., Agarwal, J., A’hearn, M.F., Angrilli, F., others, 2015. On the nucleus structure and activity of comet 67P/Churyumov-Gerasimenko. Science 347, aaa1044. doi:10.1126/science.aaa1044.

Smith, P.H., Tamppari, L.K., Arvidson, R.E., Bass, D., Blaney, D., Boynton, W.V., Carswell, A., Catling, D.C., Clark, B.C., Duck, T., DeJong, E., Fisher, D., Goetz, W., Gunnlaugsson, H.P., Hecht, M.H., Hipkin, V., Hoffman, J., Hviid, S.F., Keller, H.U., Kounaves, S.P., Lange, C.F., Lemmon, M.T., Madsen, M.B., Markiewicz, W.J., Marshall, J., McKay, C.P., Mellon, M.T., Ming, D.W., Morris, R.V., Pike, W.T., Renno, N., Staufer, U., Stoker, C., Taylor, P., Whiteway, J.A., Zent, A.P., 2009. H2O at the Phoenix Landing Site. Science 325, 58–61. doi:10.1126/science.1172339.

Steel, D., 1998. Distributions and moments of asteroid and comet impact speeds upon the Earth and Mars. Planetary and Space Science 46, 473–478. doi:10.1016/S0032- 0633(97)00232-8.

Sugita, S., Schultz, P.H., 1999. Spectroscopic characterization of hypervelocity jetting: Comparison with a standard theory. Journal of Geophysical Research: Planets 104, 30825–30845. doi:10.1029/1999JE001061.

Sugita, S., Schultz, P.H., Adams, M.A., 1998. Spectroscopic measurements of vapor clouds due to oblique impacts. Journal of Geophysical Research: Planets 103, 19427–19441. doi:10.1029/98JE02026.

Thompson, S.L., 1990. ANEOS Analytic Equations of State for Shock Physics Codes Input Manual (No. SAND89- 2951 • UC–404). Sandia National Laboratories, Albuquerque, NM.

Thompson, S.L., Lauson, H.S., 1972. Improvements in the CHARTD Radiation– Hydrodynamics Code III: Revised Analytic Equation of State (No. SC–RR– 710714). Sandia National Laboratories, Albuquerque, NM. 81

Tables

Table 1 - List of materials used in the models and their equations of state.

Equation Material Model Purpose Library Name of State Dunite Impactor ANEOS DUNITE_MOL Basalt Target ANEOS QUARTZ_14PT_MOL Ice Impactor/Target ANEOS WATER-ICE_MOL Tuff Target SESAME WET_TUFF CO2 Atmosphere Ideal Gas CO2

Table 2 - ANEOS parameters for appropriate materials (Table 1)

Valueb for:

Variablea Description Units Dunitec Basaltd Icee Number of elements N - 3.00 2.00 2.00 in material

Switch for EoS type: 2 = Gas with ionization, 3= Solid- Type liquid-gas without - 4.00 4.00 4.00 ionization, 4= Solid- liquid-gas with ionization

ρ0 Reference density g/cm3 3.32 2.65 1.11

T0 Reference temperature eV 0.025 0.025 0.0184f

P0 Reference pressure dyn/cm2 0.00 0.00 0.00

Reference sound speed in linear shock- S0 cm/s 6.60E+05 3.68E+05 1.80E+05 particle velocity relationship

Reference Gruenisen Γ - 0.82 0.62 0.58 coefficient

Reference Debye temperature, >0 = high temperature θ approximation form, eV 0.06 0.06 -0.05 <0 = complete Debye functions for solid model

Constant in linear Hugoniot shock- S1 - 0.86 2.12 1.30 particle velocity relation Note: Table 2 continued on next page.

Table 2 continued

Valueb for:

Variablea Description Units Dunitec Variablea Icee

3x the limiting value of the Gruenisen 3*C24 - 2.00 2.00 2.00 coefficient for large compressions

Zero temperature Es erg/g 1.30E+11 1.24E+11 4.70E+10 separation energy

Tm Melting temperature eV 0.19 0.17 0.02

Parameter for low- C53 density critical point - 0.00 6.00E+11 0.00 modification

Parameter for low- C54 density critical point - 0.00 0.80 0.00 modification

Thermal conductivity ergs/(cm s H0 coefficient (not 0.00 0.00 0.00 eV) included)

Temperature dependence of C41 thermal conduction - 0.00 0.00 0.00 coefficient (not included)

Lowest allowed solid ρmin g/cm3 0.00 0.00 0.00 density

Density at beginning D1 of solid-solid phase g/cm3 4.65 3.50 0.00 transition Note: Table 2 continued on next page.

Table 2 continued

Valueb for:

Variablea Description Units Dunitec Variablea Icee

Density at end of D2 solid-solid phase g/cm3 4.49 4.30 0.00 transition

Pressure at center of D3 phase transition at dyn/cm2 6.60E+11 2.10E+11 0.00 zero T

dP/dη at high- D4 pressure phase - 3.50E+12 1.80E+12 0.00 transition

D2P/dη2 at high- D5 pressure phase - 1.30E+13 6.00E+12 0.00 transition

Heat of fusion for Hf - 0.00 0.00 1.95E+09 melt transition

Ratio of liquid to ρliq solid density at melt - 0.00 0.00 -1.00 point

Upper bound in cold Up compression relation - 0.00 0.00 0.00 for expanded states

Lower bound in cold L0 compression relation - 0.00 0.00 0.00 for expanded states

Parameter of liquid α - 0.00 0.00 0.00 EoS correction

Parameter of liquid β - 0.00 0.00 0.00 EoS correction Note: Table 2 continued on next page.

Table 2 continued

Valueb for:

Variablea Description Units Dunitec Variablea Icee

Parameter of liquid γ - 0.00 0.00 0.00 EoS correction

Interpolation parameter in C60 - 0.00 0.00 0.40 Gruneisen coefficient model

Interpolation parameter in C61 - 0.00 0.00 0.00 Gruneisen coefficient model

Interpolation C62 parameter in free - 0.00 0.50 0.32 energy equation

Flag for ionization Flag model, 0=Saha, - 1.00 0.00 1.00 1=Thomas-Fermi

Energy shift for Eshift reactive chemistry - 0.00 0.00 0.00 model

Entropy shift for Sshift reactive chemistry - 0.00 0.00 0.00 model

Number of atoms in Atoms - 0.00 2.00 0.00 molecular clusters

Molecular cluster Ebind - 8.00 5.00 3.20 binding energy

Number of rotational n_rot - 0.00 2.00 0.00 degrees of freedom Note: Table 2 continued on next page.

Table 2 continued

Valueb for:

Variablea Description Units Dunitec Variablea Icee

Length of molecular Rbond - 1.00 0.00 1.00 bond

Number of n_vib vibrational degrees of - 1.25 1.00 2.40 freedom

Vibrational Debye Theta_vib - 0.00 0.17 0.00 temperature

Flag for Mie potential LJ_flag - 0.00 1.00 0.00 or Morse potential

Power in Mie a_exp - 0.00 1.70 0.00 potential

Atomic number of Z(1) - 0.00 0.00 0.00 element 1

Atomic fraction of COT(1) - 0.00 0.00 0.00 element 1

Atomic number of Z(2) - 0.00 0.00 0.00 element 2

Atomic fraction of COT(2) - 0.00 0.00 0.00 element 2 a Variable names and descriptions (Thompson and Lauson, 1972; Thompson, 1990; Pierazzo et al., 1997; Melosh, 2007) b Note that these values are library definitions for EoS parameters for CTH version 11.2 unless otherwise noted. In many cases, zeros indicate that the parameter is unused, rather than being a value of zero. c Pierazzo, April 2003 (Note in EoS document) d Melosh (2007) e Pierazzo, April 2003 (Note in EoS document) f Set within the input deck to below the melting point of ice.

Table 3 - Vaporization criteria used in the vaporization subroutine

Peak Temperature Densitya Material Pressure (GPa) (K) (g/cm3) Dunite 186 5763.9552 4.0E-04 Basalt 172 9961.3028 4.0E-04 Ice 4.5 328.5234 4.0E-04 372.73654b 4.0E-04 Tuffc 172 9961.3028 4.0E-04 a Three orders of magnitude lower than the lowest density material (ice) b JANAF (Chase et al., 1995) temperature for vaporization of ice c Because tuff is volcanic ash, the parameters were taken to be the same as basalt.

Speed

6 km/s

40 km/s

20 km/s

12 km/s

20 km/s

12 km/s

20 km/s

12 km/s

40 km/s

20 km/s

12 km/s

Impactor

Type

Impactor

Porous ice

Solid dunite

layer

basalt

500 m

100 m

500 m

(1) 500 m ice (1) 500 m

Thickness of

(2) 500 m porous(2) 500 m

tuff tuff

basalt

layer

ice layer

ice

ice layer

overlain by (1) solidiceoverlain by (1)

Solid basalt

Solid

Solid basalt

Target Type

layer and (2) porousand (2) layer basalt

Solid basalt

Solid basalt overlain by thin, porous

Solid basalt overlain by thick, porous

Solid basalt overlain by thick wet

Solid basalt overlain by a thin, porous a thin, porous Solid basalt overlain by

Control

Property

Type Tested Type

Target Property

Control, Impactor

Impactor PropertyImpactor

Primary Scenario

List of scenarios (models) tested. (models) of scenarios List

Scenario Table 4 Table

Table 5 - Mass of impact-vaporized material normalized by projectile mass for the different vaporization determination methods

Vaporization Determination Methoda Peak Scenario Pressure Temperature Temperatureb Density 1 0.00 0.00 - 1.32 2 0.00 0.00 - 2.64 3 1.26 5.E-04 - 6.09 4 0.28 8.E-04 - 5.21 5 10.33 1.E-13 - 15.52 6 4.01 0.28 0.13 4.66 7 0.64 0.05 0.05 2.08 8 0.99 0.13 0.11 3.06 9 2.07 0.40 0.33 5.78 10 4.68 1.71 0.49 3.45 11 0.00 4.E-09 - 3.26 12 1.15 5.E-04 - 5.66 13 0.88 0.33 0.31 3.80 14 0.84 0.36 0.34 8.98 15 0.80 0.41 0.39 7.20 16 2.58 0.31 0.29 15.01 a See Table 3 for the vaporization criteria for each determination method and the text for more details b JANAF (Chase et al., 1995) temperature for vaporization of ice

110

160

150

180

190

540

15 km

2.3 km

Uprange

270

350

100 m

15 km

280

340

140

150

320

100

660

1230

15 km

2.3 km

Orthogonal

(Scenario 1)

100 m

15 km

210

110

410

230

560

230

530

120

500

160

100

100 m

40 km

Control 40 km/s (Scenario 5)

Control 20 km/s (Scenario 3)

Control 12 km/s (Scenario 2)

Control 06 km/s

350

780

870

410

130

260

150

360

140

2350

3790

1290

2650

5 km

30 km

Thick Ice Surf (20 km/s) (Scenario6) (20 Thick Icekm/s) Surf

Thin Ice Surface 06 km/s (Scenario 7) Thin Ice Surface 06 km/s (Scenario

Downrange

720

240

820

920

240

110

130

400

130

270

2030

2370

30 km

2.3 km

750

430

160

840

160

490

110

120

130

100 m

30 km

Altitude

Scenario

wind speed

Wind Speed Results tracers Wind for select

Sensor Location

Table 6 continued on next page on next 6 continued Table

Distance Impact from

Note:

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Table 6 Table

110

130

300

230

30 km

2.3 km

Uprange

110

210

100 m

30 km

230

250

260

300

160

670

310

100

590

290

5 km

30 km

Orthogonal

30 km

2.3 km

300

320

110

330

530

170

110

1190

100 m

30 km

Tuff 20 km/s12) (Scenario

Tuff 12 km/s11) (Scenario

Comet 12 km/s (Scenario 13) Comet (Scenario 12 km/s

Layers (20 km/s) (Scenario 10) Layers (20 (Scenario km/s)

250

530

100

710

110

130

340

110

210

420

230

1530

1410

1230

2530

1210

5 km

30 km

Thin Ice Surface 20 km/s (Scenario 9) Thin Ice Surface 20 km/s (Scenario

Thin Ice Surface 12 km/s (Scenario 8) Thin Ice Surface 12 km/s (Scenario

Downrange

150

490

660

790

120

140

400

250

1230

1260

2420

1060

30 km

2.3 km

180

100

390

200

510

130

330

100

160

100 m

30 km

speed

Altitude

Scenario

speed

wind speed

Sensor Location

inward wind speed inward

outward wind speed outward

Distance Impact from

Note: Table 6 continued on next page on next 6 continued Note: Table

Average wind speed Average wind

Max inward wind Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max Table 6 continuedTable

130

110

30 km

2.3 km

Uprange

100

100 m

30 km

180

340

250

210

5 km

30 km

Orthogonal

30 km

2.3 km

300

380

140

410

100 m

30 km

Comet 40 km/s (Scenario 16) Comet (Scenario 40 km/s

Comet 20 km/s (Scenario 14) Comet (Scenario 20 km/s

840

650

160

2400

1390

5 km

30 km

Downrange

650

120

380

810

2360

30 km

2.3 km

280

120

140

100 m

30 km

Altitude

Scenario

Sensor Location

Distance Impact from

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward Table 6 continuedTable

110

290

240

220

15 km

2.3 km

Uprange

270

160

100 m

15 km

km/s (Scenario 12) km/s (Scenario

250

230

670

210

380

15 km

2.3 km

Tuff 20

Comet 20 km/s (Scenario 14) Comet (Scenario 20 km/s

Comet 12 km/s (Scenario 13) Comet (Scenario 12 km/s

Orthogonal

100 m

15 km

130

300

260

220

1080

15 km

2.3 km

Uprange

100

110

230

350

100 m

15 km

120

340

140

310

570

360

15 km

2.3 km

Layers (Scenario 10)

Comet 40 km/s (Scenario 16) Comet (Scenario 40 km/s

Control 20 km/s (Scenario 3)

Orthogonal

100 m

15 km

Altitude

Scenario

Orthogonal and uprange wind speed data, included second peak (see text), for select (seesecond scenarios Orthogonal uprange peak text), wind speed included and data,

Sensor Location

Distance Impact from

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward

Average wind speed Average wind

Max inward wind speed Max inward

Max outward wind speed Max outward Table





obstacle

set up at

Pressure

differential differential





Vapor

surface

of obstacle

interaction

downrange downrange





vapor

ambient)

surface

High density

N/A

Water

Silicate

2) Water

Primary

1) Water; 1) Water;

2) Silicate

1) Silicate;

vapor type

wind streaks

)





Vapor Vapor

(>0.8

generated





Winds

directed

outward





high

wind wind

speeds

surface

Sustained





wind wind

High

speeds

surface

(>70 m/s)

Scenario Summary: Likelihood to produce impact Likelihood to produce Scenario Summary:

Scenario Table Table

Figure Captions

Figure 1 - Diagram of tracer setup in the CTH model. Ten tracers fill the distance between 100 m and 5 km of the target surface, followed by one tracer at 10 km altitude and another at 15 km altitude. Note that the tracers at 0 along the x-axis are actually 15 km away from the impact point in the y-direction.

Figure 2 - Comparison of the density produced in a) an asteroid impact and b) a comet impact at 20 km s-1, taken from a tracer positioned just above the obstacle at 25 km downrange. High-density ejecta (arrows) overpower the signal in (a). The inset in (a) removes this material to show data for times <45 s, with the axes being the same as in (a).

The asteroid impact reduces atmospheric density to below that of the ambient Martian atmosphere (dotted line). Alternatively, the comet impactor generates a vapor plume that increases the atmospheric density.

Figure 3 - The effects of an obstacle placed 25 km downrange of the impact point are documented in this x-velocity data plot through time. Impact direction is from the right.

The colors correspond to the velocity scale on the right such that warmer colors represent motion (of vapor) to the right of the images and cooler colors represent motion to the left.

The apparent line (change in color) directly above the crater is due to colormap scale applied to the data. Cool colors depict downrange-moving material (vapor), while warm colors depict uprange-moving vapor. Curved arrows indicate vortices that form downrange of the obstacle. This figure was taken from Scenario 5 described in the text (a dunite impactor striking a basaltic target overlain by 500 m of ice at 45° and 12 km s-1).

Images denote 2D slices of the 3D model.

Figure 4 - Same as Figure 3 except for Scenario 13 described in the text (an ice impactor striking the surface of Mars at 45° and traveling 20 km s-1).

Figure 5 – Plots of (a) pressure and (b) x-velocity reveal double peaks in the tracer data.

The initial peak corresponds to the atmospheric shockwave (see text). The second peak is only found in uprange and some orthogonal tracer data. Because the impactor is modeled at the surface and does not travel through the atmosphere, the double peak is not due to any wake effects. Instead, it represents the front of the vapor plume expanding around the impact point. These plots are examples taken from Scenario 13 described in the text (an ice impactor striking the surface of Mars at 45° and traveling 20 km/s). Both

(a) pressure and (b) velocity plots are from Tracer 53, 2.3 km in altitude 15 km uprange of the impact point.

Figures

Figure 1

Figure 2

100

Figure 3

101

Figure 4

102

Figure 5

103

CHAPTER THREE:

Distribution, Observations, and Implications of Impact-Wind Streaks on Mars

Stephanie N. Quintana1

and

Peter H. Schultz1

1Department of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI 02912

104

Abstract

Impact-wind streaks are a unique subset of both crater-related wind streaks and craters that have a radial bright and dark pattern in Thermal Emission Imaging System

(THEMIS) nighttime infrared images. Impact-wind streaks are thermally bright features that extend from preexisting craters and ridges radial to a parent impact crater. They are interpreted to form from intense winds and surface scouring as part of the impact process.

Previous studies proposed a vapor-driven wind formation process and concluded that certain target or impactor properties could generate the vapor needed in order to drive surface-scouring winds. A global distribution study is presented here in order to assess constraints on the formation of impact-generated wind streaks based on their global occurrence and expression. This study found only 12 impact-wind streak craters on Mars.

A case study of seven such craters (Xainza, Pál, Mojave, Kotka, Prao, and two unnamed craters near Isidis Planitia) revealed that the craters exist in varying terrain, topographic settings, and geologic units. No clear relationship between target settings is apparent.

Thus, we conclude that the most likely explanation for impact-wind streaks is formation though impactor-derived (cometary) vapor-driven winds. This result has implications for the estimated comet impact flux on Mars.

105

1. Introduction

Previous studies (Schultz and Wrobel, 2012; Schultz and Quintana, 2013;

Quintana and Schultz, 2014; Quintana et al., 2015; Quintana and Schultz, 2016; Quintana et al., 2016; Quintana and Schultz, 2017; Schultz and Quintana, 2017) recognized peculiar features on Mars that are reminiscent of wind streaks, which radiate away from a central impact crater. These features are distinct from other wind streaks on the planet in that they are 1) radial to an impact crater, 2) appear bright in nighttime infrared images of the Thermal Emission Imaging System (THEMIS), and 3) are long-lasting. Typical wind streaks extending from craters respond to global wind patterns and extend in the same direction over a broad region. They also tend to appear dark in the THEMIS nighttime infrared images because they are composed of finer material than their surroundings (e.g.,

Pelkey et al., 2001). Other wind streaks on Mars (particularly the low albedo streaks in visible wavelengths) are variable on human timescales (Thomas et al., 1981). They may change length and orientation, or even disappear altogether.

The features discussed in the current study are interpreted to be a subset of crater- related wind streaks that form as a part of the impact cratering process. Because the streaks are concomitant with the impact process, we term them impact-wind streaks

(Quintana et al., 2016; Quintana and Schultz, 2017; Schultz and Quintana, 2017). As will be shown, they are distinct from crater rays or secondary cratering and manifest as thermally bright streaks within an overall environment of radial and alternating bight and dark diffuse patches, as viewed with THEMIS nighttime infrared images. The streaks

(typically in pairs) extend from preexisting topographic obstacles (crater rims or ridges) and radiate from a large, well-preserved impact crater. Schultz and Wrobel (2012) and

106 later Schultz and Quintana (2017) characterized impact-wind streaks in the context of their environment and proposed potential processes of formation. Quintana et al. (2016,

2015, Chapter 1) and Quintana and Schultz (2014, 2017, Chapter 2) further explored the most likely hypothesis of formation that the thermally bright features are wind streaks formed by impact vapor expansion into the atmosphere of Mars.

Impact-generated wind streaks differ from crater rays and secondary cratering.

Rays and secondary craters are related to ballistic ejecta deposits, whereas impact- generated wind streaks are a manifestation of surface modification from vapor-driven winds. Tornabene et al. (2006) identify several Martian rayed craters (much smaller than the craters in this study) with THEMIS images and describe them as filamentous and thermally dark. Impact-wind streaks, instead, are thermally bright and are only found extending from topographic obstacles such as preexisting craters (abbreviated PE craters) or ridges. Most importantly, Schultz and Quintana (2017) determined that impact wind- related streaks develop before the deposition of ejecta.

The objective of this contribution is to explore constraints on the formation of impact-generated wind streaks based on their global occurrence and expression, including the effects of topography and geologic setting. We first consider the global distribution of impact wind streaks and trends relative to various global datasets. Then location and surface topography are shown to have specific effects on streak expression through case studies of a portion of impact-wind streak craters. Finally, we discuss the implications of impact-wind streak crater locations and the role of impact vaporization.

2. Background

2.1 Thermal Emission Imaging System (THEMIS)

107

Thermal inertia is a useful tool for characterizing wind streaks, as demonstrated in studies using the Thermal Emission Spectrometer (TES) (Pelkey et al., 2001). Because thermal inertia is a measure of the resistance of a material to a change in temperature, such measurements can be used to distinguish between composition and morphology

(grain size or exposed competent substrates) on the surface (Kieffer et al., 1973; Fergason et al., 2006a, 2006b) that otherwise may be indistinguishable at visible wavelengths.

Several researchers (e.g., Wechsler and Glaser, 1965; Fountain and West, 1970; Presley and Christensen, 1997a, 1997b, 1997c, Fergason et al., 2006a, 2006b) established the relationship between particle size and conductivity, which in turn relates surface thermal inertia values to an effective particle size. While thermal inertia cannot uniquely characterize surface particle sizes, generalizations are still useful. Bedrock and gravel, for instance, exhibit a high thermal inertia because they heat (and subsequently cool) slowly, and will therefore appear bright in thermal emission images. Relatively finer grain sizes, in contrast, appear dark because they have a low thermal inertia.

Consequently, thermal inertia measurements can give some insight into the physical nature of that material (Kieffer et al., 1973; Christensen, 1982; Fergason et al., 2006a,

2006b).

The Mars Odyssey spacecraft carried the THEMIS instrument with much higher resolution (100 m/pixel in nighttime temperature data) than the resolution of the TES system (up to 3 km/pixel). Nighttime temperature observations are particularly useful because the physical effects on thermal inertia of the surface are clearest when albedo and slope effects are at a minimum (Kieffer et al., 1973; Kieffer, 1977; Christensen et al.,

2001; Fergason et al., 2006b). Here, we use THEMIS mosaics to locate and characterize

108 impact-wind streak craters and their relation to topography, geologic setting, and crater age.

2.2 Impact-Wind Streaks

Schultz and Wrobel (2012) explored thermally bright (NT-B, for THEMIS nighttime bright) streaks around Hale in a detailed analysis of the surface expressions from the impact and concluded that the streaks resulted from winnowing or scouring by impact-generated winds. The winds, which were attributed to be an atmospheric response to the expanding impact-vapor plume, scoured the surface when they formed vortices on the lee side of PE craters. The streaks not only radiate from Hale but also a region just downrange from the rim due to the downrange motion of the vapor plume.

Moreover, streaks and associated scouring of Hale-facing topography was identified in excess of 1000 km downrange.

A detailed study of Santa Fe Crater investigated the nature and sequence of emplacement of wind streaks extending from PE craters (Schultz and Quintana, 2017).

High-resolution Context Camera (CTX) and High-Resolution Imaging Science

Experiment (HiRISE) images revealed that the streaks are associated with crenulated ridges and yardangs, providing further evidence of wind erosion. Schultz and Quintana

(2017) proposed two scenarios in an atmospheric-wind model to explain the formation of the impact-wind streaks: a) an atmospheric shockwave-driven blast wind, initiated upon contact of the impactor and planet surface, or b) winds driven by impact-vapor expansion and atmospheric coupling. The authors favored scenario (b) based on the sequence of arrival, the pattern of erosion, and extent of the features.

109

Quintana et al. (2015, 2016, Chapter 1) tested the two scenarios with laboratory experiments performed at the NASA Ames Vertical Gun Range (AVGR). Impacts of a

Pyrex projectile into a powdered dolomite target at ~5 km s-1 produced a traceable impact-vapor plume. Tracers and detection devices allowed for observations of wind initiation and development throughout the impact. Results from this study indicated that impact-generated vapor expands separately from the contact-coupled atmospheric shockwave. It is this vapor expanding against the atmosphere, and not the shockwave, that is directly responsible for winds.

In an ensuing study, Quintana and Schultz (2017, Chapter 2) further explored the development of vapor driven winds with computational models. While the experiments demonstrated that impact vapor could drive wind development and be distinguished from an air shock, the modeling study explored the ways in which impact vapor could be generated on Mars. A suite of planetary-scale models with the shock physics code CTH

(McGlaun et al., 1990; Hertel et al., 1993) revealed that certain target properties (volatiles on the surface or subsurface) or impactor properties (speed, composition) could produce favorable conditions for impact wind development.

3. Global Distribution of Impact-Wind Streaks

Only 12 craters exhibit NT-B wind streaks associated with preexisting topography.

Cyclic polar mantling may mask streaks at high latitudes or result in a different expression, e.g., long run-out flows (Schultz, 1992; Wrobel et al., 2006). As a result, areas poleward of 60° north and south are not included in this survey. To avoid confusion, we use the terminology impact-wind streak crater to refer to the 12 craters

110 that exhibit streaks extending from PE craters. Figure 1 is a reference for the terminology used here.

An additional 35 craters exhibit a radial, alternating bright and dark pattern in the

THEMIS nighttime infrared images. For these craters, the pattern is not related to disturbances associated with preexisting relief (e.g., craters). For simplicity, these craters are termed radial thermal streak craters. In some cases, radial thermal streaks indicate ejecta deposits or secondary cratering due to ballistic ejecta large enough to avoid deceleration in the Martian atmosphere. In other cases closer to the parent crater rim, the radial thermal streaks relate to ejecta flows containing contrasting size distributions.

Elsewhere, the radial thermal streaks relate to surface modification by winds. Impact- wind streaks are therefore a subset of the latter case and are characterized as the only radial thermal streak craters to have NT-B streaks extending from preexisting topography.

Figure 2 is a global distribution of radial thermal streak craters, with impact-wind streak craters highlighted in black. The distribution of these craters is somewhat random, with the exception of a clear dearth of craters within the Tharsis Rise, Arabia Terra,

Medusa Fossae, and the high latitudes. For an explanation of these regions, it is helpful to consider the regional conditions. The pattern of erosion in these areas indicates the presence of thick deposits of aeolian dust and ice undergoing different amounts of erosion (Schultz and Lutz, 1988; Tanaka, 1999). Figure 3 maps the same crater distribution over OMEGA dustcover (a, Ody et al., 2012) and TES thermal inertia (b,

Christensen and Moore, 1992; Christensen et al., 2001), respectively. These maps give a measure of surface characteristics and how they relate to radial thermal streak craters.

Areas with few of these craters tend to have low thermal inertia (so-called stealth regions,

111

Muhleman and Butler, 1991; Edgett et al., 1997; Karunatillake et al., 2009) and high dustcover. Such regions are less likely to have the thermal contrast necessary to observe

NT-B streaks. Similarly, craters in high-latitude regions (poleward of 60°) may express vapor-wind effects differently than those at lower latitudes.

Removing the stealth regions and high-latitude areas, it is clear that impact craters with NT-B streaks are found in a variety of different terrains across Mars. Just over half of the population formed in the northern lowlands. One is in young volcanic plains from

Elysium Mons, whereas another is within the Isidis basin (with yet another just exterior to the basin). Furthermore, several are located around Chryse Planitia: one in a chaos region between the massif rings of the Chryse basin; another, within Chryse Planitia; and one, within the Kasei Valles. The last impact-wind streak crater in the northern hemisphere is located in the mantled and etched terrain near 0° latitude. Hence, these impacts clearly do not follow a trend in target type.

Impact-wind streak craters in the southern highlands are equally as varied. Three are within an early volcanic unit within Hesperia Planum, one is within a larger crater

(Huygens), and another is on the edge of the Argyre basin. More generally, the radial thermal streak craters are even more widely distributed across the southern highlands.

Figure 3c maps the distribution of radial thermal streak craters over contours of the highest percentages of stoichiometrically equivalent water mass fraction

(Karunatillake et al., 2014). The near-surface water is based on Mars Odyssey Gamma

Ray Spectrometer (GRS) data. Even though GRS is only sensitive to decimeter depths, the majority of the 12 impact-wind streak craters clearly do not correspond to areas of high water content.

112

4. Case Studies

The following six case studies illustrate the variety of locations and terrains where impact-wind streaks occur on Mars. For additional examples, the reader is directed to detailed accounts of Hale Crater (Schultz and Wrobel, 2012) and Santa Fe Crater (Schultz and Quintana, 2017). Table 1 lists all 12 impact-wind streak craters, their locations, sizes, and information regarding the wind streaks emanating from the PE craters. The case studies presented below are outlined.

Each case study will generally follow the same description sequence. The crater of interest will be introduced, followed by a brief description of the surrounding area and terrain, including geological units (defined by Tanaka et al., 2014), roughness (from the

MOLA roughness RGB map composite by Kreslavsky and Head, 2002), and TES thermal inertia (from Christensen and Moore, 1992; Christensen et al., 2001). TES thermal inertia is considered separately from the THEMIS nighttime infrared because it provides a convenient marker for whether the area would otherwise be a likely candidate for impact-wind streak craters. As noted previously, stealth regions (with very low thermal inertia) are unlikely candidates. This study uses units of ‘tiu’ (e.g., Putzig et al.,

2004) for thermal inertia, rather than the more cumbersome J m-2 K-1 s-1/2. Finally, we provide a general thermal characteristic of the area, from THEMIS nighttime infrared images, followed by more specific details about each case study crater.

4.1 Xainza Crater

Xainza (Figure 4) is a 24 km diameter crater in Meridiani Planum at 0.78° north latitude and 3.94° west longitude. It is located approximately 180 km northeast of the

Mars Exploration Rover (MER) Opportunity landing site. The region is relatively

113 smooth, especially in the south (Kreslavsky and Head, 2002), and is of moderate TES thermal inertia (200-300 tiu, Christensen and Moore, 1992; Christensen et al., 2001).

Xainza is within a Hesperian and Noachian highland mound-forming unit (Schultz and

Lutz, 1988; Tanaka et al., 2014) composed of undifferentiated and friable unconformable sedimentary sequence and impact facies. The terrain is mantled and etched, and engulfs the MER Opportunity Rover site and Miyamoto crater to the south. The smoothest region around Xainza also corresponds with the Hesperian and Noachian highland unit, locally undergoing exhumation (Schultz and Lutz, 1988).

The overall thermal characteristic, based on THEMIS nighttime infrared images of the region is intermediate to poor, meaning that a few small craters have thermally bright rays and structural contrast is apparent through infrared images. The mantled deposits have particularly low thermal contrast. Xainza itself has a well-defined radial bright and dark streaking thermal pattern that extends ~12 crater radii. Additionally,

Xainza has at least seven sources (i.e., PE craters) of NT-B streaks that are most prominent to the north. Topographic views of the area north of Xainza indicate that the region is more eroded such that courser materials may be more readily available. The farthest PE crater with a clear wind streak measures 162 km north-northwest from the center of Xainza, and it has a rim height of ~40 km (as determined in the JMARS software from MOLA topography). The longest streak measures nearly 15 km long and is located ~70 km west-southwest of Xainza. Interestingly, the rim of the PE crater with the longest wind streak is only 11 m high.

4.2 Pál Crater

114

Pál (-31.31°, 108.70°) is a 71 km diameter crater in Hesperia Planum, northeast of the Hellas Basin (Figure 5). The region directly around Pál is smooth, though the region gets moderately rough to the south beyond ~4 crater radii from Pál (Kreslavsky and Head,

2002). TES thermal inertia is low in this region (150-250 tiu, Christensen and Moore,

1992; Christensen et al., 2001).

Pál lies within the Early Hesperian volcanic ridged plains unit interpreted to be related to Tyrrhenus Mons (Tanaka et al., 2014), which is located ~600 km to the north- northwest. The Early Hesperian volcanic unit is defined by planar deposits and may be up to tens of meters thick. Within the impact-wind streak crater population, Pál is one of only four craters to have impacted a volcanic unit. Additionally, Tanaka et al. (2014) mapped an Amazonian and Hesperian impact unit that describes the impact materials

(ejecta, melt deposits) related to Pál. Only four of the 12 impact-wind streak craters were identified as having this Amazonian and Hesperian impact unit (out of >350 craters on

Mars with such a unit).

Tanaka et al. (2014) describe the Early Hesperian volcanic unit as having variable daytime infrared brightness. Similarly, the region has intermediate thermal contrast in nighttime thermal infrared. Most of the area appears relatively bland except for a few small craters that have thermally bright ejecta deposits. Yet, Resen (a ~7 km diameter crater) 200 km north of Pál exhibits spectacularly bright rays in the nighttime infrared images. The thermal character of Pál and its surrounding region is well defined to the north, and particularly to the northeast, where streaks and the light/dark pattern extend out ~10 crater radii. No characteristic light/dark radial pattern or NT-B streaks exist south of Pál.

115

Consequently, the 31 PE craters that have the most distinct bright wind streaks are located north and northeast of the crater where the presence of underlying materials with higher thermal inertia not only can be excavated by small craters but also can be differentially eroded by impact-generated winds. The farthest streak from Pál is located over 380 km away, and the PE crater bearing this streak has a rim height of ~50 m. The longest streak is 58 km long and emanates from a crater almost 300 km from Pál. In this case, the PE crater rim is 120 m high.

4.3 Mojave Crater

Mojave, located at 7.48° latitude and -32.99° longitude, lies between the Tiu

Valles and Chryse Chaos terrains, just north of a high-standing island or plateau north of

Hydraotes Chaos (Figure 6). Mojave is nearly 60 km in diameter and lies at the intersection of three geologic units: the middle Noachian highland unit, which is intersected by the Hesperian transition and the Hesperian transition outflow units.

Mojave is another of the four impact-wind streak craters with a clearly defined

Amazonian and Hesperian impact unit (Tanaka et al., 2014). Additionally, the crater occurs between the massif rings of the Chryse basin. The region has moderate to high

TES thermal inertia values (300-600 tiu in some areas, Christensen et al., 2001;

Christensen and Moore, 1992) and is located in an area that is mostly quite smooth.

South of the crater, the terrain becomes slightly rougher, related to the base of the plateau just south of Mojave (Kreslavsky and Head, 2002).

The THEMIS nighttime infrared thermal contrast in this area is quite good.

Smaller craters display bright halos, indicating an accessibility of bright, coarse materials near the surface. Thermal contrast is also apparent in the valley floors, which are much

116 brighter than the surrounding plateaus and highlands unit. Mojave itself displays a very good thermal signature, with a clear light/dark radial pattern that extends ~8 crater radii.

Mojave has 31 exceptionally bright impact-wind streaks. The farthest streak extends from a crater 200 km northwest of Mojave. MOLA data (through JMARS) did not yield a value for the rim height for this PE crater. The longest streak measures 22 km long, and extends south-southeast of Mojave from a crater ~120 km with a 33 m rim height.

The relationship between the wind streaks with the surrounding topography is more complex and provides more insight into the process of streak formation. Previously discussed examples occurred in relatively level plains. The area around Mojave, in contrast, changes in elevation by about 1700 m from the floor of the Chaos (where

Mojave impacted) to the surrounding high-standing islands. Yet topography appears to have little effect on the streaks and radial pattern around Mojave. For example, Mojave formed on a valley floor with facing cliffs that had no effect on streak development.

Moreover, NT-B streaks extend from PE craters radial to Mojave on the floor of an eroded crater south of Mojave (southernmost arrowhead in Figure 6a), yet the plateau between Mojave and the degraded crater is over 2000 m above the valley floor.

Consequently, the source of the wind streaks was not disrupted by either the plateau or the abrupt decrease in elevation to the crater floor.

An additional topographic effect manifests just east of the plateau cliff. Here, a 2 km-diameter crater (Figure 6a, blue arrowhead) displays bright impact-wind streaks in

THEMIS nighttime infrared. Tracing the streaks back toward Mojave reveals an apparent origin south of the crater. Schultz and Wrobel (2012) noted that most wind

117 streaks radiate from an area downrange (according to inferred impactor trajectory) from the rim of Hale. The downrange bias for Mojave-related streaks from the 2 km-diameter crater here, however, appears unique. The orientation of the cliff face likely re-directed the vapor plume as it expanded downward, thereby resulting in a slight re-direction of the streaks.

4.4 Kotka Crater

Kotka is a 40 km diameter crater that lies on the eastern edge of the Tartarus

Colles (relict Noachian highlands materials) at 19.3° north latitude and 170° east longitude (Figure 7). The crater is occurs in a late Amazonian volcanic unit, which superposes earlier Amazonian and Hesperian volcanic units to the west (Tanaka et al.,

2014). To the east, ejecta overlay a Hesperian and Noachian transition unit, along knobby relict relief of eroded highland materials. Kotka is one of only four impact-wind streak craters in volcanic plains and the only one in a young volcanic unit.

The region around Kotka exhibits variable roughness. While areas to the south and east are rough, the volcanic plains directly north and southwest of Kotka are smooth

(Kreslavsky and Head, 2002). The crater is, furthermore, on the edge of a radar-dark

‘stealth’ region having a low TES thermal inertia (Christensen and Moore, 1992;

Christensen et al., 2001), most likely related to the Medusa Fossae Formation (to the south) or Tharsis (to the east). The NT-B streaks around Kotka primarily extend to the west and southwest into the smoother, higher TES thermal inertia regions.

The THEMIS nighttime infrared contrast of the area is relatively good to the east

(into the Tartarus Colles). The brightest features in the area are long, global circulation- derived wind streaks and some of the knobs and hills of Tartarus Colles. Directly around

118

Kotka, the radial bright/dark pattern is muted and is dominated by darker streaks. Yet, nine NT-B streaks are clearly visible out to ~8 crater radii. The farthest feature is just over 200 km away to the northwest and surprisingly generated scouring vortices from a rim height of less than 5 m. The longest wind streak extends from a PE crater with rim height of 60 m only 80 km (southwest) from Kotka.

4.5 Two Unnamed Craters in or near Isidis Planitia

Two craters, one in Isidis Planitia and the other just outside Isidis Planitia, display impact-wind streaks (Figure 8). Eastern Isidis and the shaded region in Figure 8c display a high nighttime infrared contrast, and many of the craters in this area exhibit thermally dark ejecta deposits. The combined thermal contrast and ejecta give many of the craters an appearance of radial thermal streak craters. However, only two of these craters exhibit clear impact-wind streaks. For convenience, these craters are labeled Isidis-1 (to the south at 10.2°, 94.3°), and Isidis-2 (just outside the basin to the north at 16.1°, 101.0°).

Isidis-1 intersects the late Hesperian lowland unit and the early Hesperian transition unit. Both geologic units are sedimentary, and are composed of units interpreted as exhumed relicts beneath a once thick ice-rich aeolian deposit (Grizzaffi and

Schultz, 1989) or fluvial, lacustrine, and marine sediments (e.g., Parker et al., 1993). The late Hesperian lowland unit is continuous throughout most of the northern plains, but is broken within Isidis by the early Hesperian transition unit. Isidis-2 is fully within the early Hesperian transitional unit, which abuts the highland-lowland boundary (Tanaka et al., 2014).

Both geological units in the Isidis Basin are smooth and planar, as demonstrated in the roughness map (Kreslavsky and Head, 2002) but is composed of numerous small

119 domes (Grizzaffi and Schultz, 1989). Isidis-1 is more interior to the Isidis basin than

Isidis-2 and is therefore on smoother terrain. The region is of moderate to high TES thermal inertia, with values of 350-600 tiu around Isidis-1 and values of 300-350 tiu around Isidis-2 (Christensen and Moore, 1992; Christensen et al., 2001). Comparatively, in THEMIS nighttime infrared, the region has very good thermal contrast. Many small craters have bright halos, indicating a near-surface (or shallow substrate) supply of course, thermally bright material from which wind streaks may form. Areas toward the exterior of the basin (outside the late Hesperian lowland unit) have exceptionally good contrast, as noted previously.

Isidis-1 (Figure 8a) exhibits a strong THEMIS infrared brightness contrast and displays a bright/dark streaking pattern out to ~8 crater radii, despite only having three

PE craters with clear NT-B streaks. The farthest such crater is 142 km away (southwest) from Isidis-1 with a high crater rim (210 m). Coincidentally, this crater also has the longest streak, which measures 14 km. Isidis-2 (Figure 8b) is quite different from its counterpart. While it has eight clear wind streak sets extending from PE craters, the thermal contrast in the surrounding area is low. The bright/dark radial pattern is almost nonexistent and is instead muted and nearly the same brightness temperature all around the crater. Furthermore, the NT-B streaks are not as bright as those around Isidis-1. Of these features, the farthest streak is 135 km to the southwest, whereas the longest streak

(17.5 km long) is 100 km to the southeast. Interestingly, the streaks clearly superpose the ejecta deposits of at least two nearby craters, including that of a radial thermal streak crater (blue arrowheads in Figure 8b).

4.6 Prao Crater

120

Finally, Prao is an impact-wind streak crater located within the much larger

Huygens Crater (Figure 9), which is filled with a middle Noachian highland unit and a late Noachian highland unit (11.2° south latitude and 56.6° east longitude). The middle

Noachian highland unit is characterized by heavily degraded, uneven topography composed of undifferentiated basin materials (interpreted as impact, volcanic, or fluvial)

(Tanaka et al., 2014). The interior of Huygens is relatively smooth, but external to the basin, the region is rough and heavily cratered (Kreslavsky and Head, 2002).

Prao is within a variable TES thermal inertia zone. The crater itself is in moderate to low thermal inertia (~200 tiu), but the thermal inertia in the south is higher at ~300 tiu.

North of Prao, TES thermal inertia measures ~150-200 tiu (Christensen and Moore, 1992;

Christensen et al., 2001). The region is similarly variable in THEMIS nighttime infrared: darker intercrater areas with bright crater rims and floors. Few small craters have bright ejecta or halos, thereby indicating an absence of easily excavated material with high thermal contrast.

Like Isidis-2 and Kotka, the area immediately surrounding Prao is rather muted and is dominated by thermally darker streaks. Two craters with similar thermal characteristics exist to the southeast, but neither exhibit NT-B streaks. A zone of avoidance is clearly visible southwest of Prao, and is indicative of a southwest-northeast impact trajectory (Gault and Wedekind, 1978). The bright/dark pattern is more apparent to the south-southeast and extends to ~9 crater radii. Consequently, most of the 12 impact-wind streak-bearing craters occur to the south-southeast, as well. Of these, the farthest is located 120 km away to the southeast with a rim height of 20 m. The longest streak is 15 km long and is located only 60 km from Prao (almost due south).

121

5. Discussion: Comparing Radial Thermal Streak Craters and Impact-Wind

Streak Craters

5.1 Radial Thermal Streak Crater Characteristics

Radial thermal streak craters have a common appearance across Mars, though the bright/dark radial streak pattern may vary in intensity and formation. Figure 10a is a plot of the maximum streak-pattern extent versus parent crater radius for all radial thermal streak craters, including impact-wind streak craters. The general trend is an increase in maximum streak-pattern extent with increasing crater radius. Most radial thermal streak craters are less than 30 km in radius (Figure 10b). For such craters, the shallower trend indicates that the radial thermal patterns do not extend as far as impact-wind streak craters (for the same size crater). More interestingly, impact-wind streak craters have a much steeper trend, i.e., bright/dark pattern extends farther for smaller crater sizes. The difference in trends for these two populations is largely due to Hale, which has a far- reaching streak-pattern even though this study included only the lateral streaks to the northeast of Hale. Due to the inferred obliquity of Hale (Schultz and Wrobel, 2012), this collision may have generated more vapor, thereby resulting in farther reaching thermal streak patterns relative to craters of comparable size (i.e., Oudemans).

For comparison, a typical streak radius (labeled ‘nominal’ here) is the radius of a circle that fits the majority of the streak-pattern (Figure 11). This metric reduces the scatter at large crater radii and yields a closer fit between impact-wind streak craters and the rest of the radial thermal streak craters. Nevertheless, impact-wind streak craters still retain a steeper trend (extending greater distances from the crater), which may reflect their mode of formation. The bright/dark streak pattern could reflect surface

122 modification from vapor-driven winds, just as in impact-wind streak craters. Because these craters also have NT-B wind streaks, it is clear that the winds were intense and far- reaching. Other radial thermal wind streak craters may not have been subjected to such intense winds or may have formed the bright/dark streak pattern by other processes, e.g., ballistic ejecta or ejecta run-out (see discussion above). If the bright/dark patterns were also related to impact-generated winds, then they represent less intense winds.

For impact-wind streak craters specifically, streak length and range (distance from parent crater) inform about the processes occurring in these impacts. Figure 12a is a plot of the maximum streak length versus scaled range from the parent crater. The longest four streaks are included for spread. Note that the data points for Hale are only taken from the northeast streak cluster (lateral to the impact direction). Downrange, streaks extend exceptionally far away and may be related to expanding vapor and debris travelling downrange (Schultz, 1992; Schultz and Wrobel, 2012).

The plot in Figure 12a exhibits a maximum due to the longest streaks related to the largest craters (with the largest vapor plumes), Hale and Pál. For comparison, Figure

12b is a plot of the same data with Hale and Pál removed. The large scatter, especially at low scaled range, is interpreted to reflect either degradation of the wind streaks or variable intensity of the winds. Generally, the wind streaks are reduced in length at greater distances from the crater. This pattern is expected as the density of the wind- driving vapor plume and velocity (and its capacity for entrainment) reduces.

In another example, Figure 13 plots maximum streak length (of the longest four streaks per crater) versus the crater radius. The scatter in this plot could be related to atmospheric conditions, the types of material available for mobilization, or the amount of

123 impact-vaporization. The turnover in the trend at larger crater radii is likely related to the effect of surface curvature on vapor expansion, wind development, or age.

5.2 Crater Ages

Most impact-wind streak craters also have radial thermal streaks, but it is unclear if a causal relationship exists between the two crater populations (impact-wind streak craters and radial thermal streak craters). For example, impact-wind streak craters might lose their bright streaks with time and appear only as radial thermal streak craters.

Relative dates based on small crater densities from the continuous ejecta deposits of each of the 12 impact-wind streak craters and a sampling of the 35 radial thermal streak craters that do not exhibit clear NT-B streaks allow for the comparison between these two groups.

Crater counts included only craters that were larger than 500 m in diameter in order to avoid self-secondary craters and to avoid the role strength plays in craters smaller than this size. Most impact-wind streak craters (excluding Pál, Tyrrhena-Hesperia-1, Isidis-1, and Isidis-2) have fewer than five craters >500 m in diameter within their continuous ejecta deposits, which introduces a large uncertainty in the ages calculated. The resulting

Table 2 lists the results (along with estimated ages based on the Hartmann (2005) chronology) and reveals that radial thermal streak craters without NT-B streaks are typically older. In two instances, however, examples in each group have the same age.

The first example is a pair of craters in the larger Huygens Crater. Prao, one of the impact-wind streak case studies above, is ~80 Myr old. Conversely, an unnamed crater

(listed in Table 2 as Huygens Rim-1 for convenience) to the southwest does not display impact-wind streaks but is dated to ~70 Myr old.

124

Another example compares craters near Isidis Planitia. Isidis-2, an impact-wind streak crater is ~1 Gyr old, and could perhaps be as old as 3 Gyr. A radial thermal streak crater just northwest of Isidis-2 (named Isidis-3 for convenience) is also ~1.5 Gyr (up to 3

Gyr) old. Furthermore, Isidis-2 clearly superposes Isidis-3. From these two examples it is therefore apparent that age did not play a factor in removing wind streaks from radial thermal streak craters. Impact-generated wind streaks must not have formed on all radial thermal streak craters, either because there was not enough thermal contrast in the area or conditions of impact for thermal streak craters did not generate the surface-scouring winds observed at the impact-wind streak craters.

5.3 Impact-Generated Winds

The most likely explanation for the presence of impact wind streaks around some craters and not others, when other aspects (thermal signatures) are the same, is the intensity of impact-generated winds. Radial thermal streak craters display a variety of morphologies, such as ballistic ejecta, secondary cratering, and later ejecta flows

(discussed above). Accordingly, some process must then be able to generate winds intense enough to expose (or mobilize) less-resistant materials with high thermal inertia in order to produce thermally bright streaks.

Wind tunnel analyses and a study of the crater-related wind streaks in Chryse

Planitia show that wind can form horseshoe vortices around topographic obstacles (e.g.,

Iversen et al., 1976b, 1976a, Greeley et al., 1978, 1980; Greeley and Iversen, 1985), as in

Figure 14. These vortices may remove material in the wake of the crater in various patterns depending on crater size, wind speed, and particle size (Iversen et al., 1973;

Greeley and Iversen, 1985). The formation of impact-wind streaks is similar, but on a

125 different scale. Winds that lead to horseshoe vortices and deflation streaks on the lee side of raised rim craters in the wind tunnel experiments are subsonic, but Chang et al. (2010) also found that horseshoe vortices could form even at hypersonic speeds when the boundary layer is broken by a topographic high.

From previous studies (Wrobel et al., 2006; Quintana and Schultz, 2017, Chapter

2), impact-generated winds may reach or exceed designated ‘tornadic speeds,’ even in laboratory experiments. Cometary impacts result in wind speeds exceeding 1 km s-1 with sustained speeds in excess of 100 m s-1 (Quintana and Schultz, 2017, Chapter 2). In such cases, centimeter-scale material, as well as super fine material, can be mobilized (Iversen et al., 1976b; Greeley et al., 1980). The mobilization of dust has been shown to accelerate the surrounding gas and initiate further entrainment (Schultz, 1992;

Nemtchinov et al., 2002). Dust-laden winds then should then resemble natural but intensified wind erosion (sandblasting), as described in various experiments and models

(Greeley and Iversen, 1985; Greeley et al., 2000; Nemtchinov et al., 2002).

The impact-wind streaks appear bright in the THEMIS nighttime infrared images.

Bright materials may be composed of large particles (gravel, boulders) or even exhumed bedrock. The fact that these impact-wind streaks are visible today indicates that intense winds were able to scour the surface to such an extent that the streaks have not been removed (or covered) since their formation; no winds have been strong enough to remove them.

6. Wind Streak Origin

Some part of the impact process must be distinct for the 12 impact-wind streak craters on Mars, and sets them apart from not only the 35 radial thermal streak craters but

126 also all other craters of approximately the same age on the planet. This process must also explain the intensity of winds required to form such long-lasting wind streaks. Previous studies argued for impact-generated vapor-driven winds (Schultz and Wrobel, 2012;

Schultz and Quintana, 2013; Quintana and Schultz, 2014; Quintana et al., 2015; Quintana and Schultz, 2016; Quintana et al., 2016; Quintana and Schultz, 2017; Schultz and

Quintana, 2017), as opposed to other phenomena such as gravity collapse of a thermally buoyant plume (Boyce and Mouginis-Mark, 2006; Boyce et al., 2015), various models requiring water to produce multi-staged ejecta deposits (e.g., Wohletz and Sheridan,

1983; Barlow and Perez, 2003), or ejecta curtain-driven winds well after flow separation

(e.g., Schultz and Gault, 1979; Schultz, 1992; Barnouin-Jha and Schultz, 1996; Barnouin-

Jha et al., 1999a, 1999b).

First, the collapse of a buoyant, dust-laden plume is not favored because of the low atmospheric density of Mars. The atmosphere may not be able to support the required buoyant plume; rather, the tenuous atmosphere results in run-away upward expansion, as discussed in Schultz and Quintana (2017). Secondly, while the outward- moving ejecta curtain may generate trailing vortices that result in extensive ejecta flows

(Schultz and Gault, 1979; Schultz, 1992; Barnouin-Jha and Schultz, 1996; Barnouin-Jha et al., 1999a, 1999b), impact-wind streaks precede emplacement of ejecta and secondary impacts (Schultz and Quintana, 2017, Chapter 1 Figure 12) and therefore cannot be the result of such ejecta flows.

Instead, this study proposes impact vaporization as the driving force behind surface-scouring winds. The source of impact-vaporization (as a target or impactor characteristic) accounts for the 12 impact-wind streak craters, independent of location.

127

Because only about half of the impact-wind streak craters might be explained by trapped volatiles in the target, impactor speed and composition provide the most likely explanation. Le Feuvre and Wieczorek (2011) note a bimodal distribution of impact speeds at Mars. Models show that the lower impact speeds (around 10 km s-1) will not produce enough vaporization to initiate impact winds (Quintana and Schultz, 2017,

Chapter 2). Higher wind speeds, however, most likely relate to cometary impactors

(Steel, 1998). Moreover, this population would account for the paucity of impact-wind streak craters without necessitating particular surface characteristics, which are not shared by all craters.

Both long-period and short-period comets may impact Mars, although short- period comets are more likely to collide with the planet because they reside in the terrestrial planetary region (Valtonen et al., 1995). Comet flux and impact rate can be estimated with dynamical models of the Oort Cloud and the evolution Oort Cloud bodies as they traverse into the inner Solar System. Variations in these models lead to great deviations in cometary impact rate predictions (most of which are for the Earth, but estimates of Mars should be similar (Valtonen et al., 1995)). For example, Hut et al.

(1987) predicts ~50% of impacts on Earth are from comets, while Weissman (1982) derived an estimate of about 15%. Both (Bailey and Stagg, 1988) and (Valtonen et al.,

1995), however, predict that a much smaller portion (<1%) of the total impacts on Earth are cometary. Estimating the comet flux on Mars is therefore a difficult task. If impact- wind streak craters are the results of cometary collisions, then they could help to constrain this estimate for Mars.

128

If impact-wind streak craters represent cometary impacts, an estimate of the effective retention age of these impacts can be made. The 12 impact-wind streak craters over the surface of Mars (excepting the polar caps >60° north and south and the stealth regions, where impact-wind streaks are unlikely to be expressed) result in an age of approximately 100 Myr to 1 Gyr, following the Hartmann (2005) model and crater chronology (Figure 15). If instead, all radial thermal streak craters are included, the retention age shifts slightly older to ~1 Gyr.

A comparison of impact-wind streak craters to a well-studied area on Mars such as Hesperia Planum reveal that the impact-wind streak population accounts for ~2% of the Hesperia Planum population, i.e., ~1 in 50 craters would be expected to have impact- wind streaks. The broader radial thermal streak craters, alternatively, account for ~4% of the Hesperia Planum population. Thus, if impact-wind streak craters are representative of comets, then their retention age and population is related to the rate of cometary impact on Mars. Such estimates are consistent with the studies by (Bailey and Stagg, 1988) and

(Valtonen et al., 1995) that indicate a very small percentage, around 1% of impacts on

Mars are cometary.

7. Discussion: Applications

7.1 Constraints on Impactor Size

The extent of streaks and the bright/dark radial pattern around impact-wind streak craters allow a first-order estimate for the amount of vaporization produced during impact.

While winds are controlled by contrasts in pressure, mobilization and entrainment depends on density. For example, the farthest wind streak from Santa Fe Crater is 80 km away. This distance may be taken as the point at which the vapor plume density reduced

129 to ambient conditions, thereby providing a first-order estimate of the vapor mass. If the vapor expanded hemispherically to a distance of 80 km from Santa Fe, the volume of the vapor plume would be ~4.2x109 m3. The vapor mass needed to fill such a volume with a density equivalent to that of ambient Mars (~1.55x10-5 g cm-3) is 1.7x1016 g. Therefore,

1.7x1016 may be taken as a lower estimate for the amount of vapor needed to form the impact-wind streaks around Santa Fe.

If Santa Fe were formed from a 1.5 km diameter dunite (asteroid) impactor, the collision would need to produce nearly three times the projectile mass of vapor in order to form the most distal wind streaks. Models (Quintana and Schultz, 2017, Chapter 2) indicate that it would be possible to create that amount of vaporization, even in a basaltic target, if the impact speed were at least 20 km s-1. Surface ice layers or thick ice is needed in order to produce that amount of vapor for an impact speed less than 20 km s-1.

Alternatively, an ice (comet) impactor of the same size would need to produce nearly 30 times the projectile mass in vapor. If, however, this vapor had come only from the ice impactor, its diameter would need to be ~3.2 km in order to generate enough vaporization to form the most distal wind streaks at Santa Fe.

7.2 Constraints on Geological Processes

Impact wind streak craters also may help constrain other Martian geological processes. For example, Pál may be an indicator of the past obliquity cycles on Mars.

The presence of impact-wind streaks north of the crater is not solely related to the impact trajectory. The distinct absence of impact-wind streaks south of Pál exposes a different process. Polar deposits may have reached much lower latitudes during times of high obliquity (e.g., Ward, 1973; Schultz and Lutz, 1988; Ward, 1992; Wisdom and Touma,

130

1993; Baker, 2001; Mustard et al., 2001; Head et al., 2003; Milkovich and Head, 2005;

Head et al., 2005; Forget et al., 2006).

Ward (1992) indicated that spin-axis obliquity may vary up to 20° on a timescale of about 100 kyr, and further, that obliquities briefly could reach 45° if Mars were to pass through a resonance. Multiple researchers (e.g., Schultz and Lutz, 1988; Zimbelman et al., 1989; Mustard et al., 2001; Head et al., 2003) noted symmetric, latitude-dependent ice-rich mantling deposits down to latitudes of 30° north and south. The authors concluded that spin-axis obliquity resulted in climate variations: greater insolation at the poles and stability of ice at lower latitudes. They determined the deposits may be due to an ice age that lasted from about 2 to 0.5 Mya, when the spin-axis of Mars was approximately 30°.

Glaciated terrain deposits also have been proposed in the eastern Hellas and were interpreted as a response to climate variations in the past (Schultz and Lutz, 1988; Baker,

2001; Forget et al., 2006). Pál may lay on the edge of the maximum extent of obliquity- driven polar deposits, which may have caused the vapor-wind process to manifest differently to the south of the crater compared with the north (Figure 16). Impact-wind streaks appear as ridges extending from PE craters in the north (as in the example in

Figure 16b). While no streaks exist in the south, a distinct scouring pattern occurs

(Figure 16c and d), which is reminiscent of high-latitude scouring around pre-pedestal craters (Wrobel et al., 2006). Head et al. (2003) argue that the obliquity could have regularly exceeded 45° before 5 Ma, which would suggest that ice could have intermittently occurred at the equator in certain areas. The inferred 100-300 Myr age of

Pál, however, indicates that polar deposits either did not reach equatorward of ~30°

131 latitude, were not pervasive enough (or persisted) to cover or remove the impact-wind streaks north of Pál, or remain buried at depth.

8. Conclusion

Impact-wind streaks form as part of the impact process. Schultz and Quintana

(2017) developed a hypothesis of impact-vapor driven winds, which scour the surface on lee side of topographic obstacles (i.e., PE crater rims or ridges). Subsequent studies explored impact-wind development in the laboratory and through computational models with the CTH code. The catalog of impact-wind streak craters presented here, in addition to their distribution and physical features, indicates that the formation of impact-wind streaks on Mars is most consistent with the effects from rare cometary impacts.

Impact-wind streak craters are broadly distributed over Mars and show little to no clustering outside of areas of high dustcover or low thermal inertia. These areas exhibit a distinct absence impact-wind streak craters. Areas such as the Tharsis Rise, Arabia Terra, and the high latitudes are not expected to have impact-wind steaks because of the low thermal contrast. Alternatively, the same process also may be expressed differently (i.e., long run-out ejecta craters (Schultz, 1992; Wrobel et al., 2006)), particularly at high latitudes.

Case studies allow for a detailed look at the variety of impact-wind streak craters on Mars. A comparison of Xainza, Pál, Mojave, Kotka, Prao, and two unnamed craters near Isidis Planitia demonstrates the range of target types and geologic terrains that exhibit impact-wind streaks. No defining characteristic is apparent that could explain the impact-wind streaks for each of these craters. If impact-wind streaks represent cometary impacts, however, then the need for a controlling target property becomes moot.

132

Furthermore, these craters could then provide an easily identifiable cometary impact record on Mars. Specifically, if impact-wind streak craters are cometary, then they represent about 2% of the impacts in a well-studied area like Hesperia Planum.

Finally, Pál represents an additional constraint on geological processes. Namely,

Pál may define the extent of polar mantling from obliquity cycles dating 100-300 Myr.

Mars Reconnaissance Orbiter Context Camera images reveal evidence of scouring, which we interpret to be a manifestation of the same vapor-wind process that occurs in materials such as polar mantling deposits.

Acknowledgements

SNQ thanks Michael Bramble and Vivian Sun for their help in attaining data and maps for this work. This material is based upon work supported by the National Science

Foundation Graduate Research Fellowship under Grant (DGE-1058262), the Mars

Fundamental Research Program Grant (NNX13AG43G), and a Graduate Research

Fellowship from the NASA Rhode Island Space Grant Consortium (NNX10AI95H).

Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science

Foundation, NASA, or the NASA Rhode Island Space Grant Consortium.

133

References

Bailey, M.E., Stagg, C.R., 1988. Cratering constraints on the inner Oort cloud: steady- state models. Mon Not R Astron Soc 235, 1–32. doi:10.1093/mnras/235.1.1.

Baker, V.R., 2001. Water and the martian landscape. Nature 412, 228–236. doi:10.1038/35084172.

Barlow, N.G., Perez, C.B., 2003. Martian impact crater ejecta morphologies as indicators of the distribution of subsurface volatiles. Journal of Geophysical Research: Planets 108, 5085. doi:10.1029/2002JE002036.

Boyce, J.M., Wilson, L., Barlow, N.G., 2015. Origin of the outer layer of martian low- aspect ratio layered ejecta craters. Icarus 245, 263–272. doi:10.1016/j.icarus.2014.07.032.

Cabrol, N.A., Wynn-Williams, D.D., Crawford, D.A., Grin, E.A., 2001. Recent Aqueous Environments in Martian Impact Craters: An Astrobiological Perspective. Icarus 154, 98–112. doi:10.1006/icar.2001.6661.

Chang, C.-L., Choudhari, M.M., Li, F., 2010. Numerical computations of hypersonic boundary-layer over surface irregularities. AIAA Paper 1572, 2010.

134

Christensen, P.R., Bandfield, J.L., Hamilton, V.E., Ruff, S.W., Kieffer, H.H., Titus, T.N., Malin, M.C., Morris, R.V., Lane, M.D., Clark, R.L., 2001. Mars Global Surveyor Thermal Emission Spectrometer experiment: investigation description and surface science results. Journal of Geophysical Research: Planets 106, 23823–23871. doi:10.1029/2000JE001370.

Christensen, P.R., Moore, H.J., 1992. The Martian surface layer, in: Mars. pp. 686–729.

Crumpler, L.S., Arvidson, R.E., Farrand, W.H., Golombek, M.P., Grant, J.A., Ming, D.W., Mittlefehldt, D.W., Parker, T.J., 2015. Opportunity In Situ Geologic Context of Aqueous Alteration Along Offsets in the Rim of Endeavour Crater.

Edgett, K.S., Butler, B.J., Zimbelman, J.R., Hamilton, V.E., 1997. Geologic context of the Mars radar “Stealth” region in southwestern Tharsis. J. Geophys. Res. 102, 21545–21567. doi:10.1029/97JE01685.

Forget, F., Haberle, R.M., Montmessin, F., Levrard, B., Head, J.W., 2006. Formation of Glaciers on Mars by Atmospheric Precipitation at High Obliquity. Science 311, 368–371. doi:10.1126/science.1120335.

Gault, D.E., Wedekind, J.A., 1978. Experimental studies of oblique impact, in: Lunar and Planetary Science Conference Proceedings. pp. 3843–3875.

Greeley, R., Iversen, J.D., 1985. Wind as a geological process on Earth, Mars, Venus and Titan. Cambridge University Press, New York.

135

Greeley, R., Papson, R., Veverka, J., 1978. Crater streaks in the Chryse Planitia region of Mars: Early Viking results. Icarus 34, 556–567. doi:10.1016/0019- 1035(78)90045-3.

Greeley, R., Wilson, G., Coquilla, R., White, B., Haberle, R., 2000. Windblown dust on Mars: laboratory simulations of flux as a function of surface roughness. Planetary and Space Science, Mars Exploration Program 48, 1349–1355. doi:10.1016/S0032-0633(00)00115-X.

Grizzaffi, P., Schultz, P.H., 1989. Isidis Basin: Site of ancient volatile-rich debris layer. Icarus 77, 358–381. doi:10.1016/0019-1035(89)90094-8.

Hartmann, W.K., 2005. Martian cratering 8: Isochron refinement and the chronology of Mars. Icarus 174, 294–320. doi:10.1016/j.icarus.2004.11.023.

Head, J.W., Mustard, J.F., Kreslavsky, M.A., Milliken, R.E., Marchant, D.R., 2003. Recent ice ages on Mars. Nature 426, 797–802. doi:10.1038/nature02114.

Head, J.W., Neukum, G., Jaumann, R., Hiesinger, H., Hauber, E., Carr, M., Masson, P., Foing, B., Hoffmann, H., Kreslavsky, M., Werner, S., Milkovich, S., van Gasselt, S., Team, T.H.C.-I., 2005. Tropical to mid-latitude snow and ice accumulation, flow and glaciation on Mars. Nature 434, 346–351. doi:10.1038/nature03359.

Hut, P., Alvarez, W., Elder, W.P., Hansen, T., Kauffman, E.G., Keller, G., Shoemaker, E.M., Weissman, P.R., 1987. Comet showers as a cause of mass extinctions. Nature 329, 118–126. doi:10.1038/329118a0.

Iversen, J.D., Greeley, R., Pollack, J.B., 1976a. Windblown dust on earth, Mars and Venus. Journal of the Atmospheric Sciences 33, 2425–2429. doi:10.1175/1520- 0469(1976)033<2425:WDOEMA>2.0.CO;2.

136

Iversen, J.D., Pollack, J.B., Greeley, R., White, B.R., 1976b. Saltation threshold on Mars: the effect of interparticle force, surface roughness, and low atmospheric density. Icarus 29, 381–393. doi:10.1016/0019-1035(76)90140-8.

Iversen, J.D., White, B.R., Greeley, R., Pollack, J.B., 1973. Simulations of Martian Eolian Phenomena in the Atmospheric Wind Tunnel. NASA Special Publication 36, 191–213.

Karunatillake, S., Wray, J.J., Gasnault, O., McLennan, S.M., Rogers, A.D., Squyres, S.W., Boynton, W.V., Skok, J.R., Ojha, L., Olsen, N., 2014. Sulfates hydrating bulk soil in the Martian low and middle latitudes. Geophysical Research Letters 41, 2014GL061136. doi:10.1002/2014GL061136.

Kieffer, H.H., 1977. Thermal Mapping of Mars: Polar and Nighttime Results, in: Bulletin of the American Astronomical Society. p. 441.

Kreslavsky, M.A., Head, J.W., 2002. Mars: Nature and evolution of young latitude- dependent water-ice-rich mantle. Geophysical Research Letters 29, 14–1. doi:10.1029/2002GL015392.

Le Feuvre, M., Wieczorek, M.A., 2011. Nonuniform cratering of the Moon and a revised crater chronology of the inner Solar System. Icarus 214, 1–20. doi:10.1016/j.icarus.2011.03.010.

Maine, A.R.W., 2017. The geologic analysis of Esira: A central pit crater on Mars (M.S.). Northern Arizona University, Arizone, United States.

137

McGlaun, J.M., Thompson, S.L., Elrick, M.G., 1990. CTH: A three-dimensional shock wave physics code. International Journal of Impact Engineering 10, 351–360. doi:10.1016/0734-743X(90)90071-3.

Mest, S.C., Crown, D.A., 2014. Geologic map of MTM-30247,-35247, and-40247 quadrangles, Reull Vallis region of Mars. J. Morphol 275, 745–759.

Milkovich, S.M., Head, J.W., 2005. North polar cap of Mars: Polar layered deposit characterization and identification of a fundamental climate signal. Journal of Geophysical Research: Planets 110, E01005. doi:10.1029/2004JE002349.

Morgan, G.A., Head, J.W., 2009. Sinton crater, Mars: Evidence for impact into a plateau icefield and melting to produce valley networks at the Hesperian–Amazonian boundary. Icarus 202, 39–59. doi:10.1016/j.icarus.2009.02.025.

Morris, A.R., Mouginis-Mark, P.J., Garbeil, H., 2010. Possible impact melt and debris flows at Tooting Crater, Mars. Icarus 209, 369–389. doi:10.1016/j.icarus.2010.05.029.

Muhleman, D.O., Butler, B.J., 1991. Radar images of Mars. Science 253, 1508. doi:10.1126/science.253.5027.1508.

Mustard, J.F., Cooper, C.D., Rifkin, M.K., 2001. Evidence for recent climate change on Mars from the identification of youthful near-surface ground ice. Nature 412, 411–414. doi:10.1038/35086515.

Nemtchinov, I.V., Shuvalov, V.V., Greeley, R., 2002. Impact-mobilized dust in the Martian atmosphere. Journal of Geophysical Research 107, 5134. doi:10.1029/2001JE001834.

Ody, A., Poulet, F., Langevin, Y., Bibring, J.-P., Bellucci, G., Altieri, F., Gondet, B., Vincendon, M., Carter, J., Manaud, N., 2012. Global maps of anhydrous minerals at the surface of Mars from OMEGA/MEx. Journal of Geophysical Research: Planets 117, E00J14. doi:10.1029/2012JE004117.

Parker, T.J., Gorsline, D.S., Saunders, R.S., Pieri, D.C., Schneeberger, D.M., 1993. Coastal geomorphology of the Martian northern plains. J. Geophys. Res. 98, 11061–11078. doi:10.1029/93JE00618.

Pelkey, S.M., Jakosky, B.M., Mellon, M.T., 2001. Thermal inertia of crater-related wind streaks on Mars. Journal of Geophysical Research: Planets 106, 23909–23920. doi:10.1029/2000JE001433.

138

Presley, M.A., Christensen, P.R., 1997a. Thermal conductivity measurements of particulate materials 1. A review. Journal of Geophysical Research: Planets 102, 6535–6549. doi:10.1029/96JE03302.

Presley, M.A., Christensen, P.R., 1997b. Thermal conductivity measurements of particulate materials 2. Results. Journal of Geophysical Research: Planets 102, 6551–6566. doi:10.1029/96JE03303.

Putzig, N.E., Mellon, M.T., Jakosky, B.M., Pelkey, S.M., Martínez-Alonso, S., Hynek, B.M., Murphy, N.W., 2004. Mars Thermal Inertia from THEMIS Data, in: Lunar and Planetary Institute Science Conference Abstracts. Lunar and Planetary Institute Science Conference 35, Abstract 1863.

Quintana, S.N., Schultz, P.H., 2014. The Formation of Crater-Related Blast Wind Streaks on Mars. Presented at the Lunar and Planetary Science Conference 45, Abstract 1971.

Quintana, S.N., Schultz, P.H., 2016. A Global Distribution of Impact-Wind Streak Craters on Mars. Presented at the Lunar and Planetary Science Conference 47, Abstract 1548.

Quintana, S.N., Schultz, P.H., 2017. Model Results for Impact-Winds on Mars. Presented at the Lunar and Planetary Science Conference 48, Abstract 1123.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2015. Experimental Results Supporting an Impact-Related Blast Wind Formation Mechanism for Some Wind Streaks on Mars. Presented at the Lunar and Planetary Science Conference 46, Abstract 2469.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2016. New Experiments in Martian Impact Vapor-Induced Wind Streak Analysis. Presented at the Lunar and Planetary Science Conference 47, Abstract 1553.

Schultz, P.H., 1992. Atmospheric effects on ejecta emplacement. Journal of Geophysical Research: Planets 97, 11623–11662. doi:10.1029/92JE00613.

139

Schultz, P.H., Gault, D.E., 1979. Atmospheric effects on Martian ejecta emplacement. Journal of Geophysical Research: Solid Earth 84, 7669–7687. doi:10.1029/JB084iB13p07669.

Schultz, P.H., Lutz, A.B., 1988. Polar wandering of Mars. Icarus 73, 91–141. doi:10.1016/0019-1035(88)90087-5.

Schultz, P.H., Quintana, S., 2013. Impact Blast Wind Scouring on Mars. Presented at the Lunar and Planetary Science Conference 44, Abstract 2697.

Schultz, P.H., Quintana, S.N., 2017. Impact-generated winds on Mars. Icarus 292, 86– 101. doi:10.1016/j.icarus.2017.03.029.

Schultz, P.H., Wrobel, K.E., 2012. The oblique impact Hale and its consequences on Mars. Journal of Geophysical Research 117, E04001. doi:10.1029/2011JE003843.

Steel, D., 1998. Distributions and moments of asteroid and comet impact speeds upon the Earth and Mars. Planetary and Space Science 46, 473–478. doi:10.1016/S0032- 0633(97)00232-8.

Tanaka, K.L., 1999. Debris-flow origin for the Simud/Tiu deposit on Mars. J. Geophys. Res. 104, 8637–8652. doi:10.1029/98JE02552.

Tanaka, K.L., Robbins, S.J., Fortezzo, C.M., Skinner Jr., J.A., Hare, T.M., 2014. The digital global geologic map of Mars: Chronostratigraphic ages, topographic and crater morphologic characteristics, and updated resurfacing history. Planetary and Space Science, Planetary Geology Field Symposium, Kitakyushu, Japan, 2011: Planetary Geology and Terrestrial Analogs 95, 11–24. doi:10.1016/j.pss.2013.03.006.

Thomas, P., Veverka, J., Lee, S., Bloom, A., 1981. Classification of wind streaks on Mars. Icarus 45, 124–153. doi:10.1016/0019-1035(81)90010-5.

Tornabene, L.L., Moersch, J.E., McSween, H.Y., McEwen, A.S., Piatek, J.L., Milam, K.A., Christensen, P.R., 2006. Identification of large (2–10 km) rayed craters on Mars in THEMIS thermal infrared images: Implications for possible Martian meteorite source regions. J. Geophys. Res. 111, E10006. doi:10.1029/2005JE002600.

Valtonen, M.J., Zheng, J.Q., Matese, J.J., Whitman, P.G., 1995. Near-Earth populations of bodies coming from the Oort cloud and their impacts with planets. Earth Moon Planet 71, 219–223. doi:10.1007/BF00612962.

140

Ward, W.R., 1973. Large-Scale Variations in the Obliquity of Mars. Science 181, 260– 262. doi:10.1126/science.181.4096.260.

Ward, W.R., 1992. Long-term orbital and spin dynamics of Mars, in: Mars. pp. 298–320.

Wechsler, A.E., Glaser, P.E., 1965. Pressure effects on postulated lunar materials. Icarus 4, 335–352. doi:10.1016/0019-1035(65)90038-2.

Weissman, P.R., 1982. Terrestrial impact rates for long and short-period comets. Geological Society of America Special Papers 190, 15–24. doi:10.1130/SPE190- p15.

Werner, S.C., Ody, A., Poulet, F., 2014. The Source Crater of Martian Shergottite Meteorites. Science 343, 1343–1346. doi:10.1126/science.1247282.

Williams, R.M.E., Malin, M.C., 2008. Sub-kilometer fans in Mojave Crater, Mars. Icarus 198, 365–383. doi:10.1016/j.icarus.2008.07.013.

Wisdom, J., Touma, J., 1993. The Chaotic Obliquity of Mars. Presented at the Bulletin of the American Astronomical Society, p. 04.09.

Wohletz, K.H., Sheridan, M.F., 1983. Martian rampart crater ejecta: Experiments and analysis of melt-water interaction. Icarus 56, 15–37. doi:10.1016/0019- 1035(83)90125-2.

Zimbelman, J.R., Clifford, S.M., Williams, S.H., 1989. Concentric crater fill on Mars-an aeolian alternative to ice-rich mass wasting, in: Lunar and Planetary Science Conference Proceedings. pp. 397–407.

141

Tables

209

120

137

(m)

Rim

Max

Height

306

305

306

305

102

100

102

316

331

321

316

Long. Long.

Lat.

135

142

135

271

1002

(km)

crater

source

Distance Distance

Streak

Closest

Longest

Shortest

Farthest

Characteristic

>54

Streak

Sources

of Wind

Number Number

Size Size

(km)

16.16

37.05

51.30

28.13

137.31

53.95

36.36

94.30

Long.

100.98

101.21

6.10

9.82

Lat.

35.69

25.11

10.23

Wind Streak Craters and Streak Statistics and Streak Streak Craters Wind

Impact

Jijiga

Hale

Name

Isidis

Hesperia

Note: Table 1 continued on next page. on next 1 continued Note: Table

Table 1 Table

142

166

118

147

(m)

Rim

Max

Height

311

312

313

312

112

113

110

113

326

327

326

324

166

169

166

Long. Long.

Lat.

120

184

386

199

201

(km)

crater

source

Distance Distance

Streak

Closest

Longest

Shortest

Farthest

Characteristic

Streak

Sources

of Wind

Number Number

Size Size

(km)

20.30

18.80

71.21

57.97

39.45

47.95

32.99

56.59

Long.

108.70

169.88

7.48

Lat.

11.17

31.31

19.28

19.25

Pal

Prao

Kotka

Name

Mojave

Santa Fe

Note: Table 1 continued on next page. on next 1 continued Note: Table Table 1 continuedTable

143

(m)

Rim

Max

Height

356

355

100

101

Long. Long.

Lat.

162

155

(km)

crater

source

Distance Distance

Streak

Closest

Longest

Shortest

Farthest

Characteristic

Streak

Sources

of Wind

Number Number

Size Size

(km)

23.62

34.80

3.94

98.73

Long.

0.78

Lat.

18.49

Name

Xainza

Tyrrhena

Hesperia

Table 1 continuedTable

144

Table 2 - Estimated ages of impact-wind streak craters (top), select radial thermal streak craters (middle), and two other well-preserved craters without thermal streaks of any kind (bottom)

푵풄(> ퟓퟎퟎ 퐦) Estimated Name 푨 Other Age Estimates (Reference) ( ) Age ퟏퟎퟔ 퐤퐦ퟐ Xainza - - - Kotka 1.83E+02 30 My - <5 Myr (Werner et al., 2014), Mojave 6.04E+01 9 My Late Hesperian (Williams and Malin, 2008) Main, 2017) Isidis-3 1.42E+04 1.5-3 Gy - Isidis-4 1.37E+04 1.5-3 Gy - Chryse-1 1.98E+04 2-3 Gy - Ejriksson 1.03E+04 1-2 Gy - Sabaea-3 1.44E+04 1.5-3 Gy - Daedalia-1 7.43E+03 1 Gy - 700 My-2 Hesperian/Amazonian boundary (Morgan Sinton 5.97E+03 Gy and Head, 2009) Tooting - - <2 Myr (Morris et al., 2010) a Based on the Hartmann (2005) chronology. b These craters have fewer than five craters >500 m on their continuous ejecta deposits, which introduces large error in the estimated dates

145

Figure Captions

Figure 1 - The crater Pál is used to describe terminology. Pál, highlighted with the yellow circle, is the impact-wind streak crater. The yellow arrows indicate several preexisting (PE) craters from which the wind streaks extend. The alternating, radial bright/dark pattern around Pál (not associated with the wind streaks) is common for most impact-wind streak craters and an additional 35 craters that do not have NT-B wind streaks.

Figure 2 - The global distribution of 12 impact-wind streak craters (black squares) compared to the distribution of 35 other radial thermal wind streak craters (white squares) over a Mars Orbiter Laser Altimeter (MOLA) shaded topography basemap.

Figure 3 - The same distribution as shown in Figure 2, with overlays of (a) OMEGA dust

(from nanophase ferric oxide, Ody et al. (2012)); (b) TES thermal inertia (Christensen and Moore, 1992; Christensen et al., 2001); (c) contours of the highest levels of stoichiometrically equivalent H2O from gamma ray spectroscopy (Karunatillake et al.,

2014). Areas of high dust cover correlate well with areas of low TES thermal inertia

(stealth regions (e.g., Muhleman and Butler, 1991; Edgett, 2002; Karunatillake et al.,

2009)). Few impact-wind streak craters are within the lowest thermal inertia areas or the areas of highest near-surface water.

Figure 4 - Figures for the case study for Xainza Crater in Meridiani Planum. (a)

THEMIS nighttime infrared mosaic of the area around Xainza (yellow arrow), with examples of impact-wind streaks extending from PE craters highlighted with yellow arrowheads. Inset shows a global view with a box around the region shown in (b-e). (b)

146

MOLA color topography of the area around Xainza. The box outlines a closer view shown in (a). (c) A section of the Tanaka et al. (2014) geologic map that corresponds to the same area and scale as in (b). In (d), TES thermal inertia (Christensen and Moore,

1992; Christensen et al., 2001) overlays the region shown in (b) and indicates that Xainza is in a moderate thermal inertia area. The area and scale are the same as in (b). (d)

Roughness (Kreslavsky and Head, 2002) around Xainza (darker colors are smoother terrain) for the same area and scale as in (b). The roughness is kilometer-scale roughness from MOLA plotted in an RGB composite (R=9.6 km, G=2.4 km, and B=0.6 km roughness). The HNhu unit around Xainza is relatively smooth, particularly to the south.

Xainza is indicated by the arrow in each image. Geologic units are as follows (Tanaka et al., 2014): mNh = middle Noachian highland unit; HNhu = Hesperian and Noachian highland undivided unit; Ahi = Amazonian and Hesperian impact unit.

Figure 5 - The crater Pál. The general figure descriptions are the same as in Figure 4.

Pál is located in the bottom left of (a) because no wind streaks are visible to the south of

Pál. The geologic units in (c) are as follows: AHi = Amazonian and Hesperian impact unit; ANa = Amazonian and Noachian apron unit; eHv = early Hesperian volcanic unit; eNh = early Noachian highland unit; eNhm = early Noachian highland massif unit; Hve

= Hesperian volcanic edifice unit; lNh = late Noachian highland unit; lHv = late

Hesperian volcanic edifice; Nve = Noachian volcanic edifice unit. Pál is in an area of (d) low to moderate thermal inertia and (e) relatively smooth (dark) terrain.

Figure 6 - The crater Mojave, located in the massif rings of Chryse basin. The general figure descriptions are the same as in Figure 4. Mojave is at the intersection of several

147 geologic terrains with many others in the region, as well: AHi = Amazonian and

Hesperian impact unit; eHt = early Hesperian transition unit; eNh = early Noachian highlands unit; lHt = late Hesperian transition unit; lNh = late Noachian highland unit; Ht

= Hesperian transition unit; Hto = Hesperian transition outflow unit; mNh = middle

Noachian highland unit; Nhu = Noachian highland undivided unit. Mojave is located in

(d) an area of low thermal inertia and (e) relatively smooth terrain.

Figure 7 - Kotka Crater, located near Tartarus Colles. The general figure descriptions are the same as in Figure 4. The radial thermal streaks around Kotka in (a) are darker than in many other impact-wind streak craters, perhaps because of the proximity to a stealth region and availability of finer material (d). The region directly surrounding

Kotka is moderately rough (light grey in (e)), whereas the area to the southeast is much rougher. To the north and southwest, the region is much smoother (darker greys). Kotka impacted into volcanic units, and the geologic units in the area are as follows: AHi =

Amazonian and Hesperian impact unit; AHv = Amazonian and Hesperian volcanic unit; eHt = early Hesperian transition unit; HNt = Hesperian and Noachian transition unit; Hve

= Hesperian volcanic edifice unit; lHvf = late Hesperian volcanic field unit; lAv = late

Amazonian volcanic unit.

Figure 8 - Isidis Planitia and the region just southeast of it hosts two unnamed impact- wind streak craters. The southernmost crater is Isidis-1 (a), which has very good contrast in THEMIS nighttime infrared and exhibits a much clear bright/dark radial pattern and bright impact-wind streaks. Comparatively, the northernmost crater, Isidis-2 (b), has poor thermal contrast with a nearly imperceptible bright/dark radial pattern, yet has more

148 impact-wind streaks than Isidis-1. Arrowheads indicate examples of impact-wind streaks around PE craters, and the blue arrowheads indicate two cases of impact-wind streaks from Isidis-2 that superpose the ejecta or thermal bright/dark pattern of other craters in the area. The globe in (a) indicates the location of (c-f). (c) MOLA color topography of the region outlined in the inset globe (a), with boxes indicating the area shown in (a) and

(b). The area and scale are the same as in (c-f). The area shaded in (c) has particularly good thermal contrast, as referenced in the text. (d) Geologic units AHi = Amazonian and Hesperian impact unit; eHt = early Hesperian transition unit; eHv = early Hesperian volcanic unit; eNh = early Noachian highland unit; lHl = late Hesperian lowland unit; lHt

= late Hesperian transition unit; lNh = late Noachian highland unit; mNh = middle

Noachain highland unit; mNhm = middle Noachian highland massif unit. (e) A portion of the global TES thermal inertia map. Isidis-1 is in an area of low thermal inertia, whereas

Isidis-2 is in moderate thermal inertia. (f) A portion of the global roughness map indicating that both Isidis-1 and Isidis-2 are located in smooth (dark grey) areas. Arrows point to Isidis-1 and Isidis-2.

Figure 9 - Prao crater, located inside the larger Huygens Crater. General figure descriptions are the same as in Figure 4. Similar to Kotka, the radial thermal pattern around Prao (in (a)) is mostly dark, which indicates that it has a finer composition than the surroundings, which is corroborated by the TES thermal inertia (d). Prao is in a smooth terrain (dark grey), like much of the interior of the Huygens crater (e). Geologic units (c) are as follows: eNh = early Noachian highland unit; mNh = middle Noachian highland unit; lNh = late Noachian highland unit.

149

Figure 10 - The maximum extent of the radial bright/dark pattern around radial thermal streak craters (blue circles) and impact-wind streak craters (black circles) over parent crater radius (Rc). The box in (a) outlines the extent of the plot in (b), and removes the streaks of the largest craters in the two groups. See text for details.

Figure 11 - For comparison with Figure 10, the ‘nominal’ extent of the radial bright/dark pattern is plotted here. The nominal extent is the radius of a circle that enclosed over half of the bright/dark pattern. Variable labels are the same as in Figure 10. Less of a difference exists between radial thermal streak craters and impact-wind streak craters when taking the nominal extent. Differences are attributed to formation process (in particular, wind versus ejecta flow processes).

Figure 12 - These plots focus on impact-wind streak craters specifically. (a) is a plot of the four longest streaks for each crater versus the scaled range (distance from preexisting crater to parent crater scaled by parent crater radius). The largest craters, Hale and Pál

are responsible for the long streaks around 6-10 r/Rc (range—parent crater to preexisting crater—over crater radius). The box indicates the area plotted in (b) and removes those two craters. The scatter is due to wind streak degradation over time.

Figure 13 - For comparison, the streak length of the longest four steaks per crater is plotted here against parent crater radius. The largest craters, Hale and Pál, indicate the effects of surface curvature, as the trend begins to fall off at large Rc.

Figure 14 - A diagram of horseshoe vortex formation on the lee of a preexisting topographic obstacle (i.e., a crater), based on (Greeley et al., 1974). Wind is deflected

150 around the obstacle. When the wind detaches from the crater wall, vortices initiate and can cause surface modification downwind.

Figure 15 - A cumulative plot of impact-wind streak craters (black circles) and radial thermal streak craters (white circles) over the surface area of Mars (excepting the caps northward of 60° latitude and the stealth regions), overlain on the chronology of

Hartmann (2005).

Figure 16 - The crater Pál may mark the edge of the maximum extent of obliquity-driven polar deposits. The THEMIS nighttime infrared mosaic in (a) clearly exhibits thermally bright impact-wind streaks to the north of Pál, but the streaks are missing to the south.

White boxes outline the Mars Reconnaissance Orbiter Context Camera (CTX) footprints.

Yellow boxes outline the extent of the images in (b-d). (b) is part of a CTX image

(D21_035272_1506_XN_29S249W). Ridges (yellow arrows) clearly extend from the

3.9 km diameter crater (~150 km from Pál), which correlate with the bright streaks in the

THEMIS image. (c) is part of CTX image B19_017207_1456_XI_31S253W, centered

~100 km from Pál. No ridges corresponding to thermally bright streaks are present, but a linear scouring pattern exists. The linear grooves can be traced to Pál (black arrow).

Farther out, (d) is part of CTX image D15_033136_1473_XN32S252W, centered ~165 km from Pál. The grooves become more muted but still trace back to the parent crater

(black arrow). The scouring pattern is reminiscent of that described around high-latitude pre-pedestal craters (Wrobel et al., 2006) and indicates a different manifestation of the same process occurring to form impact-wind streaks.

151

Figures

Figure 1

152

Figure 2

153

Figure 3

154

Figure 4

155

Figure 5

156

Figure 6

157

Figure 7

158

Figure 8

159

Figure 9

160

Figure 10

161

Figure 11

162

Figure 12

163

Figure 13

164

Figure 14

165

Figure 15

166

Figure 16

167

CHAPTER FOUR:

Cometary Impacts and Impact-Wind Streaks on Mars

Stephanie N. Quintana1

and

Peter H. Schultz1

1Department of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI 02912

168

Abstract

Impact-wind streaks are thermally bright features radiating from certain craters on Mars.

They are proposed to be the result of impact vapor-driven winds that scour the surface downrange of preexisting (PE) obstacles, such as craters or ridges. Only twelve craters exhibit such streaks. Previous work assessed vapor and wind production in the laboratory, the generation of enhanced vaporization on Mars, and constraints on the formation of impact winds based on the global distribution of impact-wind streaks on Mars. These prior efforts concluded that comets are a likely source for the impact vaporization needed to generate surface-scouring winds. As a result, this contribution first assesses the effects of a cometary impact in the laboratory with a nylon projectile. The easily volatilized impactor produces pervasive winds downrange. Second, a study of specific craters on

Mars with abrupt topographic changes allows for further development of the impact vapor-driven wind model and for testing against other formation processes. Finally, computational simulations assess the possibility of streak formation around Jijiga crater, one of the examples of impact-wind streaks forming despite extreme topographic changes.

These simulations further support the vapor-driven wind model of streak formation.

169

1. Introduction

Impact-wind streaks are a subset of crater-related wind streaks on Mars that occur around only 12 craters on the planet. Primarily visible in Thermal Emission Imaging

System (THEMIS) nighttime infrared images, the wind streaks appear as bright features that extend from smaller PE craters or other topographic obstacles around a larger parent crater. From a detailed analysis of the streaks, their parent craters, and their geologic setting (Schultz and Wrobel, 2012; Quintana and Schultz, 2016; Schultz and Quintana,

2017, Chapter 3), as well as laboratory experiments and computational models (Wrobel et al., 2006; Quintana and Schultz, 2014; Quintana et al., 2015, 2016; Quintana and Schultz,

2017, Chapter 1, Chapter 2), we proposed that the wind streaks are formed by impact- vapor driven winds.

In this model, the collision of an asteroid or comet with Mars generates an expanding vapor plume that mobilizes winds as it expands into the surrounding atmosphere. Such winds are intense and long lasting, interact with the surface (via vortices from a topographic obstacle), and can scour the surface in the lee (downwind) of obstacles. Where such obstacles are PE craters, the winds mobilize and rework ejecta deposits, plucking and redepositing thermally bright material (gravel and boulders) that then stands in relief in long ridges (Schultz and Wrobel, 2012; Schultz and Quintana,

2017). Figure 1 is an example from the crater Mojave that identifies two PE craters with ridges seen in Context Camera (CTX) images that correspond to thermally bright streaks in THEMIS nighttime infrared images.

Previous work focused on specific aspects of the impact-wind streak forming process as summarized below. This contribution combines these aspects with new

170 experiments and models in order to focus on the most likely source for the impact-wind streak craters: cometary impacts. It culminates in a case study of two craters near the

Chryse Planitia area of Mars that explores the effect of topography on streak formation.

2. Background

Tests for the formation model of the thermally bright features (Schultz and

Quintana, 2017) requires addressing three conditions. It was first necessary to demonstrate that vapor expansion into an atmosphere could induce surface-interacting winds. Second, a source for high amounts of impact vapor production on Mars was required. Finally, the formation model had to address the paucity of impact-wind streak craters across Mars. The following background provides the basis for these conditions.

2.1 Vapor-Driven Winds

In high-explosive and nuclear detonations on Earth, the shockwave and vapor plume expand together resulting in a destructive blast wind (Glasstone and Dolan, 1977).

The low density of the Martian atmosphere, however, results in the impact blast (overpressure) rapidly expanding and equilibrating (Schultz, 1992). Moreover, previous studies (Wrobel et al., 2006; Schultz and Wrobel, 2012) inferred vapor-driven processes are the cause of surface scouring at high latitudes and thermally bright streaks around the crater Hale, respectively. A series of experiments (Quintana et al., 2015, 2016) at the

NASA Ames Vertical Gun Range (AVGR) investigated the possible role of vapor expansion on wind streak formation compared to the role of an atmospheric shockwave generated by the first moment of contact between the projectile and target.

Various studies (Schultz, 1988, 1996; Bruck Syal and Schultz, 2014; Schultz and

Eberhardy, 2015) demonstrated that impact vaporization can occur in easily-volatilized

171 materials, even at the relatively low impact speeds (< 6km/s) achievable at the AVGR.

Here, experiments used Pyrex projectiles and powdered dolomite targets in order to 1) generate an expanding impact-vapor plume, and 2) provide insight for computational models. Such experiments demonstrated a genetic relationship between the passage of the vapor plume and the initiation of winds, revealed by detachment and mobilization of particles off vertical dusty pipe cleaners (Quintana et al., 2016, Chapter 1). Although masked in nominal atmospheres (air, argon) by overlapping processes, impacts into low- density gas (helium) clearly exposed a separation between the shockwave and the vapor plume due to the faster moving shockwave. In this case, passing winds did not entrain particles off the pipe cleaners until well after the atmospheric shock had passed.

Application of these results to Mars indicates that 1) the thin atmosphere eventually would separate the shockwave from the expanding vapor plume, and 2) the vapor expansion through the Martian atmosphere should drive longer lasting, surface-modifying winds.

2.2 Impact Vaporization

A suite of models using the CTH family of shock physics codes (McGlaun et al.,

1990; Hertel et al., 1993) explored impact vapor generation in planetary-scale impacts that simulated the present-day Martian environment (temperature, surface atmospheric pressure, and gravity) (Quintana and Schultz, 2014, 2017). Because impact-wind streak craters are rare, enhanced impact vaporization (compared to a nominal impact on Mars) may be needed to account for the observed streaks. Hence, models tested target and impactor conditions that could lead to enhanced vaporization from a control case, represented by a dunite (astroidal) impact into a solid, basaltic plain.

172

Model results revealed that objects impacting Mars at speeds above 12 km s-1 could produce impact vaporization, but volatiles, rather than silicates, were required in order to develop sufficient vapor that could interact with the surface (Quintana and

Schultz, 2014, 2017). High-speed (>20 km s-1) comets produced winds that were consistent with observations: outward-moving, sustained winds over the length of the calculation, with wind speeds over 80 m s-1 (and maximum speeds over 300 m s-1).

While some cases of volatiles at or near the surface also produced sufficient vapor and winds that met the streak-forming criteria, such cases have limited applications on Mars.

Specifically, impact winds would simply modify the volatile rich surface, which would later disappear along with the surface expression. Comets, however, account for both the paucity of impact-wind streak craters and their global distribution.

2.3 Impact-Wind Streak Craters

A global survey revealed only 12 impact-wind streak craters on Mars, and their ages ranged from less than 10 Myr to about 3 Gyr according to the Hartmann (2005) correlation (Chapter 3). This study mapped the global distribution of impact-wind streak craters and explored commonalities among them. The most notable trend is that impact- wind streak craters are less likely to form in dusty regions where the local thermal contrast is insufficient to reveal impact-wind streaks in nighttime infrared images. A case study of seven impact-wind streak craters (Xainza, Pál, Mojave, Kotka, Prao, and two unnamed craters in the vicinity of southeastern Isidis Planitia) found that the streaks form in a variety of terrains, geologic units, and settings. Such craters do not have a clear dependence on locality, as nearly equal numbers are found in the northern lowlands and the southern highlands. The only consistent trend between impact-wind streak craters

173 was their formation on relatively smooth surfaces at the scale of MOLA-derived roughness (Kreslavsky and Head, 2002). Thus, no controlling target factor can be attributed to impact-wind streak formation.

3. Cometary Origin of Impact-Generated Winds

Previous studies concluded that several possible conditions could account for impact-wind streaks, including substantial surface ice, thick subsurface ice, and cometary impacts. The first two alternatives, however, are inconsistent with the observed wind- streak distribution: lower latitudes with minimal evidence for thick (500 m) subsurface or surface volatiles over the last billion years. As a result, this contribution explores the possible role of a cometary impact through several different strategies: (a) laboratory experiments using an easily volatized impactor; (b) specific examples on Mars where topography may help to distinguish among different processes; and (c) computational modeling applied to these specific examples.

3.1 Methods

Laboratory experiments performed at the NASA Ames Vertical Gun Range

(AVGR) incorporated a different impactor and target setup than that described in

(Quintana et al., 2016, Chapter 1). A nylon projectile (as opposed to Pyrex) provided a better comet analog for investigating the effects of impactor vaporization in the laboratory. Nylon easily absorbs moisture and has a thermal decomposition temperature of 583 K, and a vaporization point of 653 K (for Nylon-6, Beyler and Hirschler, 2002).

The target consisted of either powdered dolomite (for comparison with previous work) or sand (in order to isolate and test the impactor vaporization characteristics). An additional trial assessed limited vaporization with a single impact of a Pyrex projectile into sand.

174

The AVGR test chamber allows for different ambient atmospheres during each test run. Most commonly, runs in this set of experiments used an atmosphere of air at 33 mbar (25 Torr). Two additional experiments compared the effects of atmospheric density: one with a helium atmosphere at the same pressure; the other, with air at 67 mbar

(50 Torr). Table 1 lists the experimental runs.

The test chamber can be instrumented with a wide array of detection devices, including several high-speed cameras and tracers placed on and around the target in order to allow visual tracking of different impact processes. This study used eight high-speed cameras that recorded in both color and black and white: Phantom® high-speed cameras with frame rates up to 28,000 fps and Shimadzu® imaging cameras at 125,000 fps.

Various cameras (with different viewpoints) captured different aspects of the impact process, aided by the use of two kinds of tracers. Vertical pipe cleaners, dusted with powdered dolomite (< 20 μm) allowed tracking and quantifying otherwise invisible winds. The pipe cleaners were arranged in lines ~5 cm apart, and placed downrange

(offset ~25°), uprange, and orthogonal to the impactor trajectory (as in Figure 2).

Additionally, millimeter-sized Styrofoam balls allowed the tracking of wind development at the surface of the target and just outside of the target container (Figure 2).

3.2 Results

The nylon projectile impact caused dramatic effects in vapor production, luminosity, and wind development, particularly downrange (Figure 3). A comparison between different impactor/target combinations found that nylon impacting into sand resulted in the greatest mobilization of dust downrange, followed by nylon into dolomite,

Pyrex into dolomite, and Pyrex into sand. In addition, the early-time downrange directed

175 vapor plume (related to the reverse winds, see Chapter 1 for more discussion) engulfed the target surface, in contrast to the plume generated by the silicate (Pyrex) impactor. As a result, the pipe cleaners positioned downrange experienced dramatic and more chaotic winds from the nylon impactor compared to the Pyrex impactor.

Far from the impact point (>15 cm), the vapor plume decelerated exponentially due to atmospheric drag (Chapter 1). With the assumption that the vapor plume expands as a hemisphere, it is therefore possible to estimate the amount of vapor produced in each case. From the drag equation, the mass of the vapor plume (mplume) is,

퐶퐷휌푎푡푚퐴푝푙푢푚푒푥 푚푝푙푢푚푒 = , (1) 2 ln(푣 ) ⁄푣표 where CD is the drag coefficient (assumed here to be 0.38), ρatm is the density of the

-5 -3 ambient atmosphere (3.90x10 g cm for air at 33 mbar), Aplume is the area of the plume

(measured in images), x is the selected distance over which the vapor plume decelerates, vo is the initial velocity of the plume, and v is the velocity of the plume measured at the distance x. Given a set of impact conditions (for example, 45°, ~5 km s-1, 33 mbar atmosphere of air), a nylon impact into dolomite produced the most vaporization, followed by Pyrex into dolomite, and nylon into sand. The derived vapor mass (from equation 1) was ~6, 4, and 2 times the projectile mass, respectively. The vapor generated by the Pyrex projectile result is consistent with previous measurements (Schultz, 1996).

Table 2 lists the parameters for each example case.

The passage of the vapor plume changed the upper texture of the dolomite target surface and is expressed by a slight darkening under certain illuminations (see Chapter 1 for more discussion). For the two impacts into dolomite (160909 and 160910), the

176 surface roughening resembled previous results, but it expanded several times faster downrange than uprange (see Table 3) due to the retained downrange impactor momentum. Downrange roughening, however, was difficult to discern because of camera frame oversaturation from the bright, luminescent vapor.

The impact into sand did not sustain surface roughening beyond an area near the impact point because of its larger grain sizes compared to dolomite. As a result, vapor expansion had to be determined from the leading edge of the plume for all other runs in this series. Because the vapor plume edge was diffuse and difficult to measure, this approach yielded larger measurement errors (Quintana and Schultz reported an error of

130 m s-1 for experiments in air, Chapter 1). Errors could be reduced if the target chamber lights were turned off. Chamber lights were kept on for all runs described here because simultaneous measurements of wind speed from pipe cleaners required the lights.

Even though the measurement uncertainty appears large, the data follow a consistent trend.

Vapor expansion speeds for the nylon projectiles are listed in Table 4, and Figure

4 plots the results (uprange values) for a) all at 45° impacts and b) all nylon/sand impacts.

The nylon impact into dolomite under a helium atmosphere generated the fastest vapor expansion speed (Figure 4a). This result is expected because 1) this impact produced the most vaporization, and 2) helium is less dense than air and allows for faster expansion.

The nylon/sand impact in air also produced a vapor plume that expanded quickly, especially at larger distances from the impact point. The speed of expansion was due to the greater amount of vaporization of the projectile that expanded above the surface, rather than large amounts entrained in the crater cavity before being released (Schultz and

177

Eberhardy, 2015). Other impacts at 45° were comparable. Figure 4b demonstrates a slight dependence on impact angle for vapor expansion uprange (the slowest is 30°), primarily at early times and distances close to the impact point.

Dust detached from the dusty pipe cleaners directly after the passage of the vapor plume and captured the initiation of impact winds. Frame-by-frame analysis of dust streamer edge movement established both the onset of winds and initial wind speeds

(Table 5). Wind speeds for the nylon projectiles were generally lower than those reported in Chapter 1. For example, a 45° impact through air in this study (nylon/sand) yielded peak impact wind speeds ~130 ± 30 m s-1 nearest to the impact (20 cm distance) and ~80 ± 30 m s-1 at a distance of 43 cm from the impact point. Comparatively, the same impact conditions with a Pyrex projectile (into dolomite) yielded ~200 ± 30 m s-1 and ~135 ± 30 m s-1, respectively. The Pyrex projectile may have incorporated some silicate vapor (generated during the jetting phase mixing with the atmosphere), which led to higher initial wind speeds. Nonetheless, the nylon impactor generated winds that were exceptionally more pervasive and turbulent than those from Pyrex/dolomite impacts.

Thus, although the initial wind speeds were lower for the nylon impactors (lower vaporization temperature), the greater amount of vapor resulted in widespread winds with greater surface interactions.

Impact angle also affected wind speeds. The fastest speeds downrange occurred at the lowest angle impacts (30°, a peak of ~210 m s-1 closest to the impact point), and decreased with increasing impact angle (up to 90°, a peak speed of ~110 m s-1). Uprange wind speeds were much slower but were difficult to measure for the 30° impact. At 45°, uprange wind speeds were fastest at a peak of ~130 ± 30 m s-1 and slowest (60 ± 30 m s-1)

178 at 60°. Winds appeared not to mobilize much material off pipe cleaners placed orthogonal to the impact direction, except for low-angle impacts where wind speeds ranged from 60 ± 25 m s-1 (30°) to 160 ± 25 m s-1 (45°). However, ejecta curtain-driven winds also interfered with some measurements, thereby making it difficult to separate ejecta-induced motion from the passage of vapor-driven winds.

For completeness, a Pyrex projectile impact into sand also provided a benchmark.

Such an impact should create little to no vapor at laboratory speeds (except for the early- time jetting). Although a clear vapor plume could not be identified, dust was still blown off the pipe cleaners downrange. Dust began to detach from the pipe cleaners in streamers ~100 μs earlier than any other case tested. High-speed imaging revealed, however, that high-speed ejecta (sand particles) passed by the pipe cleaners at about this time, and the resulting wake mobilized the dust. Later, dust was drawn into atmospheric vortices generated in front of the advancing ejecta curtain. Therefore, even though dust was mobilized off the downrange pipe cleaners, the absence of a vapor plume (and related winds) underscores the role of impact vaporization in generating impact-driven winds.

4. Topographic Case Studies

Experiments not only illustrate the role of impact vaporization but also prompt possible tests in order to isolate further the underlying processes on Mars. If, for example, the wind streaks result from basal flows moving across the surface rather than vapor- driven winds, then topographic changes should disrupt, if not stop, the advancing flow.

Among the 12 craters having impact-wind streaks on Mars, two examples formed near

179 sudden topographic changes and allow for further tests of the impact winds model: the craters Mojave and Jijiga (Figure 5).

The first example is Mojave crater (7.48°E, -32.99°N) within Tui Valles, one of the outflow channels terminating in Chryse Planitia. Mojave (Figure 5a and b) is a 58 km diameter crater that impacted just north of a relict island within the channel. The region exhibits very good nighttime thermal infrared contrast, resulting in clear expression of the radial bright and dark pattern around Mojave. The 31 impact-wind streaks around

Mojave radiate in all directions (some are indicated in Figure 5a by arrowheads), but most notably formed both on the plateau and on the floor of an unnamed eroded crater south of the plateau. A shadowing effect just south of the plateau manifests in a dark strip (~10-14 km wide) within the unnamed crater where wind streaks are not present.

But beyond this zone, several PE craters formed wind streaks. Hence, the inferred winds remained sustained even though they passed from the impact site on the valley floor to the plateau (over 2000 m in elevation change) and then scoured the eroded crater floor

(another 2000 m elevation change) to form wind streaks.

The second example is the 16 km diameter crater Jijiga (25.11°E, -53.95°N), which formed on a relict island near the end of Kasai Valles (Figure 5c and d). The thermal contrast surrounding the crater in THEMIS nighttime infrared is poorer than that around Mojave, but the radial bright and dark thermal pattern is especially clear on the channel floor. The pattern extends even over other islands in the channel and onto the bluffs and smooth plains to the south. A shadowing effect is apparent around the island on which Jijiga impacted, again indicated by the ~6 km wide dark band in the THEMIS nighttime infrared. Beyond this zone, however, five impact-wind streaks formed on the

180 channel floor. A few of these streaks are highlighted with arrowheads in Figure 5c and

Figure 6. Even though Jijiga is located on a local topographic high, winds scoured the channel surface more than 500 m below. Wind streaks exist on a slight (~100 m) rise and behind another plateau within the channel with respect to Jijiga (Figure 6).

4.1 Methods

With Jijiga as a model, the following section outlines a laboratory experiment and a series of CTH computational simulations that explore the effects of an abrupt change in topography on development and passage of winds. Laboratory experiments at the AVGR illustrate the effect of a vapor plume traversing topography (see Figure 7) and provide a physical basis for the additional computational simulations. These simulations allow for more detailed examination of the specific circumstances of the topography at Jijiga Crater.

Because the vast majority of natural impacts are non-vertical (Gilbert, 1893;

Shoemaker, 1962), a three-dimensional CTH (version 11.2) model allows simulating a variety of impacts on Mars at angels aside from 90°. Three-dimensional models break the axisymmetry of the less complicated two-dimensional models and therefore require greater computing power and run time. CTH includes an adaptive mesh refinement

(AMR, Crawford, 1999) scheme that allows the user to specify resolution where it is needed most, therefore reducing computation time. The analytical equation of state

(ANEOS) package, with updated molecular definitions (Melosh, 2007), defined the solid materials, while the atmosphere was modeled as an ideal gas. ANEOS is a common equation of state (EoS) used in the planetary community because it is relatively stable and defines geological materials well.

181

The CTH model incorporated a simplified recreation of the terrain around Jijiga crater (Figure 8) such that the impact centered on a plateau between two channels. The plateau stood at a height of 500 m from the channel floor on one side and 1 km on the other. Target material consisted of a basaltic crust with the Brittle Damage with

Localized Thermal Softening (BDL) strength and damage model (Crawford and Schultz,

2013). Environmental conditions mimicked averages of present-day Mars: 213.5 K for

-3 the surface temperature, 6.5 mbar CO2 (with a density of 1.55e-5 g cm ) for surface atmospheric pressure, and 3.71 m s-2 for gravitational acceleration. Finally, the impactor was a sphere of ~60% porous water ice, which impacted the surface at either 12 km s-2 or

20 km s-2 and an angle of 45°. Models of a dunite impactor into a solid basaltic crust and a basaltic crust overlain with water-ice allowed for further comparisons. Unless otherwise noted, the impactor was 1.5 km in diameter. In order to reduce computation time, the calculation began with the impactor at the target surface.

The models saved data regarding density, material boundaries, x-velocity, peak pressure, and temperature through time at intervals of 0.1 s. Seventy-two tracers recorded density and wind speed at fixed locations and at a temporal resolution of 5 ms.

These tracers were placed in the same locations as described in Chapter 2 (i.e., at 15, 25,

30, 40 km downrange; 15 km uprange; and 15 km orthogonally, with heights from 100 m to 15 km above the surface). While impact-wind streaks on Mars typically occur much farther from the impact crater, this tracer positioning allows for longer observation within a restricted timeframe. The same process is likely to continue farther out, as well.

4.2 Results

182

Data plots of density, material boundaries, and x-velocity, combined with tracer data, provided an overview of the impact process and allowed assessment of impact wind development. The results presented in Tables 6 and 7 indicate that any of the scenarios with volatiles may lead to the development of impact winds within the topographically low channel downrange of the impact point. While each scenario generated some impact-vapor, the asteroidal control case (a dunite impactor striking a basaltic target) did not induce high-density vaporization at the surface, and therefore is unlikely to develop strong impact winds.

Comet impactors, however, generated significant vapor that expanded into (and across) the topographic low of the channel. The initial impact winds developed within the channel exceeded 400 m s-1 (at a distance of 30 km from the impact point) for comet impactors 1.5 km in diameter impacting at both 12 km s-1 (Scenario 2) and 20 km s-1

(Scenario 3, Figure 9). Comparatively, a smaller impactor (Scenario 4) produced peak wind speeds in excess of 200 m s-1 (see Table 7). At later times, the winds decayed to a lower speed and reversed direction. For example, the 1.5 km diameter comet impacting at 12 km s-1 had an average wind speed of 140 m s-1. As a result, the later reverse winds could not significantly modify effects of the initial, outward flow. Regardless of wind direction, each comet impact scenario produced a pressure differential at the channel wall.

Surface ice models also developed high-density impact-vapor, which initiated outward-directed winds that extended into the topographic low at relatively early times.

For example, an extremely thick ice layer (Scenario 5, Figure 10) caused impact- vaporization that drove wind speeds exceeding 1 km s-1 (at a distance of 30 km from the impact point) for both impact speeds of 12 km s-1 and 20 km s-1. Only the thin-ice case

183 developed winds counter to observations (inward-directed), even though all other criteria for impact wind development were met. This scenario behaved similarly to the comet scenarios (and in particular, Scenario 4) except that it produced overall higher amounts of vaporization.

Table 7 reports near-surface winds for each scenario at selected tracers placed downrange, uprange, and orthogonal to the impact trajectory. Downrange, wind speeds directed outward are negative. Uprange and orthogonally, however, negative values correspond to inward-directed winds. A maximum inward-directed speed of zero indicates that winds did not blow inward, but it does not preclude the formation of vortices (as seen in Figure 10, for example).

5. Discussion

5.1 Comet Impacts

Cometary impacts provide the most consistent cause for impact-vaporization capable of driving impact winds since they would occur independent of geologic setting.

In experiments, nylon projectiles provided a simple proxy for volatile-rich comets. Such tests revealed that most of the vapor remains above the surface, in contrast with the silicate projectiles into dolomite that contain and redirect the vapor upward (away from the surface). The lower expansion speed from a nylon projectile reflects its lower vaporization temperature (relative to a Pyrex impactor into dolomite), but also results in a greater amount of vapor that interacts with the surface. Even then, the experiments documented wind speeds of 90-200 m s-1, levels that would still be considered severe tornadic winds on the Earth (Hyndman and Hyndman, 2016).

184

Similarly, an icy cometary impact would have lower vaporization temperatures, lower expansion speeds, and greater surface modification. CTH models, in fact, revealed that comet impactors resulted in a more ground-hugging vapor plume as compared to other cases (asteroid impacts into basalt or ice layers in the target) (Quintana and Schultz,

2014, 2017). Therefore, this low, early plume behavior can be expected on Mars and would cause preconditioning of the surface in advance of the impact winds by stripping away fines and the upper dust veneer.

5.2 Topographic Effects

The unusual topographic settings of Jijiga and Mojave craters allow tests of the role of other processes including base surge (Boyce and Mouginis-Mark, 2006; Boyce et al., 2015), gravity collapse of a buoyant plume (Boyce and Mouginis-Mark, 2006; Boyce et al., 2015), or flow separation (Barnouin-Jha and Schultz, 1996, 1998, Barnouin-Jha et al., 1999a, 1999b). Topographic changes would be expected to redirect or terminate basal or ejecta flows. At Jijiga crater, wind streaks exist within the low-standing channel surrounding the crater. Vapor expanded into the channel and drove winds outward in order to form wind streaks, in some cases despite the presence of other relict islands within the channel. At Mojave, wind streaks formed on top of a topographically higher plateau just to the south and on the floor of an unnamed crater that intersects the plateau.

As illustrated in laboratory experiments, an impact vapor plume expands downward (as well as upward) into topographic lows while rapidly moving across the surface. This process would allow wind streaks to form even where ejecta flows are unlikely (in these cases, crossing upward over on a plateau and then into an adjacent low). Computational

185 fluid dynamics codes would provide an additional approach to this study and will be considered in the future.

Finally, impacts may induce significant dust mobilization and entrainment, which are also not modeled with CTH. Previous studies (e.g., Adushkin and Nemchinov, 1994;

Nemtchinov et al., 2002) explored the effects of dust ejected in vertical impacts within the ejecta curtain. The impact winds mode would expect dust mobilization from winds due to vapor expansion and atmospheric coupling to occur at great distances from the impact point well before the arrival of ejecta. Furthermore, Nemtchinov et al. (2002) expected only a small vapor mass in their model, whereas the current contribution examines examples of enhanced impact vaporization cases on Mars. Nonetheless, wind speeds estimated from the CTH models presented here could easily mobilize particles exceeding centimeters in size, well above erodible grains (sized between 20 and 600 μm,

(Greeley et al., 1980, 1992; Nemtchinov et al., 2002)). Even though it is not modeled in

CTH, dust and sand entrainment in the high-speed winds expected from impact winds would then cause surface modification (e.g., sand blasting) similar to, but more intense and longer lasting than, natural wind erosion.

6. Conclusion

This contribution tested the hypothesis that impact-wind streaks could form from cometary impacts by comparing the response of a volatile-rich impactor and by assessing the effects of topography on disrupting wind streak formation around two craters on Mars.

Results from this study support cometary impacts as a potential driver of impact winds and wind streak formation. The following conclusions can be made from the laboratory experiments:

186

1) A nylon impactor (comet analog) produced an early-time vapor that engulfed

the surrounding surface and traveled close to the ground downrange, in

contrast with vapor generated primarily from the target.

2) Nylon impacts into dolomite produced the greatest vapor mass (scaled to

projectile mass), followed by Pyrex into dolomite and nylon into sand. This

result reflects the added contribution by the easily volatized impactor.

3) Wind initiation (visualized by dolomite dust detachment from dusty pipe

cleaners) directly followed the passage of the vapor plume, rather than the air

shock created at the moment of contact between projectile and target.

4) Although vapor-wind speeds for nylon projectiles were slower than those

produced by Pyrex impactors into dolomite (due to the lower vaporization

speed of nylon), the resulting winds were longer lasting.

A case study of two craters in unusual topographic settings assessed the effects of wind streak generation and tested other streak formation models. While other impact- wind streak craters occur in relatively smooth terrains, both Jijiga and Mojave introduce topographic obstacles on a scale unlike any other for this type of crater on Mars. Impact winds crossed over deep channels and high plateaus for Jijiga and Mojave yet still formed the wind streaks. This independence from topographic effects requires a wind-generating process driven from above, rather than flowing low across the surface. These case studies revealed the following:

1) Impact-wind streaks formed on the channel floor ~500 m below the

impact point for Jijiga Crater. Radial bright/dark thermal patterns clearly

187

extend onto the channel floor, are undeflected over other islands, and

extend onto surrounding (higher) terrain.

2) Mojave Crater exhibits bright impact-wind streaks in all directions across

breaks in topography without being affected by its location (a

topographically low channel adjacent to a high plateau). Impact-wind

streaks formed both on top of the plateau and across an eroded crater floor

below the plateau.

3) CTH models with a simplified Jijiga topographic profile reveal that any

scenario that includes volatiles (whether in the target or in the impactor)

could potentially develop impact winds within a channel below the impact

point.

4) Comet impactors developed longer lasting impact winds within the

channel with outward-directed winds exceeded 400 m s-1 (for 1.5 km

diameter ice impactors). Later, less-intense reverse winds (and

turbulence) developed, but they would not significantly modify effects of

the initial flow.

5) Basal ejecta flows should be deflected or stopped by obstacles, in contrast

with the impact-wind streaks observed around Jijiga and Mojave.

Acknowledgements

This material is based upon work supported by the National Science Foundation

Graduate Research Fellowship under Grant (DGE-1058262), the Mars Fundamental

Research Program Grant (NNX13AG43G), and a Graduate Research Fellowship from the

188

NASA Rhode Island Space Grant Consortium (NNX10AI95H). Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation, NASA, or the NASA Rhode Island Space Grant Consortium. The authors wish to thank AVGR crew (Don Bowling, Charles Cornelison, Alfredo Perez, Adam Parish, and Jon-Pierre

Wiens) for their hard work and for making these experiments possible. We also thank

David Crawford for this guidance and many helpful discussions regarding CTH modeling.

189

References

Adushkin, V.V., Nemchinov, I.V., 1994. Consequences of Impacts of Cosmic Bodies on the Surface of the Earth, in: Hazards Due to Comets and Asteroids. Presented at the Hazards Due to Comets and Asteroids.

Barnouin-Jha, O.S., Schultz, P.H., 1998. Lobateness of impact ejecta deposits from atmospheric interactions. Journal of Geophysical Research 103, 25–739. doi:10.1029/98JE02025.

Beyler, C.L., Hirschler, M.M., 2002. Thermal Decomposition of Polymers, in: SFPE Handbook of Fire Protection Engineering 2, Section 1, Chapter 7. pp. 111–131.

Boyce, J.M., Wilson, L., Barlow, N.G., 2015. Origin of the outer layer of martian low- aspect ratio layered ejecta craters. Icarus 245, 263–272. doi:10.1016/j.icarus.2014.07.032.

Bruck Syal, M., Schultz, P.H., 2014. Spatially-Resolved Spectroscopy of Impact- Generated Vapor Plumes. Lunar and Planetary Science Conference 45, Abstract 2760.

Crawford, D., 1999. Adaptive Mesh Refinement in CTH (No. SAND99–1118C). Sandia National Laboratories, Albuquerque, NM.

Meteorite Impacts and Planetary Evolution, USRA, Sudbury, Canada, Abstract 3047.

Gilbert, G.K., 1893. The Moon’s Face: A Study of the Origin of Its Features. Philosophical Society of Washington.

Glasstone, S., Dolan, P.J., 1977. The Effects of Nuclear Weapons. US Department of Defense and US Department of Energy.

Greeley, R., Lancaster, N., Lee, S., Thomas, P., 1992. Martian aeolian processes, sediments, and features, in: Mars. pp. 730–766.

Hartmann, W.K., 2005. Martian cratering 8: Isochron refinement and the chronology of Mars. Icarus 174, 294–320. doi:10.1016/j.icarus.2004.11.023.

Hyndman, Donald, Hyndman, David, 2016. Natural Hazards and Disasters. Cengage Learning.

Kreslavsky, M.A., Head, J.W., 2002. Mars: Nature and evolution of young latitude- dependent water-ice-rich mantle. Geophysical Research Letters 29, 14–1. doi:10.1029/2002GL015392.

McGlaun, J.M., Thompson, S.L., Elrick, M.G., 1990. CTH: A three-dimensional shock wave physics code. International Journal of Impact Engineering 10, 351–360. doi:10.1016/0734-743X(90)90071-3.

Melosh, H.J., 2007. A hydrocode equation of state for SiO2. Meteoritics and Planetary Science 42, 2079–2098. doi:10.1111/j.1945-5100.2007.tb01009.x.

Nemtchinov, I.V., Shuvalov, V.V., Greeley, R., 2002. Impact-mobilized dust in the Martian atmosphere. Journal of Geophysical Research 107, 5134. doi:10.1029/2001JE001834.

191

Quintana, S.N., Schultz, P.H., 2014. The Formation of Crater-Related Blast Wind Streaks on Mars. Lunar and Planetary Science Conference 45, Abstract 1971.

Quintana, S.N., Schultz, P.H., 2016. A Global Distribution of Impact-Wind Streak Craters on Mars. Lunar and Planetary Science Conference 47, Abstract 1548.

Quintana, S.N., Schultz, P.H., 2017. Model Results for Impact-Winds on Mars. Lunar and Planetary Science Conference 48, Abstract 1123.

Quintana, S.N., Schultz, P.H., Horowitz, S.S., 2016. New Experiments in Martian Impact Vapor-Induced Wind Streak Analysis. Lunar and Planetary Science Conference 47, Abstract 1553.

Schultz, P.H., 1988. Impact vaporization of volatile-rich targets; experimental results and implications, Lunar and Planetary Science Conference 19, Abstract 1039.

Schultz, P.H., 1992. Atmospheric effects on ejecta emplacement and crater formation on Venus from Magellan. Journal of Geophysical Research: Planets 97, 16183– 16248. doi:10.1029/92JE01508.

Schultz, P.H., 1996. Effect of impact angle on vaporization. Journal of Geophysical Research: Planets 101, 21117–21136. doi:10.1029/96JE02266.

Schultz, P.H., Eberhardy, C.A., 2015. Spectral probing of impact-generated vapor in laboratory experiments. Icarus 248, 448–462. doi:10.1016/j.icarus.2014.10.041

Schultz, P.H., Quintana, S.N., 2017. Impact-generated winds on Mars. Icarus 292, 86– 101. doi:10.1016/j.icarus.2017.03.029.

Schultz, P.H., Wrobel, K.E., 2012. The oblique impact Hale and its consequences on Mars. Journal of Geophysical Research 117, E04001. doi:10.1029/2011JE003843.

Shoemaker, E.M., 1962. Exploration of the Moon’s Surface. American Scientist 50, 99– 130.

192

Tables

Table 1 - List of experimental runs

Impact Atmospheric Impact Projectile Run Speed Target Type Pressure Angle Type (km/s) (mbar) 160902 5.2 45 Dolomite Pyrex 32.72 160903 5.3 45 20-30 Sand Nylon 33.33 160904 4.79 45 20-30 Sand Pyrex 33.17 160909 4.93 45 Dolomite Nylon 33.58 160910 5.08 45 Dolomite Nylon 33.80 160911 5.6 30 20-30 Sand Nylon 33.64 160912 4.96 60 20-30 Sand Nylon 33.20 160913 5.33 90 20-30 Sand Nylon 33.29 160914 5.51 45 20-30 Sand Nylon 33.01 160915 5.64 45 20-30 Sand Nylon 66.43

193

Table 2 - List of parameters (from Chapter 1) used to estimate vapor plume mass with equation (1)

Uprange Vapor Expansion m v v x m 풎풑풍풖풎풆 Run Description proj o plume (g) (m/s) (m/s) (cm) (g) 풎풑풓풐풋풆풄풕풊풍풆 160902 Pyrex/Dolomite 0.2978 900 680 24 1.15 3.9 160909 Nylon/Dolomite 0.1557 960 790 20 0.96 6.2 160914 Nylon/Sand 0.155 800 550 16 0.25 1.6

194

Table 3 - Surface roughening results for impacts into dolomite (under an atmosphere of 33 mbars air, unless otherwise noted)

Uprange Downrange Distance Distance Impact Overall Overall Impact from from Run Speed Speed Speed Angle Impact Impact (km/s) (m/s) (m/s) (cm) (cm) 5 740 11 1540 8 730 15 1400 10 710 18 1230 12 690 20 1120 160902a 5.2 45 14 670 23 1070 (Pyrex/Dolomite) 16 660 25 1010 19 660 27 950 21 650 29 910 23 650 32 890 9 610 7 2020 11 640 20 2820 160909a 4.93 45 14 650 28 2630 16 660 - - 19 650 - - 3 590 7 2000 160910b 4 490 21 2890 5.08 45 (He atmosphere) 6 450 28 2640 8 420 - - a -1 60 Typical measurement error (in m s ) for runs in air: downrange ±40 and uprange ±30 b -1 600 60 Typical measurement error (in m s ) for runs in helium: downrange ±230 and uprange ±50

195

Table 4 - Vapor plume expansion speeds under 33 mbar of air (unless otherwise noted)

Uprange Distance Impact Overall Impact from Run Speed Speeda Angle Impact (km/s) (m/s) (cm) 6 900 8 860 10 800 12 730 160902 5.2 45 13 680 (Pyrex/Dolomite) 15 670 18 680 20 680 22 680 8 1180 9 960 10 810 12 720 160903 5.3 45 13 650 14 620 16 610 17 590 19 570 160904 4.79 45 No Clear Vapor (Pyrex/Sand) 6 960 9 910 11 870 160909 4.93 45 13 820 15 790 18 790 6 1550 10 1200 160910 5.08 45 13 1020 (He atmosphere) 16 980 19 910 Note: Table 4 continued on next page

196

Table 4 continued

Distance Impact Overall Impact from Run Speed Speeda Angle Impact (km/s) (m/s) (cm) 4 670 7 680 160911 5.6 30 8 660 10 640 12 780 6 850 7 750 160912 4.96 60 9 700 11 680 13 650 9 950 10 1080 160913 5.33 90 12 720 13 650 14 630 8 780 9 700 11 680 160914 5.51 45 12 610 13 590 14 550 6 920 8 790 160915 10 740 (67 mbar 5.64 45 11 670 atmosphere) 12 610 13 570 a -1 130 90 Typical measurement error (in m s ) for air: ±50 ; and for helium: ±80

197

110

(m/s)

Winds

Reverse

100

160

(m/s)

Winds

Outward

ORTHOGONAL

(cm)

from

Impact

Distance

100

110

130

120

100

(m/s)

Winds

Reverse

120

160

130

150

(m/s)

Winds

Outward

UPRANGE

(cm)

from

Impact

Distance

frame analysis of dust detachment from dustycleaners pipefor frame analysis of dust detachment

110

130

100

140

120

100

120

(m/s)

Winds

Reverse

170

110

170

130

100

110

100

150

180

100

130

120

(m/s)

Winds

Outward

DOWNRANGE

(cm)

from

Impact

Distance

5.3

5.08

4.93

4.79

Speed

(km/s)

Impact

Initial wind speeds as measured by frame as measured Initial wind speeds

Angle

160910

160909

160904

160903

Shot

Note: Table 5 continuedon next page.

nylon impactors (unless noted (unlessotherwise). nylon impactors Table 5 Table

198

110

(m/s)

Winds

Reverse

(m/s)

Winds

Outward

ORTHOGONAL

(cm)

from

Impact

Distance

210

260

160

210

(m/s)

Winds

Reverse

130

110

(m/s)

Winds

Outward

UPRANGE

(cm)

from

Impact

Distance

100

120

110

130

110

130

110

(m/s)

Winds

Reverse

120

100

120

130

100

110

130

100

210

(m/s)

Winds

Outward

DOWNRANGE

(cm)

from

Impact

Distance

5.6

5.51

5.33

4.96

Speed

(km/s)

Impact

Angle

160914

160913

160912

160911

Shot

Note: Table 5 continued on next page on next 5 continued Note: Table Table 5 continuedTable

199

(m/s)

Winds

Reverse

(m/s)

Winds

Outward

ORTHOGONAL

(cm)

from

Impact

Distance

160

140

160

(m/s)

Winds

Reverse

100

(m/s)

Winds

Outward

UPRANGE

(cm)

from

Impact

Distance

100

(m/s)

Winds

Reverse

) uprange: ±30 m/s ) uprange:

) orthogonal to impact trajectory: ±25 m/s trajectory: impact ) to orthogonal

) downrange: ±30 m/s ±30 ) downrange:

180

230

180

210

(m/s)

Winds

Outward

DOWNRANGE

(cm)

from

Impact

Distance

30 sand

5.64

Speed

(km/s)

Impact

Angle

160915

Shot

Typical measurement error in wind speed (in m s m (in speed wind error in measurement Typical

Impact of Pyrex into 20 of into Pyrex Impact

a Table 5 continuedTable

200





Low

Vapor Vapor

Extends into

Topographic Topographic





High

Vapor Vapor

Density

Surface

Type

Water

Vapor Vapor

Silicate

Primary

1) Water; 1) Water;

2) Silicate

generated winds generated



Generated Wind Development



Vapor Vapor

Generated





Winds

Directed

Outward





High

Wind

Speeds

Criteria Vapor Supporting

Surface

Sustained





m/s)

(>70

High

Wind

Speeds

Surface

(km/s)

Speed

Impact

Model results and scenario likelihood of producing impact vapor producing impact likelihood and scenario of results Model

Basalt

(Dunite)

diameter) diameter)

100 m ice 100 m

500 m ice 500 m

(Ice, 1 km (Ice, 1 km

overlain by

(Ice) Basalt

Type

Target

(Impactor)

Scenario

Table 6 Table

201

Table 7 - Near-Surface (590 m altitude) wind speeds at various distances from the impact point for each model scenario

Sensor Location Downrange Lateral Uprange Distance from Impact 30 km 40 km 15 km 15 km Scenario Asteroid Impact into Basalt at 20 km/s (Scenario 1) Max outward windspeed -1650 -1560 -160 130 Max inward windspeed 60 440 730 -300 Average windspeed -550 -300 180 -50 Comet Impact at 12 km/s (Scenario 2) Max outward windspeed -510 -3000 -180 0 Max inward windspeed 1260 2010 240 -240 Average windspeed 140 360 -40 -120 Comet Impact at 20 km/s (Scenario 3) Max outward windspeed -440 -860 -140 520 Max inward windspeed 320 890 190 -250 Average windspeed 0 70 -30 -30 Small (1 km diameter) Comet Impact at 20 km/s (Scenario 4) Max outward windspeed -210 -920 -160 80 Max inward windspeed 600 1160 80 -130 Average windspeed 200 350 -50 -50 Asteroid Impact into 500 m-Thick Ice Layer at Surface at 12 km/s (Scenario 5) Max outward windspeed -1660 -1450 -180 1100 Max inward windspeed 40 0 340 0 Average windspeed -690 -310 10 260 Asteroid Impact into 500 m-Thick Ice Layer at Surface at 20 km/s (Scenario 6) Max outward windspeed -960 -1250 -200 430 Max inward windspeed 420 30 440 -40 Average windspeed -50 -280 -10 110 Asteroid Impact into 100 m-Thick Ice Layer at Surface at 20 km/s (Scenario 7) Max outward windspeed -230 0 -140 110 Max inward windspeed 580 800 280 -340 Average windspeed 210 340 -10 -100

202

Figure Captions

Figure 1 - Two bright impact-wind streaks southeast of Mojave Crater are captured by the Mars Reconnaissance Orbiter Context Camera (CTX) and display ridges downrange that correspond to the bright streaks in the THEMIS nighttime infrared image. The

MOLA color topography image (a) provides context. The box corresponds to the area in

(b), which is a THEMIS nighttime infrared image mosaic of the southeast region of

Mojave Crater (in upper left). The box in (b) corresponds to the area in (c), which is a

CTX image (B02_010566_1874_XI_07N031W) of the two craters with arrows indicating the ridges that correspond to the thermally bright streaks in (b).

Figure 2 - AVGR target and camera setup. Insets show an oblique view of the pipe cleaner placement in (light grey) and around (dark grey) the target. A standard arrangement (a) was used for the uprange and lateral pipe cleaners. Downrange (b), the first pipe cleaner (closest to the impact) was cut in half in order to avoid being hit with high-speed debris. Downrange pipe cleaners were positioned ~40° off-center (15° farther than shown) in order to avoid debris for 30° impact angles. An additional three cameras

(a Phantom V12 color, V12 black and white, and a Shimadzu imaging camera) were placed above the target to view the impact.

Figure 3 - A comparison of the vaporization produced from the impact of (a) a Pyrex projectile into dolomite at 5.2 km s-1; (b) a nylon projectile into dolomite at 4.9 km s-1; (c) a Pyrex projectile into sand at 4.8 km s-1; and (d) a nylon projectile into sand at 5.5 km s-1.

All frames are ~250 μs after a 45° impact through 33 mbar air. The initial, luminous vapor plume from the nylon projectile can clearly be identified in (b) and (d). This plume

203 also travels much closer to the ground than the plume in (a). In (c), no vapor plume is visible. Note that some powdered dolomite is mobilized from pipe cleaners by high- speed ejecta (see text for details).

Figure 4 - Uprange vapor expansion speeds (in 33 mbar of air) for (a) all experimental runs at 45° and (b) nylon/sand impacts at different angles, plotted to the same scale. Both plots are for uprange expansion speeds. The scatter in (a) is likely due to the many different impact conditions included (Pyrex/dolomite, nylon/dolomite, and nylon/sand).

See Table for more information about each experimental run. The results in (b) indicate the effects of impact angle, with lower angles resulting in slower expansion speeds uprange. Trend lines are only to guide the eye through the data. The error bars reflect uncertainty in the measurement of the plume edge. The edge was diffuse and difficult to measure, especially with the target changer lights on. Surface roughening (a more accurate measurement, see Chapter 1 and Run 160902) was only observed in the

Pyrex/dolomite case.

Figure 5 - Panels on the top correspond to Mojave Crater and those on the bottom correspond to Jijiga. (a) THEMIS nighttime infrared image mosaic of Mojave crater.

Yellow arrowheads denote PE craters with wind streaks. (b) Mars Orbiter Laser

Altimeter (MOLA) topography of the profile A-A’ in (a). (c) THEMIS nighttime infrared images of Jijiga crater with PE craters noted. (d) MOLA profile B-B’ from (c).

Figure 6 - Additional Profiles from Jijiga crater (black arrow) demonstrate that the process that formed the impact-wind streaks from PE craters (yellow arrowheads) navigated different obstacles. In Profile A-A’ (top panel), the slope of the channel floor

204 increases by ~100 m from the floor to near the PE crater. In Profile B-B’ (bottom panel), another plateau stands between Jijiga crater and the PE crater. Basal ejecta flows should be deflected or stopped by such obstacles.

Figure 7 - An example of free expansion of impact-vapor. Self-luminescent vapor expands from the impact point downrange in all directions, including into the well between the target container and the test chamber platform (arrow). The well is outlined with white dotted lines, and the yellow dashed line outlines the path of the vapor. The

Shimadzu imaging camera frame is ~100 μs after the impact of a Pyrex projectile into powdered dolomite at 30° into a vacuum.

Figure 8 - A two dimensional cross-section of a three dimensional CTH model demonstrates the simplified recreation of the terrain around Jijiga Crater. In order to reduce computation time, the calculation begins with the impactor already at the target surface. The model testes included a basaltic crust and a dunite impactor (shown here) in addition to tests of an impactor composed of porous ice (comet). Other scenarios tested layers of ice at the surface, as well as the effects of a dunite (asteroid) impactor. Impact speeds ranged from 12 to 20 km s-1. Box denotes the area shown in Figures 9 and 10.

Figure 9 - The effect of topography documented through a sequence of CTH x-velocity data plots from Scenario 3 (an ice impactor striking a basaltic target at 45° and 20 km s-1).

The colors correspond to the velocity scale on the top right. Warm colors indicate vapor motion to the right and cool colors indicate motion to the left. Vapor extends into the channel (arrow) at 1.6 s. When the vapor plume reaches the edge of the channel at 3.2 s, a vortex (curved arrow) develops and travels back through the channel. Maximum

205 outward wind speeds exceed 800 m s-1. The sustained (average) wind speed is directed inward, but the speed of 70 m s-1 would not alter effects from the initial outward winds.

Images denote 2D slices of the 3D model.

Figure 10 - Same as Figure 9 except for Scenario 5 (a dunite impactor striking a basaltic target overlain by 500 m of ice at 45° and 12 km s-1). Vapor enters the channel (arrow) at

3.8 s and travels along the channel floor to the end of the channel. Vortices (curved arrows) result after passage of the vapor plume and when the plume reaches the end of the channel. More vapor (open arrow) reaches the channel floor at 15 s and expands outward, overpowering the vortices by ~20 s. Both the maximum and sustained winds are directed outward and exceed 300 m s-1.

206

Figures

Figure 1

207

Figure 2

208

Figure 3

209

Figure 4

210

Figure 5

211

Figure 6

212

Figure 7

213

Figure 8

214

Figure 9

215

Figure 10

216