Understanding the Threshold of Detection of Aeolian Features in Magellan Synthetic Aperture Radar Data Using Venusian and Terrestrial Dune Fields

Home , Pollack (Martian crater), Simpson Desert

University of Tennessee, Knoxville TRACE: Tennessee Research and Creative Exchange

Masters Theses Graduate School

8-2020

Louisa Xian-Yue Rader University of Tennessee, [email protected]

Follow this and additional works at: https://trace.tennessee.edu/utk_gradthes

Recommended Citation Rader, Louisa Xian-Yue, "Understanding the Threshold of Detection of Aeolian Features in Magellan Synthetic Aperture Radar Data Using Venusian and Terrestrial Dune Fields. " Master's Thesis, University of Tennessee, 2020. https://trace.tennessee.edu/utk_gradthes/6093

This Thesis is brought to you for free and open access by the Graduate School at TRACE: Tennessee Research and Creative Exchange. It has been accepted for inclusion in Masters Theses by an authorized administrator of TRACE: Tennessee Research and Creative Exchange. For more information, please contact [email protected]. To the Graduate Council:

I am submitting herewith a thesis written by Louisa Xian-Yue Rader entitled "Understanding the Threshold of Detection of Aeolian Features in Magellan Synthetic Aperture Radar Data Using Venusian and Terrestrial Dune Fields." I have examined the final electronic copy of this thesis for form and content and recommend that it be accepted in partial fulfillment of the equirr ements for the degree of Master of Science, with a major in Geology.

Bradley J. Thompson, Major Professor

We have read this thesis and recommend its acceptance:

Molly C. McCanta, Christopher M. Fedo

Accepted for the Council: Dixie L. Thompson

Vice Provost and Dean of the Graduate School

(Original signatures are on file with official studentecor r ds.)

Understanding the Threshold of Detection of Aeolian

Features in Magellan Synthetic Aperture Radar Data Using

Venusian and Terrestrial Dune Fields

A Thesis Presented for the

Master of Science

Degree

The University of Tennessee, Knoxville

Louisa Xian-Yue Rader

August 2020

ACKNOWLEDGEMENTS

Thank you to my advisor, Brad Thomson, for the opportunity to pursue my graduate degree and support through this journey. Without all your feedback and encouragement, I would not have been able to accomplish this. I would also like to thank Molly McCanta and Chris Fedo for providing advice and edits on my thesis.

I would like to thank CJ Leight for always being there to make me smile and encouraging me every step of the way. I could not have done this without the support from Megan Mouser,

Laura Breitenfeld, and my office desk pod buddy, Cole Nypaver. They have all cheered me on and provided moral support when I needed it. Finally, thank you to my parents for always supporting and believing in me. They have always encouraged my passions and helped me achieve more than I could ever imagine on my own.

iii

ABSTRACT

The spatial resolution required for the detection of geomorphologic features on planetary surfaces is critical to developing a reproducible process for surface feature mapping and identification. The Magellan synthetic aperture radar (SAR) data of Venus have a spatial resolution of 75 m/pixel, which obfuscates features below about a kilometer in scale, while allowing analysis of larger-scale features (Saunders et al., 1991; Rader et al., 2019). Smaller- scale features, such as aeolian bedforms, are important for understanding erosional processes and the redistribution of sediment on the surface. (Rader et al., 2019). In order to properly detect and describe aeolian features on the surface of Venus and get a handle on their implications (e.g., for long-term wind patterns), we must understand the unique, diagnostic characteristics of aeolian features in low resolution SAR images and how these Magellan images can be interpreted (Rader et al., 2019). Here we show how to identify the diagnostic characteristics of dune fields in both visible and radar wavelength images and quantify the differences between them, using Earth as an analog for Venus. Aeolian features of the scale observed on Venus are also found in hyper- arid desert regions on Earth, with this study focusing on the Simpson Desert in Australia and the

Namib Desert in Namibia (e.g., Bristow et al., 2007; Lancaster, 2009). The crestlines of linear dunes at these two terrestrial sites were mapped in optical (VIS) images at high resolution and in both high and low resolution SAR images as analogs for the venusian fields. The five venusian localities selected had a representative population of features mapped and then compared to the terrestrial dune fields (Greeley et al., 1992; Weitz et al., 1994; Rader et al., 2019). The mapped images were compared to identify diagnostic characteristics that can be used to identify other potential aeolian fields. These results demonstrate how dune fields can be identified on Venus and suggest possible reasons for the scarcity of potential venusian aeolian fields.

TABLE OF CONTENTS

1. Introduction ...... 1

2. Background ...... 5

2.1 Synthetic Aperture Radar ...... 5

2.2 Magellan Orbiter ...... 6

2.3 Sentinel-1 ...... 7

2.4 Optical Imagery ...... 8

2.5 World Imagery Layer ...... 8

2.6 Jet Propulsion Laboratory Topography Data ...... 9

2.7 Aeolian Features ...... 9

2.7.1 Dunes ...... 10

2.7.1.1 Linear Dunes ...... 11

2.7.1.2 Transverse Dunes ...... 12

2.7.2 Yardangs ...... 12

2.7.3 Simpson Desert ...... 14

2.7.4 Namib Desert ...... 15

2.7.5 Venusian Features and Working Hypothesis...... 15

2.8 Identification Resolution ...... 17

3. Methods and Data ...... 19

3.1 Terrestrial SAR Data Processing...... 19

3.2 Terrestrial ArcGIS Mapping ...... 21

3.3 Magellan SAR Data Processing ...... 22

3.4 Magellan ArcGIS Mapping ...... 23

3.5 Venusian Topographic ArcGIS Mapping ...... 24

4. Results ...... 26

4.1 Simpson Desert ...... 26

4.1.1 VIS at 0.5 m/pixel ...... 26

4.1.2 SAR at 13.5 m/pixel ...... 27

4.1.3 SAR at 75 m/pixel ...... 28

4.2 Namib Desert...... 29

4.2.1 Linear Dunes in VIS at 0.5 m/pixel ...... 29

4.2.2 Linear Dunes in SAR at 13.5 m/pixel ...... 30

4.2.3 Linear Dunes in SAR at 75 m/pixel ...... 31

4.2.4 Transverse Dunes in VIS at 0.5 m/pixel ...... 32

4.2.5 Transverse Dunes in SAR at 13.5 m/pixel ...... 32

4.2.6 Transverse Dunes in SAR at 75 m/pixel ...... 33

4.3 Venusian Localities ...... 34

4.3.1 Aglaonice Feature Field...... 34

4.3.2 Fortuna-Meskhent Feature Field ...... 35

4.3.3 Guan Daosheng Feature Field ...... 35

4.3.4 Stowe Crater Feature Field ...... 36

4.3.5 Mead Crater Feature Field ...... 36

4.3.5.1 Magellan SAR at 75 m/pixel...... 36

4.3.5.2 Jet Propulsion Laboratory Topographic Data ...... 37

5. Discussion ...... 38

5.1 Interpretation of Results ...... 38

5.1.1 Simpson Desert ...... 38

5.1.2 Namib Desert ...... 38

5.1.3 Venusian Feature Fields ...... 40

5.1.4 Dune Defect Density in Aeolian Fields ...... 41

5.1.5 Venusian Feature Orientation ...... 42

5.2 Diagnostic Characteristics of Linear Features Observed in Synthetic Aperture Radar

Images ...... 44

5.3 Implications for Dune Identification ...... 44

5.3.1 Identifying Dune Fields in Synthetic Aperture Radar Images ...... 44

5.3.2 Future Identification of Venusian Dune Fields in Magellan SAR Images ...... 45

6. Summary ...... 47

LIST OF REFERENCES ...... 49

APPENDIX ...... 58

VITA ...... 122

vii

LIST OF FIGURES

Figure 2.1. 1 A schematic diagram of synthetic aperture radar (SAR) instrument...... 59

Figure 2.7.1. 1 A diagram of different named dune types...... 60

Figure 2.7.2. 1 A yardang field in the Kumtag Desert...... 61

Figure 2.7.3. 1 An outline of the Simpson Desert...... 62

Figure 2.7.4. 1 The Namib Desert...... 63

Figure 2.7.5. 1 Theoretical and experimental data for sediment transport on Venus...... 64

Figure 2.8. 1 A graph describing feature identification resolutions...... 65

Figure 3.1. 1 The Sentinel-1B data download software...... 66

Figure 3.1. 2 End tag of Sentinel-1B SAR images used...... 67

Figure 3.2. 1 An example of the mapping the 13.5 m/pixel SAR Sentinel-1B imagery...... 68

Figure 3.2. 2 Terrestrial feature field characteristics...... 69

Figure 3.3. 1 A map of potential aeolian fields on Venus...... 70

Figure 3.3. 2 Venusian feature field characteristics...... 71

Figure 3.4. 1 An example of mapping multiple field extents in SAR imagery...... 72

Figure 3.5. 1 The topographic data from the Jet Propulsion Laboratory...... 73

Figure 4.1.1. 1 Bifurcation of the Simpson Desert in VIS at 0.5 m/pixel...... 74

Figure 4.1.1. 2 Area of the Simpson Desert in VIS...... 75

Figure 4.1.1. 3 Area of Simpson Desert mapped in VIS...... 76

Figure 4.1.2. 1 Area of the Simpson Desert mapped in high resolution SAR...... 78

Figure 4.1.2. 2 Area of the Simpson Desert in high resolution SAR...... 79

Figure 4.1.3. 1 Area of the Simpson Desert in low resolution SAR...... 81

Figure 4.1.3. 2 Area of the Simpson Desert mapped in low resolution SAR...... 82

Figure 4.2.1. 1 Area of the Namib Desert in VIS...... 84

viii

Figure 4.2.1. 2 Area of the Simpson Desert mapped in VIS...... 85

Figure 4.2.2. 1 Area of the Namib Desert in high resolution SAR...... 87

Figure 4.2.2. 2 Area of the Namib Desert mapped in high resolution SAR...... 88

Figure 4.2.3. 1 Area of the Namib Desert in low resolution SAR...... 90

Figure 4.2.3. 2 Area of the Namib Desert mapped in low resolution SAR...... 91

Figure 4.2.4. 1 Coastal area of the Namib Desert in VIS...... 93

Figure 4.2.4. 2 Coastal area of the Namib Desert mapped in VIS...... 94

Figure 4.2.5. 1 Coastal area of the Namib Desert in high resolution SAR...... 95

Figure 4.2.5. 2 Coastal area of the Namib Desert mapped in high resolution SAR...... 96

Figure 4.2.6. 1 Coastal area of the Namib Desert in low resolution SAR...... 97

Figure 4.2.6. 2 Coastal area of the Namib Desert mapped in low resolution SAR...... 98

Figure 4.3.1. 1 Aglaonice field mapped in Magellan SAR...... 99

Figure 4.3.2. 1 Fortuna-Meskhent field mapped in Magellan SAR...... 100

Figure 4.3.3. 1 Guan Daosheng field mapped in Magellan SAR...... 102

Figure 4.3.4. 1 Stowe field in Magellan SAR...... 103

Figure 4.3.5.1. 1 Mead field mapped in Magellan SAR...... 104

Figure 4.3.5.2. 1 Mead field mapped in JPL topography...... 105

Figure 5.1.1. 1 Comparison of resolutions in the Simpson Desert...... 106

Figure 5.1.1. 2 Comparisons of resolutions of bifurcation in the Simpson Desert...... 107

Figure 5.1.2. 1 Comparison of resolutions in the Namib Desert...... 108

Figure 5.1.2. 2 Comparison of dune defects in the Namib Desert...... 109

Figure 5.1.2. 3 Comparison of zoomed in coastal area of Namib Desert...... 110

Figure 5.1.2. 4 Comparison of the coastal area of the Namib Desert...... 111

Figure 5.1.3. 1 Features in the Aglaonice field on Venus...... 112

Figure 5.1.3. 2 Features in the Fortuna-Meskhent field on Venus...... 113

Figure 5.1.3. 3 Features in the Mead field on Venus with an elevation profile...... 114

Figure 5.1.5. 1 Rose diagram of Mead field...... 116

Figure 5.1.5. 2 Rose diagram of Aglaonice field...... 117

Figure 5.1.5. 3 Rose diagram of Fortuna-Meskhent field...... 118

Figure 5.1.4. 1 A plot of dune defect density vs. dune spacing...... 115

Figure 5.2. 1 A truth table of linear feature characteristics observed in multiple resolutions and wavelengths...... 119

Figure 5.3.1. 1 Different areas in the Namib Desert in SAR showing radar backscatter differences...... 120

Figure 5.3.2. 1 A map of Venus comparing localities to ejecta parabolas...... 121

1. Introduction

Observations of the geomorphology of a planet’s surface are a critical component for understanding its geologic history and inferring the nature of the processes that have shaped its surface. On Venus, this task is hampered by an optically impenetrable layer of planet-encircling clouds. These thick, dense clouds composed primarily of CO2 and N2 can be divided into a fast- moving upper atmosphere and a slow, hot lower atmosphere (e.g., Zolotov, 2018). The thick atmosphere limits the scientific instruments that can be used in missions to Venus but makes it an ideal target for radar observations.

Synthetic aperture radar (SAR) is a remote sensing technique that utilizes the microwave region of the electromagnetic spectrum. SAR data of surfaces are often gathered by orbiters that are using radar wavelengths to image a surface, because the technique takes advantage of the motion of the spacecraft to mimic a larger aperture than the physical instrument possesses

(Bamler and Hartl, 1998; Moreira, 2013). This design saves space and mass, which can particularly help with extraterrestrial missions that are designed to be as lightweight as possible to reduce the amount of fuel needed to leave Earth’s orbit (e.g., Bamler and Hartl, 1998). This instrumentation design was first used to image the surface of Venus with the Venera 15 and

Venera 16 probes sent by the Soviet Union in 1983 and 1984 respectively (e.g., Ivanov, 1988;

Kwok and Johnson, 1989).

In 1989 NASA launched the Magellan mission to Venus with the intent to map the surface of Venus using SAR with a 12.6 cm S-band wavelength radar (Ford et al., 1989;

Johnson, 1991; Arvidson et al., 1992). This mission produced the highest resolution global mosaic of the planet to date, generating coverage of 98 percent of the venusian surface from three mapping cycles (Ford et al., 1989; Johnson, 1991; Arvidson et al., 1992). The Magellan

1 synthetic aperture radar data have a spatial resolution of 75 m/pixel (Ford et al., 1989; Saunders et al., 1990; Johnson, 1991; Arvidson et al., 1992). The spatial resolution of these images can easily show features on the scale of tens of kilometers, though they can obfuscate features below a kilometer in scale.

Features, such as aeolian bedforms, that can reach up to a few kilometers in size on

Venus are just at the threshold of detectability in the Magellan SAR images. Geomorphological features of this scale include any simple craters, lava graben, and aeolian bedforms. Aeolian features can be used to inform both global and regional wind patterns and sediment transport models from current and paleo regimes. Observations of geomorphologic features created from the motion of sediment are important to understanding the surface-atmosphere interactions.

The term “aeolian features” refers to the erosion or deposition of sediment due to surface winds. Aeolian processes that create these features include entrainment, transport, and deposition of sand-sized grains (e.g., Lancaster, 2009; Nichols, 2009). Sediment-carrying winds have been proven to exist on Venus based on grain transport models (Iversen and White., 1982) and the surface wind speed measurements by the Venera missions, (e.g., Avduevskii et al., 1976).

Potential aeolian features have been observed as fields on Venus in several locations near clusters of impact craters in the synthetic aperture radar images (Greeley and Arvidson, 1990;

Greeley et al., 1992; Weitz et al., 1994). These observed locations can be further classified as mega-dune fields and yardang fields (Greeley et al., 1992; Weitz et al., 1994). These observed feature fields are believed to be dunes due to speckle pattern in the radar that was similar to other dunes fields observed in radar (Arvidson et al., 1991). Improved characterization of these aeolian fields could provide information on various local wind characteristics, the transport processes of sediment, and the relative quantity of available sediment on Venus.

The appearance of aeolian fields in remote imagery can be distorted in high resolution imagery; further observation of how features change at different wavelengths. Observations of the potential aeolian bedforms and quantitative characteristics of the sediment transport systems require high resolution observations of the different feature fields to accurately map the features and examine the smaller components of dunes. The dune-like features that have been observed on Venus are on the scale of kilometers, with a maximum width of 0.5 to 1 km (Greeley et al.,

1992; Weitz et al., 1994; Rader et al., 2019). With the Magellan data, these features appear as moderately radar-bright pixelated streaks; this resolution level (75 m/pixel) masks details of the bedforms, obscuring potential ripples or angular differences (Arvidson et al., 1991; Greeley et al., 1992; Weitz et al., 1994; Rader et al., 2019).

Aeolian features of the scale observed on Venus are also found on Earth in hyper-arid deserts. In particular, the Namib Desert on the western coast of Namibia and the Simpson Desert in central Australia both are large ergs with sinuous, linear dunes. These dune fields have both active and inactive regions, with complex, smaller features associated with them (e.g., ripples, bifurcation). Both of these dune fields have been imaged with synthetic aperture radar from the

European Space Agency’s Sentinel-1B mission and optical satellite Landsat 8 imagery from

NASA. The Sentinel-1B radar data have a spatial resolution of 5 m/pixel, and the Landsat 8 images have an average spatial resolution of 3.5 m/pixel. While terrestrial dune fields are not perfect analogs for venusian dune fields, they can be used to track and describe the way different dune characteristics manifest in various wavelength imageries.

These terrestrial analogs are used to quantify the threshold of detection of comparably sized dune fields by downsampling the data to 75 m/pixel and mapping features at both full and reduced resolutions. Here I show how two terrestrial analogs, Simpson Desert and Namib Desert,

3 can be used to observe how the appearance of dune crestlines changes in synthetic aperture radar and at various spatial resolutions. The terrestrial localities with known dune fields can be compared to previously identified potential venusian dune fields at 75 m/pixel resolution SAR imagery. The results of this project will help us anticipate what higher resolution views of the venusian surface might reveal in terms of aeolian features.

2. Background

2.1 Synthetic Aperture Radar

Synthetic aperture radar (SAR) is an airborne or spaceborne side-looking radar system that uses microwave energy to illuminate a surface. Synthetic aperture radar uses a single antenna as both a transmitter and receiver to record the surface in a pushbroom-like process

(Figure 2.1.1, all figures are given in Appendix A). SAR takes advantage of the motion of the spacecraft to simulate a larger aperture than is physically present to improve the spatial resolution. The instrument uses a series of pulses sent out by the transmitter and records the times that the pulses return from interacting with the surfaces as a backscatter value. The resulting data are formed into a grayscale image of the surface with radar-bright areas correlating to more energy being transmitted back to the detector and radar-dark areas are surfaces that reflected back little of the original pulse (e.g., Bamler and Hartl, 1998; Moreira, 2013).

The levels of surface radar brightness can correlate to the reflective properties of the material, such as specular or diffuse reflection, or to its dielectric properties. Both properties contribute to the reflectivity value of each pixel. Other properties that affect radar are grain size and roughness of a surface, and the angle of incidence of the detector to the feature or the surface

(Moreira, 2013). SAR inherently contains speckle noise from the scattering properties of materials, and the scattering centers of each resolution cell interacting with each other, creating a salt-and-pepper appearance when viewed at full resolution (Lee, 1981; Argenti et al., 2013).

SAR uses the Doppler effect to sharpen the resolution of the image in the azimuthal direction

(Bamler and Hartl, 1998). The Doppler effect refers to the difference in wavelength from an object that the spacecraft is moving towards and the wavelength from an object the spacecraft is

5 moving away from. This Doppler shift can be used to calculate the features various distances away from the detector and to reduce distortion (Bamler and Hartl, 1998; Moreira, 2013).

2.2 Magellan Orbiter

The SAR instrument on the Magellan spacecraft was an active system, using S-band 12.6 cm wavelength (Ford et al., 1989; Johnson, 1991; Arvidson et al., 1992; Wall et al., 1995). The

Magellan spacecraft operated from its May 4, 1989 launch on the Space Shuttle Columbia until

October 13, 1994, spending over 4 years in orbit around Venus gathering data (Ford et al., 1989;

Johnson, 1991; Arvidson et al., 1992). The SAR instrument produced single band images with a resolution of 75 m per pixel (Ford et al, 1989; Johnson, 1991). The quasi-polar elliptical orbit of

Magellan had a pericenter above 10°N latitude (Ford et al., 1989; Johnson, 1991). Magellan lasted for three cycles: left-looking, right-looking, and left-looking, with over 90% of the total surface covered in cycle 1, around 25% of the surface viewed from both directions, and around

98% of Venus having been mapped by the end of the third cycle (Ford et al., 1989; Saunders et al., 1990; Johnson, 1991).

The venusian atmosphere is composed of 96.5 percent carbon dioxide, with sulfur dioxide forming opaque clouds sitting in the middle atmosphere layer, reflecting 85 percent of the sunlight and severely limiting visibility from space (Zolotov, 2018). The denseness of the atmosphere limits the wavelengths to visibility “windows” that are the specific wavelengths that are not opaque through the thick cloud deck (Mueller et al., 2008). These windows are limited by the presence of CO2, N2, and SO2, primarily (Mueller et al., 2008). One of these windows is in the radio wavelength region, which can be used to image the surface geomorphology.

The Magellan spacecraft was designed for the venusian surface to rotate underneath for one full day for each mapping cycle, to image all 360° longitudes (Ford et al., 1989; Saunders et al., 1990; Johnson, 1991; Wall et al., 1995). The high-gain antenna had a 3.7 m diameter designed for both telecommunication and the synthetic aperture radar instrument (Ford et al.,

1989; Wall et al., 1995). By using a Block Adaptive Quantizer, Magellan was able to compress the data volume to allow for a better spatial resolution for a given data rate while preserving amplitude resolution (Kwok and Johnson, 1989). The SAR system on Magellan produced the highest resolution orbital images of Venus available at the time, and still constitute the highest resolution comprehensive global mosaic of the surface.

2.3 Sentinel-1

The terrestrial data used in this project are from the European Space Agency’s

Copernicus program. The program has seven developed missions, all named Sentinel, with different remote sensing objectives (Berger et al., 2012). Sentinel-1 is a radar imaging orbiter mission with two satellites — Sentinel-1A, launched on April 3, 2014, and Sentinel-1B, launched April 25, 2016 — flying in constellation using C-band radar in a twelve-day near-polar

Sun-synchronous orbit at 693 km altitude (Torres et al., 2012). The spatial resolution of the instruments is 13. 5 m per pixel, with a wavelength range of 3.8–7.5 cm (Torres et al., 2012).

Sentinel-1B has a right look direction with the incidence angle ranging from 30-42° (Torres et al., 2012). Both Sentinel-1 satellites are still active and are planned to last for at least seven years, with fuel for twelve years, likely ending sometime between 2024 and 2029. The Sentinel-1 mission objective is to monitor the Earth’s land masses and oceans and provide the highest resolution data quickly and easily (e.g., Hornacek et al., 2012).

2.4 Optical Imagery

Remotely sensed images in the visible region (VIS, ~350-700 nm) of the electromagnetic spectrum are multispectral or hyperspectral data products that can be used to view the true color of the surface. VIS imagery is also referred to as optical imagery, as the multispectral data products are built to mimic what we see by eye despite capturing as few as three wavelengths.

Orbital VIS instruments rely on the Sun to illuminate the planetary surface from space allowing the VIS instrument to measure reflectivity.

2.5 World Imagery Layer

The World Imagery Geographic Information System Layer is a freely available global mosaic imaged in the VIS wavelengths produced by ESRI and made available through Arc GIS.

The layer uses a variety of high resolution satellite and aerial imagery stitched together to produce a single continuous base map. The different remote sensing imaging contributors include many different public and governmental institutions — such as Earthstar Geographics,

National Centre for Space Studies (CNES)/Airbus Defense and Space, DigitalGlobe, U.S.

Geological Survey, U.S. Department of Agriculture Farm Services Agency, and GeoEye. The

VIS imagery used in this research over both the Namib Desert and the Simpson Desert is from

DigitalGlobe at 0.5 m/pixel spatial resolution. DigitalGlobe uses four multispectral channels: blue (450–520 nm), green (520–600 nm), red (630–690 nm), near-IR (760–890 nm).

DigitalGlobe contributes to the worldwide coverage of the Earth at high resolution from a variety of satellites and missions (e.g., WorldGlobe-1, WorldGlobe-2).

2.6 Jet Propulsion Laboratory Topography Data

The Jet Propulsion Laboratory at the California Institute of Technology has produced topographic data with a spatial resolution 0.5 km/pixel by combining two left-look cycles of

Magellan SAR data with different incidence angles (Nunes et al., 2018; Nunes et al., 2020). The

Magellan orbiter also contained a radar altimeter instrument, however that data product has a spatial resolution of 10 km/pixel at the equator and up to 25 km/pixel at higher latitudes, meaning that the features of interest would be indistinguishable from surrounding local topography (Bondarenko et al., 2006). The topographic data generated is higher resolution the elevation data gathered from the Magellan orbiter, allowing features on the scale of kilometers to be seen. The data products are represented as narrow vertical strips of images where the two look angles overlapped, producing a stereo image. This causes the data to have very limited coverage at the highest possible resolution in the areas of interest in this study, limiting the applications to one site.

2.7 Aeolian Features

Aeolian features are produced from the motion of sediment by wind through the air to create observable bedforms on planetary surfaces. Aeolian processes consist of three modes of transport — creep, saltation, and suspension (e.g., Nichols, 2009). Creep refers to particles rolling across a surface via wind, saltation refers to particles being moved through a series of jumps that land on the surface, and suspension refers to particles being entrained in the fluid or air with turbulence providing enough upward motion to stay within the body of the fluid (e.g.,

Nichols, 2009). Aeolian features are formed through saltation of sand-sized grains by wind through a system beginning at the sediment source, then being transported through the air, before

9 landing in the sediment sink. Sediment can be formed through physical, or mechanical, weathering of larger rocks often caused by erosion (Lancaster, 2009). The path of sediment transport is dictated both by the wind direction and more resistant pre-existing topography

(Lancaster, 2009; Nichols, 2009). Erosional aeolian features are formed as the sediment-carrying winds pass through a region of softer material, while depositional aeolian features are formed as the deposited sand builds into self-organizing bedforms (Lancaster, 2009; Nichols, 2009).

2.7.1 Dunes

Dune fields have been recognized on several planetary surfaces, including Earth, Mars,

Venus, and Titan (e.g., Tsoar, 1978; Tsoar, 1982; Lancaster, 1982; Tsoar, 1989; Bristow et al.,

2000; Tsoar, 2001; Colombini, 2004; Claudin and Andreotti, 2006; Hugenholtz et al., 2012;

Diniega et al., 2017). Dune fields can occur in a desert and are also referred to as an erg or a sand sea (e.g., Lorenz and Zimbelman, 2014). Sand dunes are formed from the saltation of sand-sized grains by winds forming large accumulations of sediment over time (e.g., Tsoar, 1978;

Lancaster, 1982; Tsoar, 1982, 1989; Bristow et al., 2000; Tsoar, 2001; Xing et al., 2001;

Colombini, 2004; Bagnold, 2012; du Pont, 2015). On Earth, there are several different types of dunes, including barchan or crescentic dunes, star dunes, transverse dunes, and longitudinal dunes (Figure 2.7.1.1) (e.g., Bristow et al., 2000; Bo and Zheng, 2011; Brookfield, 2011). On both Earth and Mars, the most common dune type is the barchan dune. Formed from winds blowing in one direction, they are wider than they are long, forming lunate structures and have a convex shape with both a steep slip face and a shallower lee face; the tails of the dunes pointing downwind (Ward et al., 1985; Greeley et al., 1993; Claudin and Andreotti, 2006; Ewing et al.,

2010; Bo and Zheng, 2011; Diniega et al., 2017). On Titan and Venus, linear dunes are the most

10 common dune type (Greeley et al., 1984; Lorenz et al., 1995; Claudin and Andreotti, 2006;

Paillou et al., 2016; Diniega et al., 2017). These dunes are different than barchan dunes, which are formed from uni-directional winds (Kocurek and Ewing, 2005; Bo and Zheng, 2011).

2.7.1.1 Linear Dunes

Linear dunes are features that have a greater length than width and form in relatively straight lines (Lancaster, 1982; Bristow et al., 2000; Colombini, 2004; du Pont, 2015). These features are identified by their characteristic parallelism and regular spacing (Lancaster, 1982;

Bristow et al., 2000; Colombini, 2004;). Linear dunes have long straight to slightly sinuous crest lines with an angle of repose of 32° (Bristow et al., 2000; Colombini, 2004). Linear dunes are also characterized with a significant inter-dune space in comparison to the dune width. Bi- directional winds with a single, dominant wind direction create linear dunes pointing in the same direction (e.g., Tsoar, 1978; Lancaster, 1982; Bristow et al., 2000; Tsoar, 2001; Colombini,

2004). The winds often blow orthogonal to the length of the dune, from the 90° or greater angle of the two winds. The dunes form fields in flat areas with little topography, persistent winds, and under arid to semi-arid climate conditions (Bristow et al., 2000; Colombini, 2004). Linear dunes typically have scalene, asymmetric cross-sections depending on the strength of the formative winds and their respective angles (Lancaster, 1982; Bristow et al., 2000; Colombini, 2004).

These features will form regardless of wind velocity, though there is an inverse correlation between wind velocity and time required to create linear dunes (Bristow et al., 2000; Tsoar,

2001; Colombini, 2004;). The length of the crestlines of the linear dunes can range from meters to kilometers, depending on the amount of available sediment, wind persistence, and regional topography (Lancaster, 1982; Bristow et al., 2000).

Long dunes that are firmed from bidirectional winds have many names; linear, longitudinal, seif, draa, uruq (Lorenz and Zimbelman, 2014). This work refers to the aeolian bedforms as linear dunes or a linear dune field. Terrestrial linear dunes form in arid equatorial regions, with a bimodal wind pattern often associated with seasonal changes (Paillou et al.,

2015). Linear dunes are characterized by their incredible size, with heights of 100 m or more.

Notable examples of terrestrial linear dunes occur in sand seas such as in the Simpson Desert in

Australia and the Namib Desert in Namibia (Folk, 1971; Twidale, 1972; Seeley, 1978; Hollands et al., 2006; Bristow et al., 2007).

2.7.1.2 Transverse Dunes

Transverse dunes are aeolian features that form from an abundant sediment supply and a single prevailing wind direction (Nichols, 2009) (Figure 2.7.1.1). The dunes are straight-crested, linear bedforms with a distinctly asymmetrical profile (Lorenz and Zimbelman, 2014). The stoss side is the shallower face, developing into the lee side (Nichols, 2009). Transverse dunes will often have more irregular patterns in the field, with smaller scale ripples developing orthogonal to the direction of the dune. Bimodal winds that are under 90° apart can also form transverse dunes, acting in a similar manner to a unidirectional wind (Paillou et al., 2015). Transverse dunes form when there is excess sediment available when the sand supply decreases, the features transition into barchan dunes (Nichols, 2009; Lorenz and Zimbelman, 2014).

2.7.2 Yardangs

Yardangs are found in feature fields similar to linear dunes, occurring on volcanic plains or on the side of a mountain for planetary surfaces, while on Earth they traditionally form in

12 hyper-arid deserts or basins (Blackwelder, 1934; Goudie, 2007). Yardangs are unique due to their positive relief erosional nature, that are formed from wind-blown sediment, rather than a fluid. The features are long, hull-like erosional features with sharp crests that are separated by narrow valleys (e.g., Blackwelder, 1934; Goudie, 2007; Paillou et al., 2015) (Figure 2.7.2.1).

Yardangs are formed from long-term erosion where wind-blown sediment has abraded weaker soil and rock into ridges (Blackwelder, 1934; Goudie, 2007). These ridges are formed from troughs initially being carved from wind-carried sediment, first forming narrow channels with steep sides and a U-shape, then widening to almost flat bottoms over time (Blackwelder, 1934;

Goudie, 2007). Over time, the channels widen and grow horizontally, and typically have transverse ripples in finer sediment across their floor (Blackwelder, 1934; Goudie, 2007). When two valleys connect, they start to form yardangs (Blackwelder, 1934). This process creates the tapered ends, with the curved sides, providing an aerodynamically streamlined form

(Blackwelder, 1934; Goudie, 2007). Yardangs have a steep windward face, with a shallower, more tapered leeward face (Blackwelder, 1934; Goudie, 2007). The yardangs form in parallel ridges that are either straight lines or slightly curved lines (Blackwelder, 1934; Goudie, 2007).

Yardangs can form at the scale of tens to hundreds of km in size, with the latter referred to as mega-yardangs (Paillou et al., 2015).

Yardangs have been recognized on Earth, Mars, and Venus. Prior to Magellan images of

Venus, it was believed that there were no yardangs present on the planet, however the resolution of the synthetic aperture radar images revealed evidence of at least one yardang field (Greeley and Arvidson, 1990; Trego, 1990; Greeley et al., 1992; Trego, 1992; Greeley et al., 1995). On

Earth, yardangs are found in deserts across the planet, within a wide range of rock types, such as sandstone, limestone, and ignimbrites (Blackwelder, 1934). On Mars, yardangs with a low width

13 to length ratios can be found in the equatorial regions of the planet (Ward, 1979; Ward et al.,

1985, Zimbelman and Griffin, 2010). On Titan, large fields of long, linear feature sets have been proposed to be yardangs (Paillou et al. 2015). Yardang fields do not correlate with general circulation models generated for a planetary surface from wind streaks, likely due to the length of time required to form a yardang field. Wind streaks are short-term aeolian features that occur on the surface and over the top of pre-existing features, while yardangs are formed from longer- term processes. Planetary yardangs are characterized by their massive scale of parallel ridges potentially indicating the presence of softer deposits over more resistant bedrock (Paillou et al.

2015).

2.7.3 Simpson Desert

The Simpson Desert is located in central Australia and forms the eastern portion of a whorl of linear dunes that is formed around the center of the continent (Figure 2.7.3.1). The

Simpson Desert covers 176,000 km2 and has a surface coloration trait is characteristically reddish due to iron oxide staining. The linear dunes found here are some of the longest on Earth, ranging from 20–24 km long with heights up to 40 m (Folk, 1970; Lancaster, 1982; Lorenz and

Zimbelman, 2014). The Simpson Desert dunes exhibit asymmetry with steeper eastern slopes, showing a bias in the transporting winds from the east (Wopfner and Twidale, 1967; Mabbutt et al., 1969; Folk, 1971; Twidale, 1981, Nanson et al. 1992). The eastern slope is at a 20° angle with a western slope at a 12°. The linear dunes are typically 150-300 m apart with interdune areas that are either vegetated or consist of exposed substrate (Lancaster, 1982).

2.7.4 Namib Desert

The Namib Desert is located on the Western Coast of Namibia and contains compound linear dunes (Figure 2.7.4.1). A bimodal wind pattern from the south-southwesterly wind blowing from the South Atlantic Ocean and an easterly, dry, mountain wind coming down the escarpment has created north-south-trending, complex linear dunes (Lancaster, 1989; Bristow et al., 2007). The southern-southwesterly wind is combined from the South Atlantic anticyclone, the upwelling from the colder South Atlantic Central Water, the cold Benguela Current, and the divergence of the South East Trades at the coast (van Zinderen Bakker, 1975, 1976; Seely,

1978). The cooling and upwelling from the winds created the arid climate with vegetationless dunes and minimal fauna present (Seely, 1978). The Namib Desert is around 34,000 km2, with a feature field with varying dune length between 20–40 km and are 5-30 m high (Lancaster, 1989).

The complex linear dunes have a main ridge 50–150 m high with a 600–1000 m width and up to

35 km in length (Lancaster, 1989; Bristow, et al. 2007; Lorenz and Zimbelman, 2014). The ridge profiles are asymmetrical with an orientation of the major slip face at 32–33° facing east to northeast. The windward slopes are 5–20° steepening near the crest. Larger, more widely spaced linear dunes occur in the central and northwestern areas of the desert with more closely spaced dunes exist nearer the margins.

2.7.5 Venusian Features and Working Hypothesis

The current working hypothesis for this research is that these observed features on Venus are dunes and yardangs. Other possibilities for these dune-like linear features are lag deposits, ergs, and mega ripples. Several of these features have been identified in multiple look angles, and their length and width measurements have been taken and are consistent with large dunes

(e.g., Greeley et al., 1992; Saunders et al., 1992). These features are likely not tectonic features, such as ridges or fractures, since they are long, sinuous lines and less bright than venusian tectonic features. The features are also more widely spaced than previously observed wrinkle ridges (e.g., Bilotti and Suppe, 1999) observed on Venus. The dune-like features are long found in parallel waves and have slightly sinuous crestlines. These features are characterized as slightly radar bright against a radar-dark surface, supporting previous theories (e.g., Greeley and

Arvidson, 1990) that the dune fields occur on the volcanic plains. These dune-like features form in large clusters, with the lines oriented primarily in the same direction. There is also a larger distance between the features than their average widths. The proposed aeolian fields are identified by these five primary characteristics detailed in section 5.2, with other similarities being noted and recorded for further study. In this present study, a range of estimations have been recorded and are congruent with previous studies done (e.g., Greeley et al., 1992; Weitz et al., 1994).

From both calculations and wind tunnel experiments, it has been demonstrated that high winds are not necessary to initiate particle motion on Venus. Florensky et al. (1977) reported the wind conditions were observed by the Venera landers; Venera landers 8 and 9 measured the average wind speeds to be 0.3–1.0 m/s at 1 m in altitude (Avduevskii et al., 1976; Keldysh,

1977) at the surface. From theoretical calculations of the Hjulström curve (Iversen and White,

1982) (Figure 3) paired with experimental data values from the Venus Wind Tunnel (Greeley et al., 1984), the theory and experimental data indicate that a threshold friction velocity of 0.03 m/s is required to initiate grain motion (Figure 2.7.5.1). Greeley et al. (1984) suggested that available grain sizes on Venus that could be moved by saltation would range from 60–2000 µm with the

16 most easily moved grain size being ~ 75 µm. Any smaller grain sizes would move through suspension, and larger grain sizes would creep along the ground.

From Venus Wind Tunnel experiments, it had been demonstrated that aeolian features are possible on the surface of Venus given a sufficient sediment supply. From the studies of terrestrial dunes fields (e.g., Blom and Elachi, 1981; Blom and Elachi, 1987), it is known that dunes will typically appear radar-bright at small look angles when the smooth dune faces are oriented near-normal to the radar beam (Weitz et al., 1994). The angle of repose on Venus is also calculated to be similar to Earth, around 32°, and the faces were orientated parallel to the radar illumination. Arvidson et al. (1991) first suggested that a dune field on Venus was visible in the

SAR images as a speckled area on volcanic plains, with the orientation of the wind streaks consistent with inferring a dominant wind direction. This thesis expands on this observation and subsequent studies (e.g., Greeley et al., 1992; Weitz et al., 1994) with a similar starting assumption that the observed features are dunes.

2.8 Identification Resolution

Feature identification resolution is defined as the level of spatial resolution (e.g., 10 m/pixel) that is needed to correctly identify various geomorphological features in remotely sensed images (Zimbelman, 2001). The range of spatial resolutions necessary to identify various features can be determined through the appearance and scale of diagnostic characteristics, though the level of preservation will also influence the range. Figure 2.8.1 illustrates the relationship between object size in pixels and angle of illumination for feature identification in in optical images. As the sun elevation above the horizon increases, the identification resolution needed for feature detection increases (Zimbelman, 2001).

Feature identification also requires the multiple working hypotheses system to allow for the interpretation of features based on the researchers and increase the reproducibility of the work (e.g., Zimbelman, 2001). While multiple hypotheses can produce divergent results, it can also allow for a reduction in human bias towards previously published findings of the same surface and allow for intellectual evolution and maturity that will lead to a more comprehensive and accurate understanding of the features in question (Chamberlin, 1897; Zimbelman, 2001).

Multiplicity refers to multiple causes acting in conjunction with each other to produce a single outcome, i.e., where a single feature can be formed through many diverse relationships acting and reacting together (Schuum, 1991). Convergence refers to multiple different causes and processes producing similar effects, directly contributing to many hypotheses of the interpretation of one feature. Outgroup testing is defined in this work as a method of exploring alternative or multiple other hypotheses of processes that could have formed the observed features. Outgroup testing to test various possible processes is used to dispute the theory of convergence, while the generation of a truth table enumerates observed characteristics of the feature and compares the likelihood of multiple converging identification hypotheses (Schumm,

1991). This process considers multiple working hypotheses and compares diagnostic characteristics against the unknown group.

3. Methods and Data

This section details the methodology for processing and analyzing with terrestrial data,

Magellan data, and venusian topographic data. The data products used are the Sentinel-1B SAR,

Magellan SAR, topographic data generated from the Magellan SAR data, and ESRI’s World

Imagery.

3.1 Terrestrial SAR Data Processing

The two terrestrial locations that have been identified for this project are the Simpson

Desert and the Namib Desert. These terrestrial dune fields have been selected as venusian analogs for their full synthetic aperture radar (SAR) data coverage, the vastness of the dune fields, and the median scale of the bedforms. SAR images of each area were selected from the

Sentinel-1 mission using the Copernicus software program (Foumelis et al., 2018). The images were then downloaded from the ESA open access data hub

(https://scihub.copernicus.eu/dhus/#/home) acquired between August 1, 2018 – August 31, 2018 from the 1B satellite orbiting Earth. The location of interest was selected through the Copernicus map interface (Figure 3.1.1). Each SAR scene was then downloaded as a series of Sentinel-1B level 1 data products that are Single Look Complex Interferometric Wide Swath images. For the

Simpson Desert dune field, eleven SAR images have been downloaded addition to five SAR images for the Namib Desert dune field (Figure 3.1.2).

Each zipped image file was individually processed using the European Space Agency’s

(ESA) Sentinel Application Platform (SNAP) 7.0 software. The first step was radiometrically calibrating the Sentinel-1B data to reduce the noise in the data, thus correcting the pixel values in the SAR image to accurately represent the radar backscatter of the surface (Foumelis et al.,

2018). The corrections that were applied are generated from the metadata of the products, and the radiometric calibration was applied to all source bands. A single Sigma0 band for each real and imaginary pair is the output for this step.

The next step included merging the three sub-swaths present in the Interferometric Wide swath mode data. The merging was done using Terrain Observation with Progressive Scans SAR

(TOPSAR)-Deburst. The swaths are bands of the same pixel area taken at slightly different times with different doppler shifts (Foumelis et al., 2018). For each image, the overlapping swaths, or bursts, were merged to produce one image. This project only used images processed with S1-

TOPS Deburst, rather than the individual swaths, consolidating the bands available to

Sigma0_VH and Sigma0_VV (Foumelis et al., 2018).

Once each image has been de-noised and the bursts have been merged, pixels of an image were averaged to produce a more realistically proportional image with a single spatial resolution

(Foumelis et al., 2018). This process converts multiple lines of pixels in range direction with multiple lines of pixels in the azimuth direction (Foumelis et al., 2018). The result of this step is to decrease the noise and produce approximately square pixel spacing in the images.

The final processing step was to use digital elevation models to perform geometric range- doppler terrain correction. This image corrects for the terrain associated with the image and project the image correctly onto the surface (Foumelis et al., 2018). SAR images are acquired with a slant-looking geometry, that can cause distortion of the image over areas of elevation change, so running a range doppler terrain correction step shifts all the pixels in each image to the correct location according to the provided DEM (Foumelis et al., 2018). The geometric terrain correction that was done for all Sentinel images was the default WGS 1984 coordinate system. Once the images were processed, they were converted and exported as geotiffs.

3.2 Terrestrial ArcGIS Mapping

The SAR images were loaded into ArcGIS and visually compared with the ESRI World

Imagery VIS layer to confirm the geotiffs were georectified at the correct scale. The background pixel value of each SAR image was set to zero to remove data gaps and excess pixels; this process allows the images to be overlaid onto each other more seamlessly. A minimum- maximum stretching was applied to the Namib Desert images to enhance the contrast around the dunes. The minimum and maximum values used for the image stretching are manually determined by querying a subset of radar-dark pixels over the dune field. The processed SAR images’ spatial resolution is 13.5 m/pixel. Field extents were manually mapped using a polygon to demarcate the edges of the field (Figure 3.2.1). This process was repeated multiple times with at least 12 to 24 hours of time between each mapping to define the field extent as best as possible. The dune fields have indistinct edges, and therefore mapping the dune field extent multiple times over a long period of time minimizes human bias and maximizes repeatability. A polyline was created in ArcGIS and the dune crestlines of a representative area were mapped at the 13.5 m/pixel spatial resolution SAR image, considered the high resolution SAR layer (Figure

3.2.1). Then, using the resampling tool in ArcGIS, the high resolution SAR images were resampled from 13.5 m/pixel to 75 m/pixel, degrading the quality of the image to match the

Magellan SAR resolution. This lower resolution SAR image is referred to as the downsampled

SAR image. A second shapefile was used to map dune crestlines at this layer spatial resolution

(75 m/pixel).

The visible (VIS) wavelength layer, used as a control in this research, is the World

Imagery layer provided by ESRI ArcGIS Online (Hu et al., 2015). The layer is a global, high- resolution mosaic of cloudless VIS images assembled from multiple terrestrial orbiters. The

21 resolution over the Simpson Desert and the Namib Desert is 0.5 m/pixel from the DigitalGlobe satellites. The VIS imagery layer was processed by the same procedure as the SAR terrestrial images, mapping the visible dune crestlines with polylines. The dune field extent was mapped multiple times, each using a different shapefile, to most closely approximate the true extent of the dune fields.

For each of the three resolutions of both the Simpson Desert and the Namib Desert, the polyline end vertices were converted to points in a separate shapefile, referred to as dune defects

(e.g., Werner and Kocurek, 1997; Werner and Kocurek, 1999; Ewing et al., 2006; Ewing et al.,

2015). Ten different measurements of the spacing between dune crestlines were taken and averaged to produce an average spacing value for the dune crestlines in each resolution at both localities (Figure 3.2.2). The dune defects were then divided by two to be counted as dune defect pairs (N) and divided by the average spacing of the dune crestlines (S) to calculate the dune defect density (ρ) for a representative area of each desert at all three resolutions. The dune defect density was plotted against the dune crestline spacing to show the denseness of dunes formed in the representative area.

3.3 Magellan SAR Data Processing

The venusian data used in this study are from NASA’s Magellan SAR instrument and were downloaded from the United States Geological Survey’s Map a Planet image processing tool (Hare et al., 2014). This tool allows a user to specify a region on the global mosaic and a coordinate system based on the location of interest and provides a georeferenced image. The downloaded Magellan data include both the left-looking SAR data and the right-looking SAR data, depending on the area (Saunders et al. 1991). The five locations that have been examined

22 are the Aglaonice feature field, the Fortuna-Meskhent feature field, the Guan Daosheng feature field, the Stowe feature field, and the Mead feature field (Figure 3.3.1, 3.3.2) (Greeley and

Arvidson, 1990; Greeley et al., 1992, 1994, 1995; Rader et al., 2019; Weitz et al., 1994). All fields occur on the volcanic plains of Venus, apart from Fortuna-Meskhent feature field that is situated on a volcanically embayed area of Ishtar terra on the tesserae. The Aglaonice field is located near the Aglaonice crater and the Carson crater. This area is referred to as the “crater farm” due to the relatively high number of impact craters (Rader et al., 2019). The Fortuna-

Meskhent features are medium radar-bright due to the contrast of the high backscatter tesserae and the very radar-bright wind streaks (Rader et al., 2019). The field is located between the

Meskhent and Fortuna tesserae and the Ishtar terra, more specifically around the Ops Corona, Al-

Uzza Undae, and Jadwiga crater (Greeley et al., 1992). The Mead features are also radar-bright with radar-dark plains, due to the difference in roughness (Rader et al., 2019). The observed bedform set is located near Mead crater, Kuro crater, and Ningal Lineae (Rader et al., 2019). The venusian localities were compared to the parabolic crater ejecta map generated by M. Gilmore from previous work (Campbell et al., 1992; Gilmore, 2016).

3.4 Magellan ArcGIS Mapping

Once each aeolian field was downloaded and processed from the USGS’s Map A Planet, each of the images were inputted into ArcGIS 10.6. For each location, the images need to be individually loaded into the program and a base map was used to confirm it was correctly scaled and georeferenced. As with the terrestrial dune fields, a shapefile was created, and the field extents are mapped with a polygon and then turned off once completed and is repeated multiple times with at least 12 to 24 hours of time between each mapping (Figure 3.4.1). This helps

23 minimize human bias in determining the mapping of individual field extents. Then a representative subset of the dunes is mapped due to the size of the dune fields (ex. ~20,000 km2).

All images that included any data gaps from the Magellan instrument have the field extent truncated at the edge of the image, causing occasional straight edges. This decision reduces uncertainty as to the area or field extent location.

For the Aglaonice, Fortuna-Meskhent, and Mead feature fields, the degree orientation for each mapped feature was calculated. The orientation values were then sorted into 10° bins and displayed on a rose diagram using R to perform the analysis. The rose diagrams generated for each venusian locality show the counts versus the polar coordinates of the orientation, illustrating the overall directional trends of the feature fields.

3.5 Venusian Topographic ArcGIS Mapping

The Magellan SAR images contain x and y coordinates in the pixels, without elevation values. The elevation global mosaic that was collected with the Magellan mission has been globally sampled to a spatial resolution of 2-3 km/pixel; individual bedforms in the five venusian localities examined in this work are not resolvable. Due to all data products not providing elevation data fine-scaled enough to measure or map any surface features, an alternative dataset was needed. The Jet Propulsion Laboratory (JPL) at the California Institute of Technology

(Caltech) produced higher resolution topographic data using both of the Magellan SAR left-look cycles, cycles 1 and 3 (Nunes et al., 2018, 2020). The topographic data were imported into

ArcGIS for the area surrounding Mead Crater and were used to confirm that the topographic data were georectified and at the same scale with the Magellan SAR data. The color ramp used is blue-orange-yellow with the blue at low elevation and yellow at high elevation with orange

24 indicating average elevation values (Figure 3.5.1). The linear features observed in the images are mapped with a polyline from a shapefile. The topographic data have significant data gaps from the processing done by JPL and are more similar to thin strips of data than a full image. This limits the field extents that can be estimated, so following the same decision as with the

Magellan SAR images, the mapped edges are truncated at the data gaps to prevent speculation.

4. Results

4.1 Simpson Desert

4.1.1 VIS at 0.5 m/pixel

The Simpson desert in the VIS layer has reddish sand with linear dunes that can be easily observed by eye. The dune crestlines are longer than they are wide and there is a defined sinuosity in the crestlines. The dunes have two faces with smaller rocks and sediment at the bases of the dunes. The linear dunes display a high density of bifurcations or dune defects (dune defect density of 0.01) and clearly show dunes that have branched off (Werner and Kocurek,

1997; Ewing et al., 2006). The image resolution is sufficient to differentiate between dunes with bifurcation and dunes that have formed near pre-existing dunes. The 0.5 m/pixel is able to show the endpoints of the dunes with a high level of accuracy and the dune slipfaces. This detail differentiates between dunes that display bifurcation and with others that have formed close by yet, distinct from its neighbors (Figure 4.1.1.1). The representative area in the Simpson dune field was chosen because the linear dunes had a clear starting point beneath a region of more resistant bedrock (Figure 4.1.1.2).

In the representative area of the dune field has been mapped in Figure 4.1.1.3, with the dunes display an overall linear trend of northwest to southeast with multiple regions of sinuosity along each feature. The full population of mapped features is shown in Figure 4.1.1.4 with the dune crestlines mapped in pink. The average dune spacing is 333.2 m with a standard deviation of 105.7 m, the average dune length is 14.06 km with a standard deviation of 16.94 km, and the average dune width of 223.9 m with a standard deviation of 61.2 m. The dunes also show the complexity of the crestlines, with one major crestline creating the dominant profile, with branching or merging segments that are only visible at close range. The inter-dune area has

26 inconsistent coloring of the material with sediment-poor regions found between the bedforms.

The sand that does appear between the features is distinctly reddish, the same color as the dune crestlines. In Figure 4.1.1.2, the surrounding bedrock is more grey or blue with more resistant layers emerging in areas. The more sediment-rich inter-dune areas are around dunes that have bifurcated. The area surrounding and in between two bifurcating dunes shows a higher density of reddish material than between two unrelated dunes.

4.1.2 SAR at 13.5 m/pixel

The Simpson dune field in SAR is seen as a radar-dark field with radar-bright bedforms

(Figure 4.1.2.1). The interdune area is consistently radar dark, with the radar-bright linear features visually aligning with dune crestlines in VIS. The spatial resolution for the processed

SAR images is 13.5 m/pixel. The radar image introduces an inherent level of background noise that the VIS images did not contain due to the scattering properties of radar. The surface features visible in the SAR image have less distinct termination points and can appear to be harder to identify from the surrounding area. The same representative area that was mapped in VIS was mapped in the SAR image at 13.5 m/pixel.

In the representative area of the dune field mapped in Figure 4.1.2.1, the dunes display an overall linear trend of northwest to southeast with multiple regions of sinuosity along each crestline, similar to the VIS images. The radar-bright features have a linear pattern with a greater length than width and semi-regular inter-dune spacing. The average dune spacing is 345.4 m with a standard deviation of 82.4 m, the average dune length is 12.76 km with a standard deviation of 10.32 km, and the average dune width of 77.9 m with a standard deviation of 40.8 m. The radar-dark surface is speckled with some radar-bright pixels and with a faint radar

27 shadow on the left side of all the dune crestlines (Figure 4.1.2.2). The full population of mapped features is shown in Figure 4.1.2.3 with the dune crestlines mapped in yellow. The SAR at 13.5 m/pixel also shows smaller dunes that are shorter in length and less radar-bright than some of the longer dunes. The SAR image does not show small scale dune characteristics, such as the slipfaces of the dunes, or any rocky material at the base of the dunes. The radar-bright features are linear patterns without additional small-scale features associated with them. The radar-dark inter-dune areas have a slight salt-and-pepper pattern from speckle noise, with each area being indistinguishable from another. There is no sign of small patches of rocky bedrock appearing between the regular spacing of the dunes. Larger radar-dark areas affect the spacing of the dunes.

4.1.3 SAR at 75 m/pixel

The Simpson dune field in SAR at 75 m/pixel has a radar-dark field with radar-bright features either due to the roughness or the tilted angle (Figure 4.1.3.1). This image was downsampled from the SAR image at 13.5 m/pixel with the same representative area mapped as the other Simpson Desert locations. The radar-bright linear features align with the crestlines in the VIS image and the higher resolution SAR image. With the downsampled image, the image has an increased proportion of noise to signal and more blurring of the features.

In the representative area, the crestlines display only a slight sinuosity that is visible enough to be mapped. The dune sides are indistinct, and the radar-shadow is unclear to non- existent. The average dune spacing is 478.9 m with a standard deviation of 123.1 m, the average dune length is 10.63 km with a standard deviation of 7.98 km, and the average dune width of

113.9 m with a standard deviation of 36.0 m. The crestlines are still visible as radar-bright streaks over a dark field with radar-dark inter-dune areas (Figure 4.1.3.1). The mapped features

28 are still visible, though most points of bifurcation evident at higher resolution are not identifiable at 75 m/pixel (Figure 4.1.3.2). The interdune areas are less radar-dark due to a higher degree of speckling and the radar-bright crestlines are more pixelated in comparison to the 13.5 m/pixel resolution at the same scale and over the same representative area. The full population of mapped features is shown in Figure 4.1.3.3 with the dune crestlines mapped in light green.

4.2 Namib Desert

4.2.1 Linear Dunes in VIS at 0.5 m/pixel

The Namib Desert has well-documented linear dunes with a north-south trend (e.g.,

Lancaster, 1995; Bristow et al., 2007). In VIS images, the eastern edges of the dune field show smaller dunes that still display distinctive crestlines. The closer to the edge of the dune field, the shorter the dunes become. The dune crestlines appear dark against the optically brighter inter- dune area. The dunes in the Namib Desert have an average dune crestline length of 14 m and display a low-to-moderate level of sinuosity. The 0.5 m/pixel spatial resolution detail shows the distinct termination points of the linear dunes, as well as the smaller linear dunes on the edge of the dune field. The dune crestlines have some visible ripples in them towards the center of the dunes, with their ends appearing smoother.

A representative area of the dune field with well-developed linear dunes was mapped and is given in Figure 4.2.1.1. The dunes displayed a minimal level of bifurcation, and the high spatial resolution allows for a differentiation between dune defects and independent dunes formed close to each other. The average dune spacing is 2.02 km with a standard deviation of

0.52 km, the average dune length is 13.99 km with a standard deviation of 2.56 km, and the average dune width of 922.7 m with a standard deviation of 176.3 m. The dunes have distinct end points with their rippled crestlines and red-orange sand. The interdune area is filled with

29 sediment and the same color as the dunes with this similarity particularly being noticeable the northern ends of the dunes. The dunes in the representative area show minimal bifurcation and low sinuosity. The average crestline to crestline spacing is 2018 m. The dunes have a regular spacing, however it varies in areas where smaller dunes have formed between larger dunes

(Figure 4.2.1.2). The full population of mapped features is shown in Figure 4.2.1.3 with the dune crestlines mapped in green.

4.2.2 Linear Dunes in SAR at 13.5 m/pixel

The Namib Desert in SAR at 13.5 m/pixel in the representative area appears as a moderately radar-bright overall surface with radar dark features (Figure 4.2.2.1). Inter-dune areas that are highly radar-bright correlate to areas with less sediment, while the inter-dune areas with more sediment have radar backscatter values more similar to the dune crestlines. This occurs because the radar wave will penetrate up to ten times the wavelength into the surface, and therefore is interacting with bedrock, rather than the sediment. The SAR image has been confirmed with the VIS image to be georeferenced using resistant bedrock outcrop on the edge of the dune field. These linear features hare a clearer pattern at a larger scale, but at a finer scale the termination points of the dunes become less distinct. The speckle noise in the radar image also contributes to the blurring of the ends of the features.

In the representative area of the dune field that is given in Figure 4.2.2.2, the dunes display the same northeast-southwest trend that is seen in the VIS image. The features with the lowest radar backscatter values are linear with a moderately regular spacing and are longer than they are wide. The full population of mapped features is shown in Figure 4.2.2.3 with the dune crestlines mapped in blue. The average dune spacing is 2.19 km with a standard deviation of 0.33

30 km, the average dune length is 15.40 km with a standard deviation of 11.30 km, and the average dune width of 767.8 m with a standard deviation of 243.9 m. The features mapped include some potential bifurcation points and few smaller, independent dunes (Figure 4.2.2.2). These dune defects were not observed in the 0.5 m/pixel VIS image, suggesting that they are a product of the radar image with a lower signal to noise ratio. The 13.5 m/pixel SAR image does not show slip faces, nor any asymmetry of a dune profile. With the radar-dark features being the dunes, there are also no radar shadows that can be observed. The larger radar-bright areas between the dunes does not affect the visibility of them, rather making those areas the easiest to map.

4.2.3 Linear Dunes in SAR at 75 m/pixel

The Namib Desert in SAR at 75 m/pixel show fewer linear features than in the SAR at

13.5 m/pixel. The representative area is the same as what was used for the 0.5 m/pixel VIS image and the 13.5 m/pixel SAR image, appearing as a radar-bright field with radar-dark features

(Figure 4.2.3.1). The 75 m/pixel SAR image was downsampled from the 13.5 m/pixel SAR image in ArcGIS for comparison. The linear radar-dark features that have been observed align with the crestlines in the other two images of the area. The downsampled image has further increased the signal to noise ratio, as well as the pixilation of the image.

The dune crestlines in the representative area display almost no sinuosity with indistinct edges. The speckle noise has obscured the termination points of the dunes, causing some inter- dune areas with lower radar backscatter values to blend in with the dune crestlines. The radar- bright pixels stand out more against the contrast, however, they predominantly occur in the inter- dune area. The crestlines are still visible and have been mapped in Figure 4.2.3.2 in orange. The full population of mapped features is shown in Figure 4.2.2.3. The mapped dunes show no

31 bifurcation and the crestlines also appear shorter in the 75 m/pixel SAR image than the other images. The average dune spacing is 2.04 km with a standard deviation of 0.37 km, the average dune length is 12.05 km with a standard deviation of 7.08 km, and the average dune width of

702.3 m with a standard deviation of 134.1 m. These data limitations are results of the spatial resolution quality of the data product, rather than a representation of the area.

4.2.4 Transverse Dunes in VIS at 0.5 m/pixel

Although the Namib Desert is known for linear dunes, on the southwestern coastal region of the field, transverse dunes that have also formed. The transverse dunes are clearly visible with a stoss side showing smaller ripples and a lee side that appears smoother. The asymmetrical shape of transverse dunes is also visible in the VIS image with beige-yellow color of the sand, rather than the orange-red color in the center of the dune field with the linear dunes (Figure

4.2.4.1). The representative area mapped in VIS shows dunes with moderate sinuosity. The average dune spacing is 851.4 m with a standard deviation of 325.2 m, the average dune length is

5.18 km with a standard deviation of 2.19 km, and the average dune width of 707.4 m with a standard deviation of 133.3 m. The transverse dunes do not come in contact with one another, though they can occur close to each other (Figure 4.2.4.2). The inter-dune area appears to be smooth with large amounts of sediment between the structures.

4.2.5 Transverse Dunes in SAR at 13.5 m/pixel

The Namib Desert in SAR at 13.5 m/pixel highlights dunes on the edge of the dune field while concealing the large dune crestlines in the center of the image. The field is radar-dark with radar-bright features seen in Figure 4.2.5.1. The radar-bright linear features align with the

32 transverse dune crestlines seen in the VIS image, though the crestlines are not very distinct from the inter-dune area, especially closer to the western coast. The radar-bright features observed in the 13.5 m/pixel SAR image display a generally low sinuosity and a moderately uniform spacing.

The average dune spacing is 952.8 m with a standard deviation of 297.0 m, the average dune length is 3.34 km with a standard deviation of 1.08 km, and the average dune width of 398.9 m with a standard deviation of 97.6 m. The features are linear, angled north-west-west south-east- east; however, they are wider than linear dunes mapped in the Namib Desert, suggesting there is a radar layover effect on the lee slope of the transverse dunes (Figure 4.2.5.2). There are fewer mapped dunes in the 13.5 m/pixel SAR image than the VIS image, due to the decreased visibility.

4.2.6 Transverse Dunes in SAR at 75 m/pixel

The Namib Desert in SAR at 75 m/pixel exhibits fewer linear features than in the SAR at

13.5 m/pixel. The 13.5 m/pixel SAR image was resampled to generate the 75 m/pixel SAR image, decreasing the speckle noise but increasing the pixelation (Figure 4.2.6.1). The background of the field is radar-dark with a few radar-bright linear features in the southeastern region. The decreased spatial resolution at the smaller scale causes the image to significantly increase the pixilation and limit the possible features that can be identified. The linear features that have been identified and mapped in this SAR image are short, linear streaks with no apparent sinuosity (Figure 4.2.6.2). The linear pattern aligns with both the 13.5 m/pixel SAR image linear features and the 0.5 m/pixel VIS image linear features, suggesting they are the only transverse dune crestlines that can be observed out of the whole field. The features trend the same north-west-west south-east-east direction, though do not have a regular apparent spacing.

The average dune spacing is 1.07 km with a standard deviation of 0.23 km, the average dune length is 2.33 km with a standard deviation of 0.95 km, and the average dune width of 385.2 m with a standard deviation of 94.3 m. The limited visibility of the transverse dune crestlines suggests a lower boundary of bedform scale for the identification resolution.

4.3 Venusian Localities

4.3.1 Aglaonice Feature Field

The Aglaonice feature field extends from 20.9°W to 20.2°W longitude and 24.85°S to

24.4°S latitude. The structures found in this field are around 3 kilometers long and 0.3 kilometers wide and are radar-bright bedforms that stand out against the radar-dark surface. The featuers in the Aglaonice field are linear, with long, skinny crestlines (Figure 4.3.1.1, Weitz et al., 994). The features are highly speckled with radar-bright linear features that are not a solid radar-bright feature, but instead are more spotted. The average dune spacing is 1.38 km with a standard deviation of 0.58 m, the average dune length is 4.79 km with a standard deviation of 3.17 km, and the average dune width of 424 m with a standard deviation of 140.0 m. The inter-feature area is radar dark and has the same appearance as the larger region that the feature field is on. A radar-bright patch was observed on the volcanic plains, that consistently appear radar-dark, suggesting that surface features exist to create the anomaly due to the feature geometry with respect to the radar look direction and incidence angle (e.g. Arvidson et al., 1992; Greeley et al.,

1992; Weitz et al., 1994). The feature field is located near the Aglaonice crater and the Carson crater. This area is referred to as the “crater farm” due to the relatively high number of impact craters (Greeley et al., 1992; Weitz et al., 1994). The average feature field extent is 1.86 km2.

4.3.2 Fortuna-Meskhent Feature Field

The Fortuna-Meskhent feature field extends from 88°E to 92°E and 65.6°N to 68°N latitude. The bedforms found in this field are between one and 5 kilometers long and 0.7 to 0.5 kilometers wide. The observed Fortuna-Meskhent structures are radar-bright linear features over a radar-dark background (Figure 4.3.2.1). The features are linear with some degree of sinuosity.

The features appear as a large population with regular spacing, with another population of different features superimposed over the first (Figure 4.3.2.2). The average dune spacing is 576.5 m with a standard deviation of 162.1 m, the average dune length is 5.97 km with a standard deviation of 2.87 km, and the average dune width of 312.7 m with a standard deviation of 63.3 m. The inter-feature area is slightly more radar-bright than the volcanic plains surrounding the region. The field is located between the Meskhent and Fortuna tesserae and the Ishtar Terra, more specifically around the Ops Corona, Al-Uzza Undae, and Jadwiga crater. The average feature field extent is 1.5*107 km2.

4.3.3 Guan Daosheng Feature Field

The Guan Daosheng feature field extends from 179°E to 182°E longitude and 61°N to

63°N latitude. The features are small enough that they are only a few pixels long and have been classified as microdunes (Weitz et al., 1994) and are around 2.5 kilometers long and 0.5 kilometers wide (Figure 4.3.3.1). The Guan Daosheng microdune field is identified as radar- bright features atop radar dark plains. The feature field is located between Gran Daosheng crater and Uluk crater. The average feature field extent is 12.11 km2.

4.3.4 Stowe Crater Feature Field

The Stowe feature field extends from 128°W to 126.9°W and 40.8°S and 42°S. Dunes found in this field are around 1 kilometer long and 0.5 kilometers wide (Figure 4.3.4.1) (Greeley et al., 1992; Weitz et al., 1994; Rader et al., 2019). The individual features are too small to map with the same level of accuracy applied of other locations. Features in the Stowe feature field are only visible in the SAR right-look data, and not in the left-look (Greeley et al., 1992; Weitz et al., 1994; Rader et al., 2019). Therefore, the data for this location from the cycle 2 SAR right look, while the other locations all use SAR left-look data (Greeley et al., 1992; Weitz et al.,

1994; Greeley et al., 1995; Rader et al., 2019). The Stowe feature field is radar-bright with radar- dark plains. The feature field was identified by a radar-bright patch in the right-look data, standing out from the radar-dark plains in the left-look data (Greeley et al., 1992; Weitz et al.,

1994; Rader et al., 2019). The feature field is located near Stowe Crater with Alima Crater and

Achek Dorsa nearby and is superimposed atop a set of northwest-southeast trending wrinkle ridges. The average feature field extent is 84.27 km2.

4.3.5 Mead Crater Feature Field

4.3.5.1 Magellan SAR at 75 m/pixel

The Mead feature set extends from 60°E to 62°E and 8°N to 10°N. The features in this field, interpreted as erosion yardangs rather than bedforms (Greeley et al., 1992), average 23 kilometers long and around 0.3 kilometers wide. The observed features are also radar-dark with radar-bright plains, due to the difference in roughness (Figure 4.3.5.1.1). The linear features appear as two sets of features curving with the western set having a positive curvature and the

36 eastern set having a negative curvature. The average yardang spacing is 1.13 km with a standard deviation of 0.62 km, the average dune length is 10.43 km with a standard deviation of 7.56 km, and the average dune width of 351.1 m with a standard deviation of 122.3 m. The two populations also have a transition region between the two where similar features appear without a clearly defined pattern. The features display no sinuosity and with a narrow wavelength and no visible bifurcation anywhere. The feature field is located near Mead crater, Kuro crater, and

Ningal Lineae. The average feature field extent is 19.95 km2.

4.3.5.2 Jet Propulsion Laboratory Topographic Data

The Mead feature field in the topographic data shows the same region of interest as the

Mead feature set in the Magellan SAR images. The linear features in the topographic data are mapped in pink in Figure 4.3.5.2.1. The high topographic elevations are orange and low elevations blue. The linear features appear to be in two sets with a slight curvature on the western half of the image and the eastern population of features shows a negative curvature. The features are linear with no sinuosity, sporadically showing a slight curvature. The features mapped only cover the available topographic data, meaning that all mapped features are truncated at data.

5. Discussion

5.1 Interpretation of Results

5.1.1 Simpson Desert

The compound linear dunes in the Simpson Desert appear in both VIS and SAR images, though drastically changed in appearances. The dune crestlines in VIS layer have a high level of sinuosity with significant variations in the dune spacing (Figure 5.1.1.1). The overall pattern of dunes has an overall uniform spacing; however, the dunes show bifurcation and smaller features can be seen nearby (Figure 5.1.1.2). The high resolution VIS image shows the difference between smaller dunes that have formed near each other and a dune that has bifurcated into two.

This difference is evident in the VIS image at 0.5 m/pixel and the high resolution SAR image at

13.5 m/pixel, though the latter becomes more vague due to the lower resolution and the radar speckle. The termination points of the dunes are also obfuscated in both of the SAR due to the speckle noise, limiting the absolute measurements that can be made at these resolutions. The regularity of the spacing of the dunes increases as the resolution decreases, which then limits the identifiable features in the inter-dune area. This inability to positively identify smaller dunes between the larger ones causes the crestlines to appear more uniform in the low resolution SAR image than they actually are in the VIS image.

5.1.2 Namib Desert

The linear dunes in the Namib Desert are known for their length and size (e.g., Lancaster,

1995; Bristow et al., 2007), which were clearly visible in the VIS images, but were significantly obscured in the SAR image. The dune crestlines were compared in VIS to the high resolution

SAR image have been observed to be the radar-dark linear features (Figure 5.1.2.1). The level of

38 sinuosity and frequency of dune defects observable in the VIS image are obscured in the high resolution SAR image, similar to the effect observed in the Simpson Desert images. However, in the Namib Desert images, the high resolution SAR images also have instances where features are observed and mapped that are not present in the 0.5 m/pixel VIS image. This phenomenon shows that some gaps in crestlines in VIS were mapped as continuous in the 75 m/pixel SAR imagery and shows a potential unreliability of SAR images without additional information (Figure

5.1.2.2). Similar to the Simpson Desert images, the inter-dune spacing becomes more uniform as the resolution decreases.

The transverse dune crestlines observed on the western coast of the Namib Desert were visible in the 0.5 m/pixel VIS image but are almost unrecognizable in the SAR images in both spatial resolutions. The dune crestlines in the VIS image show the transverse dunes clearly, though in the SAR at 13.5 m/pixel there were significantly fewer features observed (Figure

5.1.2.3). In the 75 m/pixel SAR image the only dune crestlines that are visible are the ones farthest from the coast (Figure 5.1.2.4). These visibility issues are likely caused by a combination of the angle of the features relative to the radar look direction and incidence angle and the quantity of sediment in the region. As discussed with the linear dunes in the Namib desert, transverse dune fields contain moderate to large amounts of sediments, limiting the differences in the radar return between dune and interdune area. The viewing angle between the instrument and the surface features also affects the radar backscatter value, causing features that are oriented too close to parallel to not appear in the SAR images. These effects are exacerbated in the 75 m/pixel

SAR image, with the added increase of pixelation, show that a dune field can be reduced to only a fraction of the overall dunes being visible.

5.1.3 Venusian Feature Fields

The previously identified venusian feature fields are found on volcanic plains with multiple craters in their vicinity (Greeley et al., 1992; Weitz et al., 1994; Rader et al., 2019).

Each locality has linear features that occur in a regularly spaced field with minimal sinuosity and a narrow width. Both Aglaonice and Fortuna-Meskhent have more distinct features in the fields than the Guan Daosheng field or the Stowe field. Aglaonice is speckled and the radar-bright features display a linear pattern that is indicitave of long, narrow linear features, similar to the appearance of the Simpson dunes in SAR at 75 m/pixel (Figure 5.1.3.1). This adds to previous work suggesting these features are dunes, specifically linear dunes populating the Aglaonice field.

Fortuna-Meskhent has many linear features that are less speckled than Aglaonice and a greater density of features. The population of features observed are narrow and long with a moderate level of sinuosity, suggesting that they are a dune-like feature (Figure 5.1.3.2). The linear features of interest are trending NW-SE with an additional population of features trending

NE to SW that have a higher backscatter value. These features have been previously identified by Greeley et al., (1992) to be wind streaks, indicating the presence of a southwesterly wind. By using the wind streaks as short-term wind vanes for the area, the original linear features can be observed to consistently intersect perpendicularly with the wind direction, suggesting that these are transverse dunes.

The potential aeolian field near Mead Crater displays curving linear features in both SAR and JPL-generated topographic data. The features display no sinuosity, though are similarly linear and narrow like the other feature fields (Figure 5.1.3.3). The radar backscatter values for the linear features in the Mead field are brighter than the linear features observed in the Fortuna-

Meskhent field or Aglaonice field, combined with the positive topographic relief pattern observed in the JPL data suggests that the Mead feature field is a yardang field.

The Stowe feature field and the Guan Daosheng feature field both display a moderately radar-bright features on a radar-dark background. The field both occur nearby multiple impact craters, though the features themselves are so small and short that they are nearly indistinguishable (Figure 5.1.3.4 and Figure 5.1.3.5). These field extents have been mapped to the accuracy that is possible at that scale, indicating they are microdunes. These features have been identified with less accuracy than with Aglaonice field, Fortuna-Meskhent field, or Mead field, but indicating aeolian processes forming the features.

5.1.4 Dune Defect Density in Aeolian Fields

For the representative areas mapped in the Simpson Desert and in the Namib Desert that across different resolutions, the loss of visibility of dunes are comparable. By comparing the two terrestrial dune sites, patterns can be observed about resolution differences apparent in the deserts. For each of the six images, the dune defect pairs are plotted against the average dune spacing on a log-log plot (Figure 5.1.4.1). The number of dune defect pairs refers to the number of terminations of a feature that are observed (Werner and Kocurek, 1997). Both locations show that the VIS image had the narrowest average dune spacing, aligning with the notion that the highest resolution images would show smaller dunes more clearly. The overall trend of the graph shows the average dune spacing to increase as the dune defect density to decrease. The 13.5 m/pixel SAR image for both dune fields has the most variability against the trend, suggesting that the difference in data product skews the measurements, rather than the spatial resolution being the only dominant factor. By comparing the VIS data points to the higher resolution SAR

41 data points, it shows that interpreting radar images can introduce significant uncertainty into the interpretation. The rest of the points are from Ewing et al., 2006, their Figure 8B. The Simpson

Desert and Namib Desert points continue with the general negative trend seen in Ewing et al.,

2006, though there is an offset with the data from this study in both the average dune spacing and the dune defect density. This is likely due to a different in mapping techniques and the representative areas chosen based on the data. The dune defect density for each location shows the density of the population of the dunes that formed. The Simpson Desert have greater overall defect density values, showing that the dunes formed tightly with minimal unused inter-dune space. The area of the Namib Desert used in this study have lower overall dune defect density values than the Simpson Desert, showing that features that formed near the edge of a dune field will be more dispersed and with a greater ratio of inter-dune space to dunes. The plot shows that poorer spatial resolution influences the dune spacing observed in the images, while the difference between radar imagery and optical imagery affects the dune defect density.

5.1.5 Venusian Feature Orientation

Out of the identified locations of Venus, the Aglaonice feature field, a representative area of the Fortuna-Meskhent feature field, and the Mead feature field in both Magellan SAR and the

JPL-generated topography were all used to generate rose diagrams of the mapped feature orientations. The diagrams can be compared to illustrate the dominant wind direction patterns, showing the quantity and preference for different angles. The Mead features in Magellan SAR show a greater spread from the two subgroups of features being used (Figure 5.1.5.1). The Mead features in the topographic data show that northeast-southwest orientation is the most dominant.

The Aglaonice field has the fewest observable features and the largest spread of orientations,

42 showing a correlation between number of features and the dominance of a single feature direction (Figure 5.1.5.2). The Fortuna-Meskhent field rose diagram is more concentrated than the Aglaonice field, with the most east-west orientation of all the localities (Feature 5.1.5.3). The feature orientation is important for aeolian fields because the dominant, structure-forming wind direction can be inferred based on the type of aeolian feature is observed.

Aglaonice field shows the greatest degree of variability with the angle, showing the effect of aeolian features that have formed in a smaller field and are forming with lower quantities of available sediment or nearby resistant topography. The wind pattern required for linear dunes are bidirectional winds with low to moderate quantities of sediment, furthering the possibility that the observed features in the Aglaonice feature field are linear dunes. The representative area of features in the Fortuna-Meskhent field appear to be more east-south-east trending, which is perpendicular to the north-south oriented wind streaks observed in that region and mapped on a global wind vector map (Greeley et al., 1992). The orthogonal relationship the features display to the pre-determined wind vectors suggests the presence of transverse dunes in the Fortuna-

Meskhent field. The two groups of features in the Mead feature field both display a high level of uniformity, with the features roughly parallel with observable curvature within each subset. The uniformity of spacing and consistency with the high elevation seen in the topographic data suggest that the positive relief features are formed through erosion, rather than deposition. On

Venus, there are few surface processes that erode away material, leaving behind large, positive relief fields of linear structures. This suggests that the features observed in the Mead feature field in both Magellan SAR and the JPL-generated topography are yardang sets.

5.2 Diagnostic Characteristics of Linear Features Observed in Synthetic Aperture Radar Images

The linear dune crestlines in both the Simpson and Namib Deserts in the SAR images have clear similarities, despite having different formation histories and conditions. The observed radar features are longer than they are wide, with a generally narrow wavelength that occur in fields with a moderate degree of sinuosity. These characteristics have been summarized in Figure

5.2.1 comparing their occurrences with different geomorphologic features that display a similar linear pattern. The characteristics that are most sensitive to a resolution change are the wavelength width and sinuosity level. The characteristics listed in Figure 5.2.1 that are the least sensitive are feature isolation and linearity.

The different surface features identified are linear dunes, transverse dunes, yardangs, wind streaks, tectonic fractures, and wrinkle ridges. These features have been selected as similar potential identifications of the linear features observed in radar. While both dune types and yardangs are aeolian structures, a yardang field would be expected to more clearly show the features that formed and would not have any bifurcation or merging of features (e.g., Goudie,

2007; de Silva et al., 2010). The outgroup testing that has been done produced the truth table of observable characteristics in SAR images that have been compared to the original venusian features. Feature identifications and discriminations of the venusian localities are further discussed in section 6.

5.3 Implications for Dune Identification

5.3.1 Identifying Dune Fields in Synthetic Aperture Radar Images

Dunes and dune fields imaged in SAR display diagnostic characteristics detailed in the previous section that can be used as a qualitative analysis of future potential aeolian sites. The

44 linear features in the Simpson Desert are radar-bright on a radar-dark background, while the same dune crestlines in the Namib Desert are radar-dark on a radar-bright background, indicating that the radar brightness value is highly dependent on the look angle. This characteristic illustrates how the appearance of the dunes can change, even in the same sand sea (Figure

5.3.1.1). However, the features found in both the Simpson Desert and the Namib Desert have several similar diagnostic characteristics that are not reliant on the radar brightness values. The results for the SAR images at 13.5 m/pixel and 75 m/pixel show that the dune defects can be obfuscated and more difficult to interpret at the lower resolution. This problem shows that the higher the resolution the radar data, the less likely the patterns will be interpreted easily with key diagnostic characteristics.

5.3.2 Future Identification of Venusian Dune Fields in Magellan SAR Images

The identified venusian aeolian sites have shown that there is a diversity in how they can appear in radar, however they are all located near multiple impact craters. The four venusian aeolian sites identified are overlapping with the impact ejecta parabolas from a map produced by

M. Gilmore (Figure 5.3.2.1). The sites also all occur in high density ejecta parabola areas, suggesting that these craters are supplying the sediment used to construct the bedforms. This also indicates that future aeolian sites could be located in regions of similarly high-density impact ejecta parabolas. The results have shown that the aeolian fields were all radar-bright features over a radar-dark background and were first identified as potential candidates from that pattern, however the research shows that more aeolian sites could exist but be hidden due to the angle of the features relative to the look angle.

For a planet that is roughly Earth-sized, it is considered surprising that Venus has so few identified aeolian sites. This work shows that more aeolian fields may exist at orientations too close to parallel or perpendicular to the look direction of the Magellan spacecraft that they did not appear in the data. The five identified aeolian sites are also large, comparable to the largest dune fields and ergs present on Earth. It is likely that there are numerous smaller aeolian fields that are below the range of identification resolution, particularly within impact crater parabolas.

6. Summary

This study explored the appearance of sand dunes in remotely sensed images in both VIS and SAR imagery at 0.5 m/pixel, 13.5 m/pixel, and 75 m/pixel spatial resolution with venusian examples and terrestrial analogs. The feature fields on Venus were suggested to be evidence of mega-dunes, without specific characteristics that indicate the identification of the dunes shown from the Magellan SAR images. To better describe the appearance of km-sized dunes on Venus, two massive terrestrial dune fields were used as analogs, the Simpson Desert in central Australia and the Namib Sand Sea on the western coast of Namibia. The results for both locations were compared at all three resolutions, showing the increase in uniformity with lower resolutions and the decrease in the number of mappable features.

The feature fields on Venus display similar characteristics to the terrestrial dunes mapped in 75 m/pixel resolution SAR data. Aglaonice field has long, moderately radar-bright features that are only a few pixels in width with no bifurcation visible. The features appear similar to the linear dunes mapped in the Simpson Desert and also match with the diagnostic characteristics for linear dunes (Figure 5.2.1). This further indicates that the features in Aglaonice field are linear dunes. Fortuna-Meskhent field has long, sinuous, moderately radar-bright features with no visible bifurcation and wind streaks occurring orthogonally to the features. This suggests the observed features are transverse dunes, similar to the transverse dunes mapped on the coast of the Namib Desert. The Guan Daosheng field displays small features that are barely distinguishable, similar to the Stowe field. The features appear as a moderately radar-bright field over a radar-dark volcanic plain, suggesting they are microdunes, a term suggested by Greeley et al., (1992) and Weitz et al., (1994). The Mead field appeared different, with no sinuosity and two sets of features that were similar yet distinct. The field also contained features that had distinct

47 edges, rather than the more diffuse edges that the other venusian fields displayed. These features appear to be different than the other venusian feature fields examined in this work but still oriented in a consistent direction. This suggests that they are aeolian features, but yardangs instead of dunes. The five venusian locations studied show a variety of aeolian features, which can provide significant insight to the sediment on Venus that is available for transport and characteristics of likely areas that future aeolian fields can be found in.

This work shows that with higher resolution SAR data, more dune defects and bifurcations would be visible in the images. The average observed dune spacing would be decreased and a higher level of sinuosity could be expected, along with a greater number of potential aeolian sites being identified at a higher resolution. The majority of the aeolian fields on Venus were oriented closer to north-south trending than east-west trending, which could imply a difficulty in identifying features that only exist orthogonal to the poles due to the north- south polar orbit of the Magellan spacecraft. The correlation between venusian aeolian localities and the impact crater ejecta parabolas may indicate that Venus could be sediment limited, with greater quantities of available sand found surrounding the ejecta parabolas. Future missions to

Venus have proposed global 30 m/pixel spatial resolution SAR data (e.g., Smrekar et al., 2016) which will hopefully provide opportunities to observe more aeolian features and answer more questions about the sediment transport processes.

LIST OF REFERENCES

Al-Hashemi, H.M.B. and Al-Amoudi, O.S.B., 2018. A review on the angle of repose of granular materials. Powder Technology, 330, pp.397-417.

Argenti, F., Lapini, A., Bianchi, T. and Alparone, L., 2013. A tutorial on speckle reduction in synthetic aperture radar images. IEEE Geoscience and remote sensing magazine, 1(3), pp.6-35.

Arvidson, R.E., Greeley, R., Malin, M.C., Saunders, R.S., Izenberg, N., Plaut, J.J., Stofan, E.R. and Shepard, M.K., 1992. Surface modification of Venus as inferred from Magellan observations of plains. Journal of Geophysical Research: Planets, 97(E8), pp.13303-13317.

Avduevskij, V.S., Vishnevetskij, S.L., Golov, I.A., Karpejskij, Y.Y., Lavrov, A.D., Likhushin, V.Y., Marov, M.Y., Mel'Nikov, D.A., Pomogin, N.I., Pronina, N.N. and Razin, K.A., 1976. Measurement of wind velocity on the surface of Venus during operation of the Venera 9 and Venera 10 space probes. Kosmicheskie Issledovaniia, 14, pp.710-713.

Bagnold, R.A., 2012. The physics of blown sand and desert dunes. Courier Corporation.

Bamler, R. and Hartl, P., 1998. Synthetic aperture radar interferometry. Inverse problems, 14(4), p.R1.

Basilevsky, A.T. and Head, J.W., 2003. The surface of Venus. Reports on Progress in Physics, 66(10), p.1699.

Berger, M., Moreno, J., Johannessen, J.A., Levelt, P.F. and Hanssen, R.F., 2012. ESA's sentinel missions in support of Earth system science. Remote Sensing of Environment, 120, pp.84-90.

Bindschadler, D.L., DeCharon, A., Beratan, K.K., Smrekar, S.E. and Head, J.W., 1992. Magellan observations of Alpha Regio: Implications for formation of complex ridged terrains on Venus. Journal of Geophysical Research: Planets, 97(E8), pp.13563-13577.

Blackwelder, E., 1934. Yardangs. Bulletin of the Geological Society of America, 45(1), pp.159- 166.

Bo, T.L. and Zheng, X.J., 2011. The formation and evolution of aeolian dune fields under unidirectional wind. Geomorphology, 134(3-4), pp.408-416.

Bondarenko, N.V., Kreslavsky, M.A. and Head, J.W., 2006. North‐south roughness anisotropy on Venus from the Magellan Radar Altimeter: Correlation with geology. Journal of Geophysical Research: Planets, 111(E6).

Bondarenko, N.V. and Kreslavsky, M.A., 2018. Surface properties and surficial deposits on Venus: New results from Magellan radar altimeter data analysis. Icarus, 309, pp.162-176.

Bristow, C.S., Bailey, S.D. and Lancaster, N., 2000. The sedimentary structure of linear sand dunes. Nature, 406(6791), p.56.

Bristow, C.S., Duller, G.A.T. and Lancaster, N., 2007. Age and dynamics of linear dunes in the Namib Desert. Geology, 35(6), pp.555-558.

Brookfield, M.E., 2011. Aeolian processes and features in cool climates. Geological Society, London, Special Publications, 354(1), pp.241-258.

Campbell, D.B., Stacy, N.J.S., Newman, W.I., Arvidson, R.E., Jones, E.M., Musser, G.S., Roper, A.Y. and Schaller, C., 1992. Magellan observations of extended impact crater related features on the surface of Venus. Journal of Geophysical Research: Planets, 97(E10), pp.16249-16277.

Chamberlin, T.C., 1897. The method of working multiple hypotheses. J. Geol, 5, pp.837-848.

Claudin, P. and Andreotti, B., 2006. A scaling law for aeolian dunes on Mars, Venus, Earth, and for subaqueous ripples. Earth and Planetary Science Letters, 252(1-2), pp.30-44.

Colombini, M., 2004. Revisiting the linear theory of sand dune formation. Journal of Fluid Mechanics, 502, pp.1-16. de Silva, S.L., Bailey, J.E., Mandt, K.E. and Viramonte, J.M., 2010. Yardangs in terrestrial ignimbrites: Synergistic remote and field observations on Earth with applications to Mars. Planetary and Space Science, 58(4), pp.459-471.

Diniega, S., Kreslavsky, M., Radebaugh, J., Silvestro, S., Telfer, M. and Tirsch, D., 2017. Our evolving understanding of aeolian bedforms, based on observation of dunes on different worlds. Aeolian research, 26, pp.5-27. du Pont, S.C., 2015. Dune morphodynamics. Comptes Rendus Physique, 16(1), pp.118-138.

Ewing, R.C., Kocurek, G. and Lake, L.W., 2006. Pattern analysis of dune‐field parameters. Earth Surface Processes and Landforms: The Journal of the British Geomorphological Research Group, 31(9), pp.1176-1191.

Ewing, R.C., Peyret, A.P.B., Kocurek, G. and Bourke, M., 2010. Dune field pattern formation and recent transporting winds in the Olympia Undae Dune Field, north polar region of Mars. Journal of Geophysical Research: Planets, 115(E8).

Ewing, R.C., Hayes, A.G. and Lucas, A., 2015. Sand dune patterns on Titan controlled by long- term climate cycles. Nature Geoscience, 8(1), pp.15-19.

Ewing, R.C., McDonald, G.D. and Hayes, A.G., 2015. Multi-spatial analysis of aeolian dune- field patterns. Geomorphology, 240, pp.44-53.

Folk, R.L., 1971. Longitudinal dunes of the northwestern edge of the Simpson Desert, Northern Territory, Australia, 1. Geomorphology and grain size relationships. Sedimentology, 16(1‐2), pp.5-54.

Ford, J.P., Blom, R.G., Crisp, J.A., Elachi, C., Farr, T.G., Saunders, R.S., Theilig, E.E., Wall, S.D. and Yewell, S.B., 1989. Spaceborne radar observations: A guide for Magellan radar-image analysis.

Foumelis, M., Blasco, J.M.D., Desnos, Y.L., Engdahl, M., Fernández, D., Veci, L., Lu, J. and Wong, C., 2018, July. ESA SNAP-StaMPS Integrated Processing for Sentinel-1 Persistent Scatterer Interferometry. In IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium (pp. 1364-1367). IEEE.

Gilmore, M.S., 2016, March. Which Venus Tesserae Are Most Pristine?. In Lunar and Planetary Science Conference (Vol. 47, p. 2621).

Goudie, A.S., 2007. Mega‐yardangs: A global analysis. Geography Compass, 1(1), pp.65-81.

Greeley, R., Iversen, J.D., Pollack, J.B., Udovich, N. and White, B., 1974. Wind tunnel studies of Martian aeolian processes. Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 341(1626), pp.331-360.

Greeley, R., Iversen, J., Leach, R., Marshall, J., White, B. and Williams, S., 1984a. Windblown sand on Venus: Preliminary results of laboratory simulations. Icarus, 57(1), pp.112-124.

Greeley, R., Marshall, J.R. and Leach, R.N., 1984b. Microdunes and other aeolian bedforms on Venus: Wind tunnel simulations. Icarus, 60(1), pp.152-160.

Greeley, R. and Arvidson, R.E., 1990. Aeolian processes on Venus. Earth, Moon, and Planets, 50(1), pp.127-157.

Greeley, R., Arvidson, R.E., Elachi, C., Geringer, M.A., Plaut, J.J., Saunders, R.S., Schubert, G., Stofan, E.R., Thouvenot, E.J., Wall, S.D. and Weitz, C.M., 1992. Aeolian features on Venus: preliminary Magellan results. Journal of Geophysical Research: Planets, 97(E8), pp.13319- 13345.

Greeley, R., Skypeck, A. and Pollack, J.B., 1993. Martian aeolian features and deposits: Comparisons with general circulation model results. Journal of Geophysical Research: Planets, 98(E2), pp.3183-3196.

Greeley, R., Schubert, G., Limonadi, D., Bender, K.C., Newman, W.I., Thomas, P.E., Weitz, C.M. and Wall, S.D., 1994. Wind streaks on Venus: clues to atmospheric circulation. Science, 263(5145), pp.358-361.

Greeley, R., Bender, K., Thomas, P.E., Schubert, G., Limonadi, D. and Weitz, C.M., 1995. Wind-related features and processes on Venus: Summary of Magellan results. Icarus, 115(2), pp.399-420.

Hansen, V.L. and López, I., 2010. Venus records a rich early history. Geology, 38(4), pp.311- 314.

Hare, T.M., Keszthelyi, L., Gaddis, L. and Kirk, R.L., 2014, March. Online planetary data and services at USGS astrogeology. In Lunar and Planetary Science Conference (Vol. 45, p. 2487).

Head, J.W., Crumpler, L.S., Aubele, J.C., Guest, J.E. and Saunders, R.S., 1992. Venus volcanism: Classification of volcanic features and structures, associations, and global distribution from Magellan data. Journal of Geophysical Research: Planets, 97(E8), pp.13153-13197.

Hollands, C.B., Nanson, G.C., Jones, B.G., Bristow, C.S., Price, D.M. and Pietsch, T.J., 2006. Aeolian–fluvial interaction: evidence for Late Quaternary channel change and wind-rift linear dune formation in the northwestern Simpson Desert, Australia. Quaternary Science Reviews, 25(1-2), pp.142-162.

Hornacek, M., Wagner, W., Sabel, D., Truong, H.L., Snoeij, P., Hahmann, T., Diedrich, E. and Doubková, M., 2012. Potential for high resolution systematic global surface soil moisture retrieval via change detection using Sentinel-1. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 5(4), pp.1303-1311.

Hu, Y., Janowicz, K., Prasad, S. and Gao, S., 2015. Enabling semantic search and knowledge discovery for ArcGIS Online: A linked-data-driven approach. In AGILE 2015 (pp. 107-124). Springer, Cham.

Hugenholtz, C.H., Levin, N., Barchyn, T.E. and Baddock, M.C., 2012. Remote sensing and spatial analysis of aeolian sand dunes: A review and outlook. Earth-science reviews, 111(3-4), pp.319-334.

Ivanov, M.A., 1988, March. The results of morphometric study of the tessera terrain of Venus from Venera 15/16 data. In Lunar and Planetary Science Conference (Vol. 19).

Iversen, J.D. and White, B.R., 1982. Saltation threshold on earth, mars and venus. Sedimentology, 29(1), pp.111-119.

Iversen, J.D., Pollack, J.B., Greeley, R. and White, B.R., 1976. Saltation threshold on Mars: the effect of interparticle force, surface roughness, and low atmospheric density. Icarus, 29(3), pp.381-393.

Johnson, W.T., 1991. Magellan imaging radar mission to Venus. Proceedings of the IEEE, 79(6), pp.777-790.

Kocurek, G. and Ewing, R.C., 2005. Aeolian dune field self-organization–implications for the formation of simple versus complex dune-field patterns. Geomorphology, 72(1-4), pp.94-105.

Kreslavsky, M.A. and Basilevsky, A.T., 1998. Morphometry of wrinkle ridges on Venus: Comparison with other planets. Journal of Geophysical Research: Planets, 103(E5), pp.11103- 11111.

Kreslavsky, M. A. and Bondarenko, N. V., 2017. Aeolian sand transport and aeolian deposits on Venus: A review. Aeolian Research, 26, pp. 29-46.

Kwok, R. and Johnson, W.T., 1989. Block adaptive quantization of Magellan SAR data. IEEE Transactions on Geoscience and remote sensing, 27(4), pp.375-383.

Lancaster, N., 1982. Linear dunes. Progress in Physical Geography, 6(4), pp.475-504.

Lauknes, T.R., 2011. Rockslide mapping in Norway by means of interferometric SAR time series analysis.

Lee, J.S., 1981. Speckle analysis and smoothing of synthetic aperture radar images. Computer graphics and image processing, 17(1), pp.24-32.

Lorenz, R.D., Lunine, J.I., Grier, J.A. and Fisher, M.A., 1995. Prediction of aeolian features on planets: Application to Titan paleoclimatology. Journal of Geophysical Research: Planets, 100(E12), pp.26377-26386.

Lorenz R. D., 2015. Heights of Furtuna-Meshknet Dunes (Al-Uzza Undae), Venus, From Magellan Radarclinometry, International Planetary Dunes Workshop, 8004.

Lumay, G., Fiscina, J., Ludewig, F. and Vandewalle, N., 2013, June. Influence of cohesive forces on the macroscopic properties of granular assemblies. In AIP Conference Proceedings (Vol. 1542, No. 1, pp. 995-998). AIP.

Malenovský, Z., Rott, H., Cihlar, J., Schaepman, M.E., García-Santos, G., Fernandes, R. and Berger, M., 2012. Sentinels for science: Potential of Sentinel-1,-2, and-3 missions for scientific observations of ocean, cryosphere, and land. Remote Sensing of Environment, 120, pp.91-101.

Marshall, J.R. and Greeley, R., 1992. An experimental study of aeolian structures on Venus. Journal of Geophysical Research: Planets, 97(E1), pp.1007-1016.

Moreira, A., 2013. Synthetic aperture radar (SAR): principles and applications. 4th Advanced Training Course in Land Remote Sensing.

Mueller, N., Helbert, J., Hashimoto, G.L., Tsang, C.C.C., Erard, S., Piccioni, G. and Drossart, P., 2008. Venus surface thermal emission at 1 μm in VIRTIS imaging observations: Evidence for variation of crust and mantle differentiation conditions. Journal of Geophysical Research: Planets, 113(E5).

Neakrase, L.D.V., 2015, May. Terrestrial Analogs for Venusian Dunes: Sub-Aqueous, Seafloor Dune Fields. In Fourth Annual International Planetary Dunes Workshop (Vol. 1843, p. 8023).

Neakrase, L.D., Klose, M. and Titus, T.N., 2017. Terrestrial subaqueous seafloor dunes: Possible analogs for Venus. Aeolian research, 26, pp.47-56.

Nunes, D.C., Mitchell, K.L., Cotton, K.J., Toole, N., Hensley, S. and Deen, R., 2018, March. Robust, Automated Stereogrammetry of Venus Magellan SAR Imagery and Preliminary Tessera Results. In Lunar and Planetary Science Conference (Vol. 49).

Nunes, D.C., Smrekar, S.E., Mueller, N., Stock, J., Davaille, A., Hensley, S., Cotton, K. and Mitchell, K., 2020. High-Resolution Stereo Topography Re-Examination of Central Artemis, Venus. LPI, (2326), p.2734.

Paillou, P., Seignovert, B., Radebaugh, J. and Wall, S., 2016. Radar scattering of linear dunes and mega-yardangs: Application to Titan. Icarus, 270, pp.211-221.

Petropoulos, B., 1988. Physical parameters of the atmosphere of Venus. Earth, Moon, and Planets, 42(1), pp.29-40.

Rader, L.X., Thomson, B.J. and McCanta, M.C., 2019, March. A Review of Surface-Atmosphere Interactions on Venus via Observable Aeolian Features. In Lunar and Planetary Science Conference (Vol. 50).

Reffet, E., Courrech du Pont, S., Hersen, P. and Douady, S., 2010. Formation and stability of transverse and longitudinal sand dunes. Geology, 38(6), pp.491-494.

Saunders, R.S., Pettengill, G.H., Arvidson, R.E., Sjogren, W.L., Johnson, W.T.K. and Pieri, L., 1990. The Magellan Venus radar mapping mission. Journal of Geophysical Research: Solid Earth, 95(B6), pp.8339-8355.

Saunders, R.S., Spear, A.J., Allin, P.C., Austin, R.S., Berman, A.L., Chandlee, R.C., Clark, J., Decharon, A.V., De Jong, E.M., Griffith, D.G. and Gunn, J.M., 1992. Magellan mission summary. Journal of Geophysical Research: Planets, 97(E8), pp.13067-13090.

Schumm, S.A. and Schumm, S.A., 1998. To interpret the Earth: ten ways to be wrong. Cambridge University Press.

Seely, M.K., 1978. The Namib dune desert: an unusual ecosystem. Journal of Arid Environments, 1(2), pp.117-128.

Shao, Y. and Lu, H., 2000. A simple expression for wind erosion threshold friction velocity. Journal of Geophysical Research: Atmospheres, 105(D17), pp.22437-22443.

Smrekar, S., Dyar, M., Hensley, S. and Helbert, J., 2016, October. VERITAS (Venus Emissivity, Radio Science, InSAR, Topo-graphy And Spectroscopy): A Proposed Discovery Mission. In AAS/Division for Planetary Sciences Meeting Abstracts# 48 (Vol. 48).

Solomon, S.C., Smrekar, S.E., Bindschadler, D.L., Grimm, R.E., Kaula, W.M., McGill, G.E., Phillips, R.J., Saunders, R.S., Schubert, G., Squyres, S.W. and Stofan, E.R., 1992. Venus tectonics: An overview of Magellan observations. Journal of Geophysical Research: Planets, 97(E8), pp.13199-13255.

Taylor, F.W., Svedhem, H. and Head, J.W., 2018. Venus: the atmosphere, climate, surface, interior and near-space environment of an Earth-like planet. Space Science Reviews, 214(1), p.35.

Torres, R., Snoeij, P., Geudtner, D., Bibby, D., Davidson, M., Attema, E., Potin, P., Rommen, B., Floury, N., Brown, M. and Traver, I.N., 2012. GMES Sentinel-1 mission. Remote Sensing of Environment, 120, pp.9-24.

Trego, K.D., 1990. The absence of yardangs on Venus. Earth, Moon, and Planets, 49(3), pp.283- 284.

Trego, K.D., 1992. Yardang identification in Magellan imagery of Venus. Earth, Moon, and Planets, 58(3), pp.289-290.

Tsoar, H., 1978. The Dynamics of Longitudinal Dunes. BEN-GURION UNIV OF THE NEGEV BEERSHEBA (ISRAEL) DEPT OF GEOGRAPHY.

Tsoar, H., 1982. Internal structure and surface geometry of longitudinal (seif) dunes. Journal of Sedimentary Research, 52(3), pp.823-831.

Tsoar, H., 1989. Linear dunes-forms and formation. Progress in Physical Geography, 13(4), pp.507-528.

Tsoar, H., 2001. Types of aeolian sand dunes and their formation. In Geomorphological fluid mechanics (pp. 403-429). Springer, Berlin, Heidelberg.

Twidale, C.R., 1972. Evolution of sand dunes in the Simpson Desert, central Australia. Transactions of the Institute of British Geographers, pp.77-109.

Wall, S.D., Mcconnell, S.L., Leff, C.E., Austin, R.S., Beratan, K.K. and Rokey, M.J., 1995. User guide to the Magellan synthetic aperture radar images.

Wang, Z.T., Wang, H.T., Niu, Q.H., Dong, Z.B. and Wang, T., 2011. Abrasion of yardangs. Physical Review E, 84(3), p.031304.

Wang, J., Xiao, L., Reiss, D., Hiesinger, H., Huang, J., Xu, Y., Zhao, J., Xiao, Z. and Komatsu, G., 2018. Geological features and evolution of Yardangs in the Qaidam Basin, Tibetan Plateau

(NW China): a terrestrial analogue for Mars. Journal of Geophysical Research: Planets, 123(9), pp.2336-2364.

Ward, A.W., 1979. Yardangs on Mars: Evidence of recent wind erosion. Journal of Geophysical Research: Solid Earth, 84(B14), pp.8147-8166.

Ward, A.W., Doyle, K.B., Helm, P.J., Weisman, M.K. and Witbeck, N.E., 1985. Global map of eolian features on Mars. Journal of Geophysical Research: Solid Earth, 90(B2), pp.2038-2056.

Weitz, C.M., Plaut, J.J., Greeley, R. and Saunders, R.S., 1994. Dunes and microdunes on Venus: Why were so few found in the Magellan data?. Icarus, 112(1), pp.282-295.

Werner, B.T. and Kocurek, G., 1997. Bed-form dynamics: Does the tail wag the dog?. Geology, 25(9), pp.771-774.

Werner, B.T. and Kocurek, G., 1999. Bedform spacing from defect dynamics. Geology, 27(8), pp.727-730.

Williams, S.H. and Greeley, R., 1994. Windblown sand on Venus: The effect of high atmospheric density. Geophysical research letters, 21(25), pp.2825-2828.

Xing, M., Wu, C. and Adams, M.J., 2011. The saltations of different sized particles in aeolian sand transport. Geomorphology, 134(3-4), pp.217-223.

Zimbelman, J.R. and Griffin, L.J., 2010. HiRISE images of yardangs and sinuous ridges in the lower member of the Medusae Fossae Formation, Mars. Icarus, 205(1), pp.198-210.

Zolotov, M.Y., 2018. Gas–solid interactions on Venus and other Solar System bodies. Reviews in Mineralogy and Geochemistry, 84(1), pp.351-392.

APPENDIX

Figure 2.1. 1 A schematic diagram of synthetic aperture radar (SAR) instrument.

The satellite depicted is interacting with the ground from the air and is detecting a single spot on the ground from the emitted pulses that become pixels in the radar image. Figure from Lauknes,

2011.

Figure 2.7.1. 1 A diagram of different named dune types.

Different named dune types that are formed from variations in the available quantity of sediment and/or varying wind directions. Barchan dunes are also sometimes referred to as crescentic dunes. Longitudinal dunes are also referred to as linear or seif dunes. Figure from Nichols, 2009.

Figure 2.7.2. 1 A yardang field in the Kumtag Desert.

An aerial image of a yardang field bordering the Kumtag Desert in Xinjiang, China. Figure from

Dong et al., 2012.

Figure 2.7.3. 1 An outline of the Simpson Desert.

The Simpson Desert field extent mapped in blue shown in context of Australia to illustrate the geographic location and scale.

Figure 2.7.4. 1 The Namib Desert.

The Namib Desert field extent mapped in blue shown in context of the southwestern coast of

Africa to show geographic location and scale.

A B

Figure 2.7.5. 1 Theoretical and experimental data for sediment transport on Venus.

(A) Threshold data from the Venus Wind Tunnel compared to theoretical extrapolation (Iversen and White, 1982) using venusian atmospheric conditions and Earth gravity. Test conditions are:

CO2 gas, density 0.0664 g/cm3, kinematic viscosity is 0.00230 cm2/sec, particle density is 2.65 g/cm3, Earth gravity. (B) Threshold data from the Venus Wind Tunnel corrected for venusian gravity compared to theoretical extrapolation, gas density is 0.0646 g/cm3, kinematic viscosity is

0.0044 cm2/sec, particle density is 2.65 g/cm3, Venus gravity is 877 cm/sec2. Figure and caption from Greeley et al., 1984.

Figure 2.8. 1 A graph describing feature identification resolutions.

A graph showing the relationship between a feature object size and the sun elevation in terms of the ability to identify or detect surface features. Figure from Zimbelman, 2001.

Figure 3.1. 1 The Sentinel-1B data download software.

The ESA Copernicus software program showing the footprints of pre-selected images of the

Namib Desert.

Figure 3.1. 2 End four-character tag of Sentinel-1B SAR images used.

A B

Figure 3.2. 1 An example of the mapping the 13.5 m/pixel SAR Sentinel-1B imagery.

A representative area of the Simpson Desert in Australia showing how the dune crestlines are mapped in the 13.5 m/pixel SAR image. (A) is the area in the 13.5 m/pixel SAR image and (B) is the 13.5 m/pixel SAR image with the linear features mapped.

Figure 3.2. 2 Terrestrial feature field characteristics.

Figure 3.3. 1 A map of potential aeolian fields on Venus.

A map of Venus in the left look Magellan synthetic aperture radar data showing the locations of the five feature fields of interest.

Figure 3.3. 2 Venusian feature field characteristics.

Figure 3.4. 1 An example of mapping multiple field extents in SAR imagery.

A representative area of Aglaonice field showing multiple field extents mapped in the 75 m/pixel

Magellan SAR image. The uncertainty to how far the features extended and if their appearance would change with different sizes was addressed by having two field extents mapped, showing both a generous and conservative estimate.

Figure 3.5. 1 The topographic data from the Jet Propulsion Laboratory.

Topographic data from JPL of the Mead Crater feature field layered over the Magellan SAR image at 50% transparency. The field extent mapped in the Magellan SAR data is shown in green, truncating at the SAR data gaps.

Figure 4.1.1. 1 Bifurcation of the Simpson Desert in VIS at 0.5 m/pixel.

Enlarged area of the Simpson Desert in VIS at 0.5 m/pixel. Arrow indicates bifurcation of a dune.

Figure 4.1.1. 2 Area of the Simpson Desert in VIS.

Representative area of the Simpson Desert in VIS at 0.5 m/pixel without crestlines mapped.

Figure 4.1.1. 3 Area of Simpson Desert mapped in VIS.

Representative area of the Simpson Desert in VIS at 0.5 m/pixel with dune crestlines mapped in pink.

Figure 4.1.1. 4 All mapped features in the Simpson Desert in the VIS image.

All mapped features in this study in the Simpson Desert in VIS at 0.5 m/pixel with dune crestlines mapped in pink.

Figure 4.1.2. 1 Area of the Simpson Desert mapped in high resolution SAR.

Representative area of the Simpson Desert in SAR at 13.5 m/pixel with dune crestlines mapped in blue.

Figure 4.1.2. 2 Area of the Simpson Desert in high resolution SAR.

Representative area of the Simpson Desert in SAR at 13.5 m/pixel.

Figure 4.1.2. 3 All mapped features in the Simpson Desert in high resolution SAR image.

All mapped features in this study in the Simpson Desert in SAR at 13.5 m/pixel with dune crestlines mapped in blue.

Figure 4.1.3. 1 Area of the Simpson Desert in low resolution SAR.

Representative area of the Simpson Desert in SAR at 75 m/pixel.

Figure 4.1.3. 2 Area of the Simpson Desert mapped in low resolution SAR.

Representative area of the Simpson Desert in SAR at 75 m/pixel with dune crestlines mapped in yellow.

Figure 4.1.3. 3 All mapped features in the Simpson Desert in low resolution SAR image.

All mapped features in this study in the Simpson Desert in SAR at 75 m/pixel with dune crestlines mapped in yellow.

Figure 4.2.1. 1 Area of the Namib Desert in VIS.

Representative area of the linear dunes in the Namib Desert in VIS at 0.5 m/pixel.

Figure 4.2.1. 2 Area of the Namib Desert mapped in VIS.

Representative area of the linear dunes in the Namib Desert in VIS at 0.5 m/pixel with the dune crestlines mapped in green.

Figure 4.2.1. 3 All mapped features in the Namib Desert in VIS image.

All mapped features in this study in the Namib Desert in VIS at 0.5 m/pixel with dune crestlines mapped in green.

Figure 4.2.2. 1 Area of the Namib Desert in high resolution SAR.

Representative area of the linear dunes in the Namib Desert in SAR at 13.5 m/pixel.

Figure 4.2.2. 2 Area of the Namib Desert mapped in high resolution SAR.

Representative area of the linear dunes in the Namib Desert in SAR at 13.5 m/pixel with the dune crestlines mapped in blue.

Figure 4.2.2. 3 All mapped features in the Namib Desert in high resolution SAR image.

All mapped features in this study in the Namib Desert in SAR at 13.5 m/pixel with dune crestlines mapped in blue.

Figure 4.2.3. 1 Area of the Namib Desert in low resolution SAR.

Representative area of the linear dunes in the Namib Desert in SAR at 75 m/pixel.

Figure 4.2.3. 2 Area of the Namib Desert mapped in low resolution SAR.

Representative area of the linear dunes in the Namib Desert in SAR at 75 m/pixel with the dune crestlines mapped in orange.

Figure 4.2.3. 3 All mapped features in the Namib Desert in low resolution SAR image.

All mapped features in this study in the Namib Desert in SAR at 75 m/pixel with dune crestlines mapped in orange.

Figure 4.2.4. 1 Coastal area of the Namib Desert in VIS.

Representative area of the transverse dunes in the Namib Desert in VIS at 0.5 m/pixel.

Figure 4.2.4. 2 Coastal area of the Namib Desert mapped in VIS.

Representative area of the transverse dunes in the Namib Desert in VIS at 0.5 m/pixel with the dune crestlines mapped in chartreuse.

Figure 4.2.5. 1 Coastal area of the Namib Desert in high resolution SAR.

Representative area of the transverse dunes in the Namib Desert in SAR at 13.5 m/pixel overlaid on the 0.5 m/pixel VIS image to preserve ocean at coastline for reference.

Figure 4.2.5. 2 Coastal area of the Namib Desert mapped in high resolution SAR.

Representative area of the transverse dunes in the Namib Desert in SAR at 13.5 m/pixel overlaid on the 0.5 m/pixel VIS image to preserve ocean at coastline for reference with the dune crestlines mapped in pink.

Figure 4.2.6. 1 Coastal area of the Namib Desert in low resolution SAR.

Representative area of the transverse dunes in the Namib Desert in SAR at 75 m/pixel overlaid on the 0.5 m/pixel VIS image to preserve ocean at coastline for reference.

Figure 4.2.6. 2 Coastal area of the Namib Desert mapped in low resolution SAR.

Representative area of the transverse dunes in the Namib Desert in SAR at 75 m/pixel overlaid on the 0.5 m/pixel VIS image to preserve ocean at coastline for reference with the dune crestlines mapped in red.

Figure 4.3.1. 1 Aglaonice field mapped in Magellan SAR.

Aglaonice feature field with two field extents mapped in green and a representative area of features mapped in yellow. Data are from Magellan SAR instrument at 75 m/pixel.

Figure 4.3.2. 1 Fortuna-Meskhent field mapped in Magellan SAR.

Fortuna-Meskhent feature field with two field extents mapped in green and a representative area of features mapped in yellow. Data are from Magellan SAR instrument at 75 m/pixel.

100

Figure 4.3.2. 2 Fortuna-Meskhent field with wind streaks atop the features.

Fortuna-Meskhent feature field with a box indicating an area where wind streaks superimposed on the linear features. Data are from Magellan SAR instrument at 75 m/pixel.

101

Figure 4.3.3. 1 Guan Daosheng field mapped in Magellan SAR.

Guan Daosheng feature field with two field extents mapped in green and a representative area of features mapped in yellow. Data are from Magellan SAR instrument at 75 m/pixel.

102

Figure 4.3.4. 1 Stowe field in Magellan SAR.

Stowe feature field with two field extents mapped in green. Data are from Magellan SAR instrument at 75 m/pixel.

103

Figure 4.3.5.1. 1 Mead field mapped in Magellan SAR.

Mead feature field with the field extents mapped in green and a representative population of crestlines mapped in yellow. Data are from Magellan SAR instrument at 75 m/pixel.

104

Figure 4.3.5.2. 1 Mead field mapped in JPL topography.

Mead feature field with the field extents mapped in green and a representative population of crestlines mapped in pink.

105

A B C

D E F

Figure 5.1.1. 1 Comparison of resolutions in the Simpson Desert.

A comparison of the representative area of the Simpson Desert shown in the 0.5 m/pixel VIS image, 13.5 m/pixel SAR image, and 75 m/pixel SAR image. (A) shows the area of the Simpson

Desert mapped in pink of the 0.5 m/pixel VIS image, (B) shows the area mapped in blue of the

13.5 m/pixel SAR image, (C) shows the area mapped in yellow of the 75 m/pixel SAR image,

(D) shows the area unmapped in the 0.5 m/pixel VIS image, (E) shows the area unmapped in the

13.5 m/pixel SAR image, and (F) shows the area unmapped in the 75 m/pixel SAR image.

106

A B C

D E F

Figure 5.1.1. 2 Comparisons of resolutions of bifurcation in the Simpson Desert.

A comparison of an enlarged area of the Simpson Desert shown in the 0.5 m/pixel VIS image,

13.5 m/pixel SAR image, and 75 m/pixel SAR image highlighting a dune defect that was lost in lower resolution. (A) shows the zoomed in area of the Simpson Desert mapped in pink of the 0.5 m/pixel VIS image, (B) shows the area mapped in blue of the 13.5 m/pixel SAR image, (C) shows the area mapped in yellow of the 75 m/pixel SAR image, (D) shows the area unmapped in the 0.5 m/pixel VIS image, (E) shows the area unmapped in the 13.5 m/pixel SAR image, and

(F) shows the area unmapped in the 75 m/pixel SAR image. The white rectangle highlights an area that changes in different resolutions.

107

A B C

D E F

Figure 5.1.2. 1 Comparison of resolutions in the Namib Desert.

A comparison of the representative area of linear dunes in the Namib Desert shown in the 0.5 m/pixel VIS image, 13.5 m/pixel SAR image, and 75 m/pixel SAR image. (A) shows the area of the Namib Desert mapped in green of the 0.5 m/pixel VIS image, (B) shows the area mapped in blue of the 13.5 m/pixel SAR image, (C) shows the area mapped in orange of the 75 m/pixel

SAR image, (D) shows the area unmapped in the 0.5 m/pixel VIS image, (E) shows the area unmapped in the 13.5 m/pixel SAR image, and (F) shows the area unmapped in the 75 m/pixel

SAR image.

108

A B C

D E F

Figure 5.1.2. 2 Comparison of dune defects in the Namib Desert.

A comparison of an enlarged area of the linear dunes in the Namib Desert shown in the 0.5 m/pixel VIS image, 13.5 m/pixel SAR image, and 75 m/pixel SAR image highlighting a dune defect that was lost in lower resolution. (A) shows the area of the Namib Desert mapped in green of the 0.5 m/pixel VIS image, (B) shows the area mapped in blue of the 13.5 m/pixel SAR image, (C) shows the area mapped in orange of the 75 m/pixel SAR image, (D) shows the area unmapped in the 0.5 m/pixel VIS image, (E) shows the area unmapped in the 13.5 m/pixel SAR image, and (F) shows the area unmapped in the 75 m/pixel SAR image.

109

A B C

D E F

Figure 5.1.2. 3 Comparison of zoomed in coastal area of Namib Desert.

A comparison of an enlarged area of the transverse dunes in the Namib Desert shown in the 0.5 m/pixel VIS image, 13.5 m/pixel SAR image, and 75 m/pixel SAR image with the SAR images overlaid on the 0.5 m/pixel VIS image to preserve ocean at coastline for reference, highlighting the dunes lost at lower resolution. (A) shows the zoomed in area of the Namib Desert near the eastern coast mapped in yellow-green of the 0.5 m/pixel VIS image, (B) shows the area mapped in pink of the 13.5 m/pixel SAR image, (C) shows the area mapped in red of the 75 m/pixel SAR image, (D) shows the area unmapped in the 0.5 m/pixel VIS image, (E) shows the area unmapped in the 13.5 m/pixel SAR image, and (F) shows the area unmapped in the 75 m/pixel

SAR image.

110

A B C

D E F

Figure 5.1.2. 4 Comparison of the coastal area of the Namib Desert.

A comparison of the representative area of transverse dunes in the Namib Desert shown in the

0.5 m/pixel VIS image, 13.5 m/pixel SAR image, and 75 m/pixel SAR image with the SAR images overlaid on the 0.5 m/pixel VIS image to preserve ocean at coastline for reference. (A) shows the area of the Namib Desert near the eastern coast mapped in yellow-green of the 0.5 m/pixel VIS image, (B) shows the area mapped in pink of the 13.5 m/pixel SAR image, (C) shows the area mapped in red of the 75 m/pixel SAR image, (D) shows the area unmapped in the

0.5 m/pixel VIS image, (E) shows the area unmapped in the 13.5 m/pixel SAR image, and (F) shows the area unmapped in the 75 m/pixel SAR image.

111

Figure 5.1.3. 1 Features in the Aglaonice field on Venus.

An enlarged area of the Aglaonice feature field highlighting the characteristics of the individual features in Magellan 75 m/pixel SAR image.

112

Figure 5.1.3. 2 Features in the Fortuna-Meskhent field on Venus.

An enlarged area of the Fortuna-Meskhent feature field highlighting the characteristics of the individual features in Magellan 75 m/pixel SAR image.

113

A B

Figure 5.1.3. 3 Features in the Mead field on Venus with an elevation profile.

A comparison of an enlarged area of the Mead feature field highlighting the characteristics of the individual features in Magellan 75 m/pixel SAR image (A) and the JPL topographic data overlaid on the Magellan SAR image(B). The cross section highlighted in blue on (A) and (B) from A–A’ is shown as an elevation profile in (C).

114

Figure 5.1.4. 1 A plot of dune defect density vs. dune spacing.

Average dune spacing in meters on a log scale is plotted against dune defect density on a log scale for the Simpson Desert and the Namib Desert in 0.5 m/pixel VIS, 13.5 m/pixel SAR, and

75 m/pixel SAR imagery and the data from Ewing et al., 2006.

115

A B

Figure 5.1.5. 1 Rose diagram of Mead field.

A rose diagram of the Mead feature field on Venus in (A) Magellan SAR data at 75 m/pixel spatial resolution using 688 mapped features and (B) the JPL generated topographic data using

70 mapped features.

116

Figure 5.1.5. 2 Rose diagram of Aglaonice field.

A rose diagram of the Aglaonice feature field on Venus in Magellan SAR data at 75 m/pixel spatial resolution using 45 mapped features.

117

Figure 5.1.5. 3 Rose diagram of Fortuna-Meskhent field.

A rose diagram of the Fortuna-Meskhent feature field on Venus in Magellan SAR data at 75 m/pixel spatial resolution using 52 mapped features.

118

Figure 5.2. 1 A truth table of linear feature characteristics observed in multiple resolutions and wavelengths.

119

A B

Figure 5.3.1. 1 Different areas in the Namib Desert in SAR showing radar backscatter differences.

A comparison of two enlarged areas of the linear dunes in the Namib Desert shown in the 0.5 m/pixel VIS image and the 13.5 m/pixel SAR image in two different regions highlighting the change in relative radar backscatter values of the two regions. (A) shows the linear dune crestlines as radar-dark on a radar-bright surface, while (B) shows the linear dune crestlines as radar-bright on a more radar-dark surface.

120

Figure 5.3.2. 1 A map of Venus comparing localities to ejecta parabolas.

A map of Venus with the five aeolian sites highlighted. Data of impact ejecta parabolas are overlaid on the map showing a correlation between their existence and the location of the identified potential aeolian fields.

121

VITA

Louisa Xian-Yue Rader was born in Xianning, China on April 22, 1996 and was adopted at six months old to Sharon Goldstein and Jim Rader and grew up in Northampton, MA. She earned a B.A. cum laude in Astronomy and Dance from Mount Holyoke College in South Hadley, MA in May 2018. She worked with Caleb Fassett at Mount Holyoke College during the summer after her sophomore year mapping craters on the permafrost regions of Mars, where she first became interested in planetary geology and GIS work. She also worked with both Darby Dyar, her undergraduate advisor, from Mount Holyoke College during her junior and senior years running the Laser Induced Breakdown Spectroscopy instrument, and with Brad Thomson for her senior year independent study project. When starting to work with Brad Thomson for her graduate degree, she was given an opportunity to switch planets from Mars to Venus, allowing her to expand her horizons, both literally and figuratively. Louisa has presented her research at the Lunar and Planetary Science Conference every year from 2017 to 2020.

122