Collections of straight, RADAR-bright, linear features, or BLFs, on Saturn's moon Titan are revealed in Cassini SAR (Synthetic Aperture RADAR) images. Most are widely distributed across the northern midlatitudes SAR on SAR swaths T18, T23, T30, T64, and T83 and in swath T56 in the southern midlatitudes. To understand the origin of these features, we compare them with terrestrial yardangs in Dunhuang, China, the Altiplano/Puna of Argentina, and the Lut Desert of Iran and with a similar morphological landform, linear dunes in the Namib Sand Sea, Namibia and on Titan. We apply a statistical classification model developed through random forests, a type of decision tree classification system, grown with terrestrial and titanian training data to the BLFs. To develop the classification, we measured sinuosity, width, spacing, and length for all of the BLFs and their possible terrestrial analogs. We interpret the features in T18, T64-1, and T83 as yardangs based upon morphological similarities between them and features in Iran and Argentina, such as overall SAR brightness, straightness, and lack of branching. Similarities exist between the BLFs and terrestrial yardangs in sinuosity and spacing—sinuosity values range from 1.00 to 1.04 for all the BLFs, and terrestrial yardangs in Iran range from 1.00 to 1.001. A generated statistical model classified a large number of yardangs in T18 and T64-1. In contrast, we interpret the BLFs in T23 and T30 as stabilized linear dunes due to similarities in sinuosity, spacing, and scale with linear dunes in the Namib Sand Sea and Titan swath T3. Stabilized linear dunes may be slightly brighter than the SAR-dark dunes due a change in dielectric constant from introduction of liquids and subsequent stabilization or from the formation of a crust over the top the feature. Sinuosities range from 1.00 to 1.37 in T23 and T30 whereas dunes in the Namib and in T3 range from 1.01 to 1.05. Branching behavior similar to dunes are also observed in BLFs in swaths T23 and T30. The BLF features in T56 in the southern hemisphere we interpret to be dune-related, likely SAR-bright (rough) inter-dune areas. We base this interpretation on the presence of SAR-dark lineations between the BLFs that may be linear dunes. The statistical model classifies few yardangs in T23, T30, and T56. We conclude that statistical classification of these features can be performed. We also show that yardang orientations may aid in the development of global climate and wind models as both current and paleo wind direction indicators.

Yardangs are streamlined ridges carved by the action of wind into consolidated yet erodible substrates. The direction of sediment transport is indicated by their elongate shapes and steep sides which redirects most of the sediment transport layer onto adjacent low-lying surfaces, resulting in lower erosion rates on the elevated areas. Despite this simple premise, the rates of erosion and transport as well as the material properties necessary to form yardangs have until now been largely unknown. This study aims to determine how material properties affect yardang dimensions in order to use yardang morphometry to derive the mechanical properties of a surface that has yet to been explored in situ. As a terrestrial analog for planetary yardangs, we use ignimbrite samples and a Digital Terrain Model (DTM) derived from drone imagery of the Campo Piedra Pomez, Argentina. For comparison, we examine a martian analog using a DTM generated by the HiRISE instrument team for a section of the Medusae Fossae Formation northeast of Aeolis Dorsa. We use the DTMs along with thin sections, porosity, density, and strength measurements of yardang materials to understand the conditions contributing to yardang morphology. This method reveals microscopic evidence of nuanced differences in terrestrial ignimbrite depositional processes that create strong, lightweight, yardang-forming ignimbrite like that suggested to occur in the Medusae Fossae Formation. On average, the CPP ignimbrite samples have 49.51 ± 0.43 percent porosity, density of 1.26 ± 0.13 g/cm^3, and uniaxial compressive strength of 4.88 ± 2.86 MPa. Using the topographic structure of yardangs in the DTMs, we automatically extract yardang polygons and characterize their length, width, height, and spacing in four directions. We ratio these measurements and find that yardang width over minimum crosswind spacing has a geometric mean near one for 4102 terrestrial yardangs (~0.7) as well as for 1269 martian yardangs (~1.3). The ratio of ~1 for closely spaced yardangs is probably caused by increased windspeeds enhancing erosion in the gaps between yardangs until the gaps achieve the same cross-sectional area as the yardangs. Finally, we use regressions of the material properties and morphometry data to suggest that if formation conditions are the same as in the Campo Piedra Pomez, the Medusae Fossae Formation surface would have 52.04 + 1.41 / - 1.37 percent porosity, density of 1.19 ± 0.02 g/cm^3 and strength of 0.64 + 0.84 / -0.36 MPa. These results indicate that topography, porosity, strength, and density attenuate the formation processes that ultimately determine the morphometric properties of yardangs. This establishes a framework on which to make progress towards a quantitative understanding of yardang morphology and evolution. The ArcGIS toolboxes and Python scripts used to obtain our results are available at https://github.com/dmcdoug/yardangtools.