A Textural Image Analysis Study of Mars-Analog Sediment: Comparisons of 2D and 3D Textural Analysis Methods, Image Resolution, and the Effects of Shadows

Sediments from Mars-analog, basaltic, aeolian (Moses Lake dunes, WA) and fluvial (Death Valley region, CA) environments were analyzed for textural properties (grain size, roundness, sphericity) to understand how well image-based analyses compare with actual sediment characteristics. Three interrelated studies determined: (1) how well high-resolution, two-dimensional (2D) image analysis compares with three-dimensional (3D) sediment textures, (2) the extent to which decreasing image resolution affects textural analyses by comparison with high-resolution images, and (3) the extent to which shadows at different incident solar angles has on textural measurements. In both aeolian and fluvial environments, 2D grain-size distributions and statistics closely match 3D properties of sediment. A comparison of grain-size distributions determined from images with different resolutions shows, for both sediments, grain-size means, modes, and sorting can be accurately determined at all resolutions and distribution frequencies and skewness can be estimated to 40 μm/pixel for the aeolian sediment and 80 μm/pixel for the fluvial sediment. A comparison of 2D and 3D sphericity (fluvial sediment only) shows 2D image analysis overestimates 3D sphericity. Sphericity measured in 2D, however, is not affected by image resolutions to 50 μm/pixel for the aeolian sediment and 100 μm/pixel for the fluvial sediment. Roundness estimates made in 2D and 3D are compared and show broad similarities for both sediments. At different image resolutions, roundness estimates show the average, mode, and standard deviation can be determined at all resolutions. Although, resolutions higher than 20 μm/pixel for the aeolian sediment and 40 μm/pixel, for the fluvial sediment yield most accurate results. Textural analysis of images taken at different times of day (aeolian sediment only) show shadow length has a negligible effect on the textural parameters measured in this study. The results determine the smallest grain size at any image resolution that can be studied, as well as the required the number of grains to define a population for sediment possessing the studied sorting characteristics