RESOLVING BASALTIC SEDIMENT CHARACTERISTICS FROM 2D IMAGES USING KNOWN 3D DATA: IMPLICATIONS FOR DESCRIBING AND INTERPRETING THE MARTIAN SURFACE

Until there is a sample return mission to Mars, textural analysis of martian sedimentary particles will rely on two-dimensional (2D) image interpretation. Such an approach has yielded interesting results at Gusev Crater, and will also now be used in Gale Crater with the successful landing of the Curiosity rover. However, 2D techniques impose limitations because of the inherent problems of evaluating three-dimensional (3D) data from 2D imagery. To better understand such limitations in basaltic sediment, we have set up an experiment using analog material that directly compares 3D textural properties (grain size, distribution, sphericity, rounding) with those determined from 2D images of the same material at two resolutions (140 µm/pix, 37 µm/pix). Crushed basalt from the Cima volcanic field was sieved into 0.5 phi grain-size fractions and define a poorly sorted unimodal distribution with an average clast size at -1 phi (2 mm). We separated approximately 300 clasts representing grain sizes from -4 to 2 phi for individual assessment. Grain edges were examined for roundness, and as expected from a deposit formed only by intense shattering, the clasts are all angular (avg = 1.4). Axial lengths of clasts that range from -4 to 0 phi in size were measured with a digital caliper and the data used to determine sphericity. The data span a range from 0.4 to 0.9 with an average value of ~0.6. Image analysis of the sediment reveals considerable changes between low- and high-resolution. At low resolution the frequency histogram is erratic showing anomalous abundance of coarse material at -3.5 phi and a dramatic drop-off of data acquisition between -1 and 0 phi, with an average phi size of about -1.5 phi. By contrast, the data collected at high resolution has a much smoother pattern as expected from simple crushing of bedrock. The average grain size in both is about -1.5 phi or about 0.5 phi off from the sieved data. Rounding at low resolution is highly variable, with values from 1-6, and an average of ~3, whereas at high resolution the data are far more skewed to low values with an average less than 2. At both high and low resolutions 2D sphericity calculations show the same results (avg = ~ 0.75), which is more spherical than the 3D data. Continued work will determine how well 2D image analysis, and at what resolution, can best approximate actual 3D results.