142-8 Adapting Image-Based Grain-Size Analysis Techniques to Characterize Poorly Sorted Regolith on Mars

Session: Geomorphology and Landscape Evolution of Mars (Posters)

Presenting Author:

Authors:

Abstract:

Martian regolith serves as a scientific archive while simultaneously dictating rover mobility and in-situ resource potential. A key characteristic of regolith is grain size distribution (GSD), typically gleaned from high-resolution imagery captured by proximity rover cameras like Curiosity’s Mars Hand Lens Imager (MAHLI). When targets are well sorted, GSDs can be characterized manually at one scale by either measuring every grain in an image or overlaying a grid onto the image and measuring the grain at each grid intersection as a representative subpopulation.

On poorly sorted regolith surfaces, grain sizes can span over three orders of magnitude, complicating GSD measurements. Image standoff distances that capture centimeter-scale rocks yield images with insufficient resolution to regularly measure micron-scale grains. Additionally, grid counting is hindered by spatial constraints: resolving relative grain proportions accurately requires 200-400 points spaced wider than the largest visible grain in the image. For heterogeneous regolith, this requires a grid layout larger than available image dimensions.

MAHLI's imaging of surfaces at multiple standoff distances enabled us to develop a method for characterizing heterogeneous regolith using two scales. We calculate the areal coverage of each grain size (rather than count-based or volumetric GSDs). First, using a 25 cm-standoff MAHLI image, we manually measure all resolvable grains down to a lower limit of five pixels (~500μm), drawing polygons around each grain. Next, using a representative 2 cm-standoff image within the 25 cm-standoff image, we use a grid count method to measure the fine grains, drawing polygons around individual grains at grid intersections. Given the higher resolution of that image, our 5-pixel limit allows us to measure grains down to ~100 μm. When completing the grid count, we exclude all polygonal areas already outlined in the 25 cm-standoff image. This generates two GSDs that may have overlapping size ranges (e.g., depending on dust, shadows, grain agglomerates, and/or cementation). We transition from count-based GSDs to areal GSDs by normalizing individual grain areas to the total grain area measured on the same scale image. We generate a single surface-wide GSD by merging the two distributions—weighting each population by its areal coverage calculated from the 25 cm-standoff image—then binning and plotting by long-axis length.

This technique enables 2D image-based grain size analysis of heterogeneous regolith targets where standard approaches fail.

© Copyright 2026 The Geological Society of America (GSA), all rights reserved.