REVISITING MULTIPARAMETER RELATIVE-AGE METHODS FOR MAPPING LATE QUATERNARY FAN DEPOSITS OF THE SODA MOUNTAINS, MOJAVE DESERT, CALIFORNIA, WITH STRUCTURE-FROM-MOTION PHOTOGRAMMETRY

Grain size and surface relief of alluvial fan surfaces derived from digital elevation models (DEMs) with clast-size resolution are used for relative-age determination and correlation of alluvial fan surfaces. Original studies relied on field measurements of individual clasts and bar and swale topography; more recent studies have focused on surface roughness derived from airborne LiDAR at scales reflecting bar and swale topography and greater. In this study, we bridge the difference in scale using low altitude aerial photography as a prototype for mapping by unmanned aircraft systems (UAS). High-resolution DEMs and orthophotos of alluvial fan surfaces ranging in age from late Pleistocene to late Holocene were produced using SfM photogrammetric analysis of photos taken from pole-mounted cameras at 5-6 m in elevation. Starting at sub-centimeter resolution, DEMs were aggregated to successively coarser resolutions ranging from 0.01-0.83 m to reflect the transition from field- to flight-scale measurements. Surfaces representing slope, in degrees, and the terrain ruggedness index were derived from the DEMs. Five proxies for surface roughness were analyzed at each aggregated size using a 3x3 moving window: standard deviation and range of elevation, standard deviation and range of slope, and mean ruggedness. Each of the roughness proxies can discriminate between late Pleistocene and younger alluvial fan surfaces and the differences are statistically significant at 0.01 level for the 0.28-m aggregated data. Range of elevation and mean terrain ruggedness can discriminate between late Pleistocene, early to mid Holocene, and mid to late Holocene surfaces. Because of the tradeoffs between flight altitude, mapping extent, and ground resolution, these preliminary results provide a target DEM resolution for future mapping using UAS. DEM resolutions between 0.15-0.30 m are fine enough to incorporate both larger clasts sizes and bar and swale topography into roughness measurements. To incorporate varnish cover into our analysis, perceived brightness of the orthophotos was also analyzed by several different metrics, but the variation of time of day of photography on the different fan surfaces created problems with lighting and shadowing.