Mapping Soils Using Predictive Models in a Great Basin Landscape Influenced by Volcanic Activity.
A. Stum and Janis L. Boettinger. Utah State Univ, 4820 Old Main Hill, Logan, UT 84322-4820
We applied the pedogenic understanding raster classification (PURC) methodology to a small watershed (~110,000 ac) in Beaver County, Utah, to map soils as part of an order 3 soil survey. Located on the fringe of the Tonoquints Volcanic Field and near the eastern extent of the Great Basin, the area has a complicated landscape composed of Proterozoic and Paleozoic sedimentary rocks and Tertiary volcanic rocks. To facilitate and simplify modeling, the area was stratified into three distinct physiographic regions: LakeBonneville lacustrine deposits, piedmont and alluvial slopes, and mountains and ridges. We approached the modeling of soils in each region independently to address the unique soil forming factors. Data derived from 10-m digital elevation models (DEMs) and Landsat 7 ETM data were used to represent soil forming factors within each stratified region. To separate Lake Bonneville deposits from the other physiographic regions, we created a raster layer using Interactive Data Language (IDL). This highest shoreline feature is not the same elevation in the study area because of local variation in isostatic rebound. Using IDL, we were able to create a raster layer that delimits the shoreline, which ranges from 1558.2 m in the south to 1561.9 m in the north. Above the LakeBonneville shoreline, mineralogy plays a very important role in soil formation. Normalized difference indices of Landsat spectral bands were used to differentiate soil parent material mineralogy: (2-3)/(2+3), (3-4)/(3+4), (4-5)/(4+5), (5-7)/(5+7), and (5-2)/(5+2). Sedimentary rock units and volcanic rock units are often easily separated using these band combinations. Within some map units, components of associations and complexes were separated using digital color orthophotography.