Modeling Soil Depth based upon Topographic and Landscape Attributes
Location
ECC 303/305
Event Website
http://water.usu.edu/
Start Date
4-5-2007 3:50 PM
End Date
4-5-2007 4:10 PM
Description
Soil depth is an important input parameter in hydrologic, ecological, irrigation, and solute transport modeling, land use planning, land resource management etc. The spatial patterns of soil depth result from the combination of geomorphologic and soil forming processes, many of which are related to topography. Presently, the soil depth data available in national databases (STATSGO, SSURGO) from the Natural Resources Conservation Service (NRCS) is quite generalized and uncertain, limiting its applicability for distributed hydrologic modeling. The objective of our work is to develop a soil depth spatial distribution prediction model for semiarid mountainous watersheds based upon the relationship between soil depth and topography and other landscape variables. Soil depth was surveyed by driving a rod into the ground until refusal at preselected georeferenced locations in Dry Creek Watershed, Boise Idaho. The survey was designed to develop prediction algorithms from 810 survey locations in within 8 sub-watersheds with different topographic settings and test the algorithms at 130 randomly distributed locations over the remaining part of the watershed. Topographic attributes were derived from a Digital Elevation Model. These included slope, aspect, curvature, distance to stream, slope length, distance to stream, wetness index, and hill slope position. Soil texture and erodibility variables were extracted from the SSURGO soil database. Land cover and rock outcrop attributes were derived from remote sensing images and high resolution aerial photographs. We present our initial comparative results using Stepwise Regression, Generalized Additive, and Random Forest models.
Modeling Soil Depth based upon Topographic and Landscape Attributes
ECC 303/305
Soil depth is an important input parameter in hydrologic, ecological, irrigation, and solute transport modeling, land use planning, land resource management etc. The spatial patterns of soil depth result from the combination of geomorphologic and soil forming processes, many of which are related to topography. Presently, the soil depth data available in national databases (STATSGO, SSURGO) from the Natural Resources Conservation Service (NRCS) is quite generalized and uncertain, limiting its applicability for distributed hydrologic modeling. The objective of our work is to develop a soil depth spatial distribution prediction model for semiarid mountainous watersheds based upon the relationship between soil depth and topography and other landscape variables. Soil depth was surveyed by driving a rod into the ground until refusal at preselected georeferenced locations in Dry Creek Watershed, Boise Idaho. The survey was designed to develop prediction algorithms from 810 survey locations in within 8 sub-watersheds with different topographic settings and test the algorithms at 130 randomly distributed locations over the remaining part of the watershed. Topographic attributes were derived from a Digital Elevation Model. These included slope, aspect, curvature, distance to stream, slope length, distance to stream, wetness index, and hill slope position. Soil texture and erodibility variables were extracted from the SSURGO soil database. Land cover and rock outcrop attributes were derived from remote sensing images and high resolution aerial photographs. We present our initial comparative results using Stepwise Regression, Generalized Additive, and Random Forest models.
https://digitalcommons.usu.edu/runoff/2007/AllAbstracts/31