Back to search

Michigan State University

Incorporating tacit knowledge of soil-landscape relationships for digital soil and landscape mapping applications

Abstract

dc:description

The purpose of this dissertation is to develop and test a quantitative model for classifying hillslope position, using digital terrain analyses. The model is calibrated by, and validated against observations of soil scientists, while working in the field. Classifying the landscape into the five standard hillslope elements, or positions, in a manner that is transferable between landscapes, can greatly help soil sampling and mapping endeavors. This dissertation begins by reviewing the role of environmental predictors in the conceptual models of soil geographers, at different scales, via a historical overview of soil maps. I conclude that the sum experience of soil science has recognized a hierarchy of scale in three of Dokuchaev's five factors of soil formation, proceeding in decreasing phenomenon scale: 1) climate, 2) parent material, and 3) relief. That relief is at the smallest phenomenon scale suggests that delineations on modern, county-level soil maps can perhaps be best disaggregated by increasing the resolution and applicability of topographic base maps used to construct those maps. Today, LiDAR technology provides high resolution, elevation data, but requires contextual information if it is to be used for mapping landscape elements and soil cover patterns. Soil geomorphic research has long recognized the relationship between soil development and the contextual metric of hillslope position. However, the definitions for hillslope positions are based on conceptual models that synthesize multiple terrain parameters, which are not always easily observed in elevation data alone. This dissertation develops and utilizes methods for calibrating analysis scales, classification breaks, and model structures to soil scientists' field observations, as a means of capturing their expert knowledge of hillslope position. The result is a quantitative definition of hillslope position, which can be applied in digital terrain analysis. The optimal analysis scales observed for slope gradient, profile curvature, and relative elevation were 9 m, 63 m, and 135 m, respectively. The best performing model differentiates areas on the landscape using a decision tree hierarchy that divides the landscape first by slope gradient (breaks at 1.4 and 2.9°), then by profile curvature and relative elevation. The final model was also tested for sensitivity to quality of the digital elevation data used to calculate the various slope parameters. Because of the importance of analysis scale, grid resolution was equally important for obtaining accurate results as uncertainty in the elevation attribute. By using the analysis scales, classification breaks, and model structure identified in this dissertation, hillslope position was determined consistently across a variety of Midwestern U.S. landscapes. The resulting hillslope position base map provides quantitative criteria for common soil map unit boundaries, study of geomorphic landscape patterns, and sampling strategies. This classification of hillslope position not only has the potential to increase soil map resolution by disaggregation of current delineations, but also offers consistency. Consistency increases the reliability of soil maps and allows for comparisons between landscapes.

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Miller, Bradley Allen
Contributors dc:contributor
  • Schaetzl, Randall J.
  • Shortridge, Ashton
  • Lusch, David
  • Kravchenko, Sasha

Subjects

dc:subject × 5

Rights

dc:rights
Statement dc:rights
  • In Copyright
Language dc:language
English

Identifiers

dc:identifier.*
Identifier
etd:46
isbn:130325669X
isbn:9781303256691
oclc:931742560
umi:3587887
local:Miller_grad.msu_0128D_12156

Chain of custody

source
Harvested from
Michigan State University
Base URL
d.lib.msu.edu/oai
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Miller, Bradley Allen. Incorporating tacit knowledge of soil-landscape relationships for digital soil and landscape mapping applications. 2013. https://doi.org/doi:10.25335/vgay-3a87