Roecker-2010-Scale_Effects

Introduction

  • Digital elevation models (DEMs) are crucial for soil mapping.

  • The scale of terrain attribute calculation impacts the representation of soil landscapes.

  • Multi-scale terrain analysis may improve data accuracy over traditional methods.

Key Concepts

Scale and Representation

  • Terrain attributes depend on the grid and neighborhood size for calculation.

  • High-resolution DEMs present more accurate soil-landscape relationships.

  • Neighborhood size affects topographical variability and represents different landscape features.

Case Studies Overview

  1. Case Study 1: Effects of varying grid and neighborhood size on terrain attributes from LiDAR.

  2. Case Study 2: Examining correlations between soil and terrain attributes with different neighborhood sizes.

Case Study 1: Systematic Effects on Terrain Attributes

Study Area

  • Locations: Gilmer County and Jefferson County, West Virginia.

  • Use of 1-m resolution LiDAR datasets over a quarter-quarter (QQ) quadrangle (~600 ha).

DEM Resampling and Terrain Calculation

  • Employed nearest neighbor approach for comparing terrain attributes.

  • Calculated attributes include slope gradient, northerness, and several curvature types.

  • Slope aspect transformed to northerness for analysis.

Results: Varying Neighborhood Size

  • Smaller neighborhood sizes (≤9 m) emphasize microtopographic features.

  • Moderate sizes (15-81 m) capture broader landscape trends (e.g., hillslopes).

  • Larger sizes (>81 m) may oversimplify and cause misrepresentation of terrain.

  • Observed differences in sensitivity between the two landscapes (Gilmer vs. Jefferson) depending on relief.

Comparison Metrics

  • Assess goodness of fit via mean difference (MD), root mean square difference (RMSD), and Pearson correlation coefficient (r).

  • Results showed significant representation changes for smaller neighborhood sizes.

Case Study 2: Correlations Response to Neighborhood Size

Study Area

  • Focused on Upper Gauley watershed, Monongahela National Forest.

  • Soil dataset collected from 97 sites using stratified random sampling.

Soil Properties Analysis

  • Relevant soil properties analyzed included pH, particle size, and nutrient concentrations.

  • Soil data stratified by geology, elevation, and stream power index.

Correlation Procedures

  • Established correlation between soil properties and terrain attributes across differing neighborhood sizes.

  • Slope curvature showed the strongest correlation, especially at optimal neighborhood sizes (117-189 m).

Conclusions

  • Terrain attributes must align with the scale of landforms for accurate soil correlation.

  • Larger neighborhood sizes effectively filter noise without losing important data.

  • For effective digital soil mapping, a neighborhood size of 81 m is optimal for representing gradients.

References

  • Literature cited includes seminal works on DEM applications, terrain modeling, and soil prediction.