Comparisons of five indices for estimating local terrain surface roughness using LiDAR point clouds

Lei Fan

DOI: https://doi.org/10.1109/Geoinformatics57846.2022.9963877

2023-01-06

Abstract:Terrain surface roughness is an abstract concept, and its quantitative description is often vague. As such, there are various roughness indices used in the literature, the selection of which is often challenging in applications. This study compared the terrain surface roughness maps quantified by five commonly used roughness indices, and explored their correlations for four terrain surfaces of distinct surface complexities. These surfaces were represented by digital elevation models (DEMs) constructed using airborne LiDAR (Light Detection and Ranging) data. The results of this study reveal the similarity in the global patterns of the local surface roughness maps derived, and the distinctions in their local patterns. The latter suggests the importance of considering multiple indices in the studies where local roughness values are the critical inputs to subsequent analyses.

Numerical Analysis

What problem does this paper attempt to address?

This paper aims to address the ambiguity in the quantitative description of terrain surface roughness and the challenges in method selection. Specifically, by comparing five commonly - used terrain surface roughness indices, the paper explores their correlations on four terrain surfaces with different levels of complexity. These terrain surfaces are represented by digital elevation models (DEMs) constructed using airborne LiDAR data. The research results reveal the similarities in the global patterns and the differences in the local patterns of the local surface roughness maps, emphasizing the importance of considering multiple roughness indices in subsequent analyses, especially when local roughness values are key inputs.

The five roughness indices used in the paper are:
1. **Root - Mean - Square Height (RMSH)**: It is used to quantify the local surface roughness of scattered elevation data and is defined as follows:
\[
\text{RMSH} = \sqrt{\frac{1}{n} \sum_{i = 1}^{n}(Z_i-\bar{Z})^2}
\]
where \(n\) is the number of selected data points, \(Z_i\) is the elevation value of the \(i\) - th data point, and \(\bar{Z}\) is the average elevation value of all selected data points.

2. **Standard Deviation of Local Detrended Residual Elevation (\(\sigma_{LDRE}\))**: The surface roughness is represented by calculating the standard deviation of the residual elevation through linear detrending of the local elevation data in a moving window.

3. **Standard Deviation of Residual Terrain (\(\sigma_{RT}\))**: The residual terrain is the difference between the original terrain (original DEM) and the smoothed terrain (smoothed DEM). The surface roughness is represented by calculating the standard deviation of the residual terrain.

4. **Standard Deviation of Slope (\(\sigma_{slope}\))**: First, calculate the slope value. The slope is defined as the rate of change of the terrain elevation, and the expression is as follows:
\[
\text{slope}=\sqrt{\left(\frac{dz}{dx}\right)^2+\left(\frac{dz}{dy}\right)^2}
\]
where \(\frac{dz}{dx}\) and \(\frac{dz}{dy}\) represent the rates of change in the \(x\) and \(y\) directions respectively. Based on the DEM, \(\frac{dz}{dx}\) and \(\frac{dz}{dy}\) of the central cell are usually calculated using a 3×3 moving window, and the specific formulas are as follows:
\[
\frac{dz}{dx}=\frac{(Z_3 + 2Z_6+Z_9)-(Z_1 + 2Z_4+Z_7)}{8L}
\]
\[
\frac{dz}{dy}=\frac{(Z_7 + 2Z_8+Z_9)-(Z_1 + 2Z_2+Z_3)}{8L}
\]
where \(L\) represents the cell size.

5. **Standard Deviation of Curvature (\(\sigma_{curvature}\))**: The curvature is obtained by calculating the second - order derivative of the DEM raster map, using the same moving window as for calculating the slope. The curvature calculation formula is as follows:
\[
\text{curvature}=2E + 2D
\]
where \(E\) and \(D\) are given by the following formulas respectively:
\[
D=\frac{(Z_4 + Z_6)/2-Z_5}{L^2}
\]
\[
E=\frac{(Z_2 + Z_8)/2-Z_5}{L^2}
\]

Through these methods, the paper explores the performance and correlations of different roughness indices at different scales, providing an important reference for subsequent research.