Rupesh Shrestha

A Comparison of Two Open Source LiDAR Surface Classification Algorithms

With the progression of LiDAR (Light Detection and Ranging) towards a mainstream resource management tool, it has become necessary to understand how best to process and analyze the data. While most ground surface identification algorithms remain proprietary and have high purchase costs; a few are openly available, free to use, and are supported by published results. Two of the latter are the multiscale curvature classification and the Boise Center Aerospace Laboratory LiDAR (BCAL) algorithms. This study investigated the accuracy of these two algorithms (and a combination of the two) to create a digital terrain model from a raw LiDAR point cloud in a semi-arid landscape. Accuracy of each algorithm was assessed via comparison with >7,000 high precision survey points stratified across six different cover types. The overall performance of both algorithms differed by only 2%; however, within specific cover types significant differences were observed in accuracy. The results highlight the accuracy of both algorithms across a variety of vegetation types, and ultimately suggest specific scenarios where one approach may outperform the other. Each algorithm produced similar results except in the ceanothus and conifer cover types where BCAL produced lower errors.

What is the effect of LiDAR-derived DEM resolution on large-scale watershed model results? ☆

This paper examines the effect of raster cell size on hydrographic feature extraction and hydrological modeling using LiDAR derived DEMs. LiDAR datasets for three experimental watersheds were converted to DEMs at various cell sizes. Watershed boundaries and stream networks were delineated from each DEM and were compared to reference data. Hydrological simulations were conducted and the outputs were compared. Smaller cell size DEMs consistently resulted in less difference between DEM-delineated features and reference data. However, minor differences been found between streamflow simulations resulted for a lumped watershed model run at daily simulations aggregated at an annual average. These findings indicate that while higher resolution DEM grids may result in more accurate representation of terrain characteristics, such variations do not necessarily improve watershed scale simulation modeling. Hence the additional expense of generating high resolution DEMs for the purpose of watershed modeling at daily or longer time steps may not be warranted.

Airborne LiDAR and Terrestrial Laser Scanning Derived Vegetation Obstruction Factors for Visibility Models

Research presented here explores the feasibility of leveraging vegetation data derived from airborne light detection and ranging (LiDAR) and terrestrial laser scanning (TLS) for visibility modeling. Using LiDAR and TLS datasets of a lodgepole pine (Pinus contorta) dominant ecosystem, tree canopy and trunk obstructions were isolated relevant to a discrete visibility beam in a short-range line-of-sight model. Cumulative obstruction factors from vegetation were compared with reference visibility values from digital photographs along sightline paths. LiDAR-derived tree factors were augmented with single-scan TLS data for obstruction prediction. Good correlation between datasets was found up to 10 m from the terrestrial scanner, but fine scale visibility modeling was problematic at longer distances. Analysis of correlation and regression results reveal the influence of obstruction shadowing inherent to discrete LiDAR and TLS, potentially limiting the feasibility of modeling visibility over large areas with similar technology. However, the results support the potential for TLS-derived subcanopy metrics for augmenting large amounts of aerial LiDAR data to significantly improve models of forest structure. Subtle LiDAR processing improvements, including more accurate tree delineation through higher point density aerial data, combined with better vegetation quantification processes for TLS data, will advance the feasibility and accuracy of data integration.

A Slope-Based Method for Matching Elevation Surfaces

A method is presented for the co-registration of overlapping elevation surfaces based on local slope analysis. Comparison and statistical analysis of local slope versus local elevation difference between overlapping surfaces allows for the estimation of both vertical and horizontal offsets between the two surfaces. This method is then used to re-align the flightlines of a Light Detection and Ranging (lidar) dataset collected in southern Idaho resulting in a dataset with significantly higher accuracy than the original. The relative horizontal accuracy is doubled, with a final value of approximately 25 to 30 cm, while the relative vertical accuracy is improved by several centimeters to a final value of approximately 6 to 7 cm.

Single and Multi-Date Landsat Classifications of Basalt to Support Soil Survey Efforts

Basalt outcrops are significant features in the Western United States and consistently present challenges to Natural Resources Conservation Service (NRCS) soil mapping efforts. Current soil survey methods to estimate basalt outcrops involve field transects and are impractical for mapping regionally extensive areas. The purpose of this research was to investigate remote sensing methods to effectively determine the presence of basalt rock outcrops. Five Landsat 5 TM scenes (path 39, row 29) over the year 2007 growing season were processed and analyzed to detect and quantify basalt outcrops across the Clark Area Soil Survey, ID, USA (4,570 km 2 ). The Robust Classification Method (RCM) using the Spectral Angle Mapper (SAM) method and Random Forest (RF) classifications was applied to individual scenes and to a multitemporal stack of the five images. The highest performing RCM basalt classification was obtained using the 18 July scene, which yielded an overall accuracy of 60.45%. The RF classifications applied to the same datasets yielded slightly better overall classification rates when using the multitemporal stack (72.35%) than when using the 18 July scene (71.13%) and the same rate of successfully predicting basalt (61.76%) using out-of-bag sampling. For optimal RCM and RF classifications, uncertainty tended to be lowest in irrigated areas; however, the RCM uncertainty map included more extensive areas of low uncertainty that also encompassed forested hillslopes and riparian areas. RCM uncertainty was sensitive to the influence of bright soil reflectance, while RF uncertainty was sensitive to the influence of shadows. Quantification of basalt requires continued investigation to reduce the influence

Improved visibility calculations with tree trunk obstruction modeling from aerial LiDAR

Viewshed and line-of-sight are spatial analysis functions used in applications ranging from urban design to archaeology to hydrology. Vegetation data, a difficult variable to effectively emulate in computer models, is typically omitted from visibility calculations or unrealistically simulated. In visibility analyzes performed on a small scale, where calculation distances are a few hundred meters or less, ineffective incorporation of vegetation can lead to significant modeling error. Using an aerial LiDAR (light detection and ranging) data set of a lodgepole pine (Pinus contorta) dominant ecosystem in Idaho, USA, tree obstruction metrics were derived and integrated into a short-range visibility model. A total of 15 visibility plots were set at a micro-scale level, with visibility modeled to a maximum of 50 m from an observation point. Digital photographs of a 1 m2 target set at 5 m increments along three sightline paths for each visibility plot were used to establish control visibility values. Trunk obstructions, derived from mean vegetation height LiDAR data and processed through a series of tree structure algorithms, were factored into visibility calculations and compared to reference data. Results indicate the model calculated using trunk obstructions with LiDAR demonstrated a mean error of 8.8% underestimation of target visibility, while alternative methods using mean vegetation height and bare-earth models have an underestimation of 65.7% and overestimation of 31.1%, respectively.