Skip to main content

DAAC Home > Get Data > Regional/Global > Vegetation Collections > Data Set Documentation

LiDAR-derived Vegetation Canopy Structure, Great Smoky Mountains National Park, 2011

Documentation Revision Date: 2015-10-16

Data Set Version: V1

Summary

This data set provides multiple-return LiDAR-derived vegetation canopy structure and their spatial distribution at 30-meter spatial resolution for the Tennessee part of the Great Smoky Mountains National Park (GSMNP). Canopy characteristics were analyzed using high resolution three-dimensional point cloud measurements gathered between February and April 2011. Vegetation types were mapped by grouping areas of similar canopy structure. The map was compared and validated against existing vegetation maps for the Park.

This LiDAR based vegetation canopy data set incorporates three-dimensional canopy structure, yielding information not available in two-dimensional images of the landscape provided by most of the traditional multi-spectral remote sensing platforms. These three-dimensional measurements provide a better representation of vegetation structure and distribution within the diverse ecosystem of the GSMNP.

This data set includes three GeoTIFFs and three supplementary CSV files.

Support Acknowledgement

This research was partially sponsored by the U.S. Department of Agriculture, U.S. Forest Service, Eastern Forest Environmental Threat Assessment Center. Additional support was provided by the Biogeochemistry Feedbacks Scientific Focus Area, which is sponsored by the Regional and Global Climate Modeling Program in the Climate and Environmental Sciences Division of the Biological and Environmental Research Program in the U.S. Department of Energy Office of Science.

Figure 1: 30 unique vegetation canopy structure classes identified for the Tennessee side of Great Smoky Mountain National Park.

Citation

Kumar, J., J. Weiner, W.W. Hargrove, S.P. Norman, F.M. Hoffman, and D. Newcomb. 2015. LiDAR-derived Vegetation Canopy Structure, Great Smoky Mountains National Park, 2011. ORNL DAAC, Oak Ridge, Tennessee, USA. http://dx.doi.org/10.3334/ORNLDAAC/1286.

Table of Contents

  1. Data Set Overview
  2. Data Characteristics
  3. Application and Derivation
  4. Quality Assessment
  5. Acquisition Materials and Methods
  6. Data Access
  7. References

Data Set Overview

Investigators

Kumar, J., J. Weiner, W.W. Hargrove, S.P. Norman, F.M. Hoffman, and D. Newcomb.

This data set provides multiple-return LiDAR-derived vegetation canopy structure and their spatial distribution at 30-meter spatial resolution for the Tennessee part of the Great Smoky Mountains National Park (GSMNP). Canopy characteristics were analyzed using high resolution three-dimensional point cloud measurements gathered between February and April 2011. Vegetation types were mapped by grouping areas of similar canopy structure. The map was compared and validated against existing vegetation maps for the Park.

This LiDAR based vegetation canopy data set incorporates three-dimensional canopy structure, yielding information not available in two-dimensional images of the landscape provided by most of the traditional multi-spectral remote sensing platforms. These three-dimensional measurements provide a better representation of vegetation structure and distribution within the diverse ecosystem of the GSMNP.

LiDAR point clouds were aggregated into a 30- by 30-meter horizontal grid and split into 1-meter vertical bins from the ground surface to a maximum height of 75 meters. Point density within each bin was used to characterize vertical structure of the vegetation canopy as a percentage of total points in each 30-meter grid cell. Cluster analysis identified 30 distinct canopy structure classes representing 8 traditional vegetation classes for GSMNP.  Data processing is described in Kumar et al. (2015).  This data set includes geotiffs (3) depicting LiDAR-derived canopy structure classes, vegetation classes mapped to LiDAR-derived canopy structure classes, and LiDAR-derived canopy structure classes mapped to vegetation classes.

Data Characteristics

Spatial Coverage

This data set covers approximately 1,400 square kilometer (540 square mile) area of the Tennessee part of the GSMNP.

Spatial Resolution

30 x 30 meters

Temporal Coverage

This data set was produced from LiDAR point cloud data gathered between February and April 2011.

Temporal Resolution

The data set is representative of vegetation canopy structure within GSMNP during the airborne LiDAR campaign.

Study Area: (All latitudes and longitudes given in decimal degrees)

Site Northern Extent Western Extent Eastern Extent Southern Extent Geodetic Datum
GSMNP (TN) 35.831601 -84.035931 -83.031695 35.433051 North American Datum, 1983 (NAD83)

 

lidar DEM

Figure 2: 1.5-meter resolution digital elevation model for the study area. 1,500-meter by 1,500-meter LiDAR tiles represented by grid overlay.

The Tennessee side of GSMNP covers an approximately 1,400 square kilometer (540 square mile) area of complex topography, with elevations ranging from 267 to 2,025 meters above mean sea level. GSMNP is ecologically rich and diverse, consisting of about 1,600 species of flowering plants, including 100 native tree species and over 100 native shrub species. The distribution of vegetation in the park is strongly influenced by topography, moisture, and other environmental gradients.

 

Data File Information

 

This data set is comprised of 3 single-band GeoTIFFs accompanied by 3 companion text files.

GeoTIFF Spatial Data Properties

Spatial Representation Type: Raster
Pixel Depth: 8 bit
Pixel Type: unsigned integer
Compression Type: LZW
Number of Bands: 1
Raster Format: TIFF
Source Type: Thematic
No Data Value: 255

Data Format and Projection

All GeoTIFFs are distributed with the following projection:

Projection: UTM Zone 17N
Datum: North American Datum 1983
Spheroid name: GRS 1980
False Easting: 500000
False Northing: 0
Central Meridian: -81
Latitude of origin: 0

File Northern Extent Western Extent Eastern Extent Southern Extent
gsmnp_canopy_structure_classes_k30.tif 35.831601 -84.035931 -83.095437 35.487784
GRSM_VegInventory_30m.reclassed.tif 35.775217 -84.008943 -83.031695 35.433051
gsmnp_canopy_structure_classes_k30.reclassed.tif 35.831601 -84.035931 -83.095437 35.487784

Data Files

gsmnp_canopy_structure_classes_k30.tif 

This GeoTIFF contains LiDAR-derived vegetation canopy structure based on 30 unique vegetation classes. The vertical canopy structures that defines each of these classes are found in: vegetation_canopy_structures_k30.csv

Note: Each row within vegetation_canopy_structure_k30.csv represents a LiDAR-derived canopy structure and each column a 1-meter vertical bin of average point densities per square meter for each canopy structure.

GRSM_VegInventory_30m.reclassed.tif

This GeoTIFF contains GRSM vegetation inventory (Madden 2014) remapped to the LiDAR-based canopy structure classes. The translation table is found in: translation_table_GRSMvegmap-k30.csv

gsmnp_canopy_structure_classes_k30.reclassed.tif

This GeoTIFF contains the LiDAR-based canopy structure classes remapped to GRSM vegetation inventory (Madden 2014). The translation table is found in: translation_table_k30-GRSMvegmap.csv

Note: See Madden (2014) for additional information regarding CEGL codes and classes.

Table 1: Mapcurves (Hargrove et al. 2006) based translation between LiDAR-derived canopy structure and traditional GRSM vegetation inventory classes

Canopy structure GSMNP Vegetation Descriptor
0 Successional or modified vegetation
1 Chestnut oak forests
2 Chestnut oak forests
3 Successional or modified vegetation
4 Chestnut oak forests
5 Northern hardwood/acid hardwood forests
6 Chestnut oak forests
7 Yellow pine forests
8 Northern hardwood/acid hardwood forests
9 Chestnut oak forests
10 Montane cove forests
11 Chestnut oak forests
12 Northern hardwood/acid hardwood forests
13 Montane oak-hickory forests
14 Northern hardwood/acid hardwood forests
15 Yellow pine forests
16 Chestnut oak forests
17 Montane cove forests
18 Montane oak-hickory forests
19 Chestnut oak forests
20 Montane oak-hickory forests
21 Spruce-fir forests
22 Northern hardwood/acid hardwood forests
23 Chestnut oak forests
24 Yellow pine forests
25 Montane oak-hickory forests
26 Chestnut oak forests
27 Ericaceous shrubs (Heath Bald type)
28 Chestnut oak forests
29 Yellow pine forests
30 Chestnut oak forests

Note: Mapcurves identifies the single vegetation type having the best fit in terms of spatial overlap with the LiDAR canopy structure clusters, though clusters are likely to overlap with many vegetation types. The table above shows the single vegetation category exhibiting the largest spatial co-registration.

For instance, the Chestnut oak vegetation type, which accounts for 43% of the mapped area, corresponds with 12 different profile types, perhaps reflecting differences in both forest age and compositional differences.

Application and Derivation

This data set was produced to identify and characterize the three-dimensional structure and spatial distribution of vegetation canopies. Vegetation canopy structure is a critically important habitat characteristic for many threatened and endangered birds and other animal species. Species composition in the GSMNP is in a state of transition due to various environmental stressors like fires, hemlock death due to the wooly adelgid, and other factors. This data set provides forest and wildlife managers with critically important information for resource management and conservation planning. 

Quality Assessment

Case studies were conducted to verify the LiDAR based canopy structures against the best available maps of vegetation in GSMNP (Madden, 2014). Cades Cove consists of woodlots interspersed within regularly mowed and burned fields to mimic a 19th century agrarian settlement. Vegetation maps produced by Madden (2014) identify this region as "Successional or modified vegetation," in correspondence with the low height vegetation class identified through the LiDAR-based analysis. Additional validation was performed at a number of "Citizen Science" phenology plots maintained by the Great Smoky Mountains Institute at Tremont (GSMIT), an area surrounded by "Montane Cove" and "Hemlock" forest, which further indicated agreement between the LiDAR canopy structures and vegetation class.

Acquisition Materials and Methods

The airborne LiDAR survey was conducted during the period of February to April 2011 by The Center for Remote Sensing and Mapping Science at the University of Georgia and Photo Science, Inc. under a U.S. Geological Survey-funded program. The survey covered the roughly 1,400 sq. km (540 sq. miles) area of the Tennessee portion of the GSMNP and Foothills Parkway. Data were calibrated and LiDAR points categorized as unclassified, ground, noise, or overlap. Data sets were split into 1,500 by 1,500-meter adjacent and non-overlapping tiles. The tiled data sets consisting of 724 tiles in "las" format, were obtained from the GSMNP Service.

workflow

Figure 3: Computational workflow for the analysis.

The raw LiDAR point cloud data was filtered to remove points at anomalously high and low elevations and points corresponding to low height vegetation within 1 meter of ground level. Grid cells where 95% or more LiDAR return points were within 1 meter of the ground surface were identified and classified as low lying elevation.

A k-means clustering algorithm was used to cluster the gridded point cloud into groups containing locations with similar vertical canopy structure. k-means clustering groups data by iteratively assigning pixels to the nearest centroid and repositioning clusters based on the mean euclidean distance of each assigned pixel to all centroids until fewer than 0.05% changed cluster assignments. The high volume of LiDAR data used to produce this data set was analyzed using a parallel version of the k-means algorithm (Kumar et al., 2011) to accelerate convergence, handle empty cluster cases, and obtain initial centroids through a scalable implementation of the triangular equality based acceleration method (Hartigan 1975).

kmeans

Figure 4: Representative vegetation canopy structures that define the 30 unique canopy structure classes.

Mapcurves (Hargrove et al., 2006) algorithm was used to identify the best translation table between LiDAR clusters and vegetation types. Mapcurves identifies the single vegetation type having the best fit in terms of spatial overlap with each LiDAR cluster.

Data Access

This data is available through the Oak Ridge National Laboratory (ORNL) Distributed Active Archive Center (DAAC).

LiDAR-derived Vegetation Canopy Structure, Great Smoky Mountains National Park, 2011

Contact for Data Center Access Information:

References

Hargrove W. W., F. M. Hoffman, and P. F. Hessburg, “Mapcurves: A quantitative method for comparing categorical maps,” J. Geograph. Syst., vol. 8, no. 2, pp. 187–208, Jul. 2006.

Hartigan J.A., Clustering Algorithms. John Wiley & Sons, 1975.

Kumar J., R. T. Mills, F. M. Hoffman, and W. W. Hargrove, “Parallel k-means clustering for quantitative ecoregion delineation using large data sets,” in Proceedings of the International Conference on Computational Science (ICCS 2011), ser. Procedia Comput. Sci., M. Sato, S. Matsuoka, P. M. Sloot, G. D. van Albada, and J. Dongarra, Eds., vol. 4. Amsterdam: Elsevier, Jun. 2011, pp. 1602–1611.

Kumar, J., J. Weiner, W.W. Hargrove, S.P. Norman, F.M. Hoffman, and D. Newcomb,  2015.  Characterization and classification of vegetation canopy structure and distribution within the Great Smoky Mountains National Park using LiDAR.  Proc. Intern. Conf. Data Mining (ICDM 2015)  

Madden M., “Overstory Vegetation at Great Smoky Mountains National Park, Tennessee and North Carolina, Reference Code: 1047498,” 2014. [Online]. Available: https://irma.nps.gov/App/ Reference/Profile/1047498