Remote sensing survey of Chinese tallow tree in the Toledo Bend Reservoir area, Louisiana and Texas
We applied Hyperion sensor satellite data acquired by the National Aeronautics and Space Administration’s Earth Observing-1 (EO-1) satellite in conjunction with reconnaissance surveys to map the occurrences of the invasive Chinese tallow tree (Triadica sebifera) in the Toledo Bend Reservoir study area of northwestern Louisiana and northeastern Texas. The rationale for application of high spectral resolution EO-1 Hyperion data was based on the successful use of Hyperion data in the mapping of Chinese tallow tree in southwestern Louisiana in 2005. In contrast to the single Hyperion image used in the 2005 project, more than 20 EO-1 Hyperion and Advanced Land Imager (ALI) images of the study area were collected in 2009 and 2010 during the fall senescence when Chinese tallow tree leaves turn red. Atmospherically corrected reflectance spectra of Hyperion imagery collected at ground and aerial observation locations provided the input datasets used in the program for spectral discrimination analysis. Discrimination analysis was used to identify spectral indicator sets to best explain variance contained in the input databases. The expectation was that at least one set of Hyperion-based indicator spectra would uniquely identify occurrences of red-leaf Chinese tallow tree; however, no combination of Hyperion-based reflectance datasets produced a unique identifier.
The inability to discover a unique spectral indicator resulted primarily from relatively sparse coverage by red-leaf Chinese tallow tree within the study area (percentage of coverage was less than 5 percent per 30- by 30-meter Hyperion pixel). To enhance the performance of the spectral discrimination analysis, leaf and canopy spectra of Chinese tallow tree were added to the input datasets to guide the indicator selection. In addition, input databases were segregated by land class obtained from an ALI-based landcover classification in order to reduce the input variance and to promote spectral discrimination of red-leaf Chinese tallow tree. Although no unique spectral identifier for red-leaf Chinese tallow tree was uncovered with these enhanced methods, in some cases predicted spatial patterns throughout the Hyperion images revealed alignment with vegetation associations within each land class that was often observed to contain Chinese tallow trees. These instances were associated particularly with the addition of helicopter-based spectra to the input databases. It was attempted to extend such predictions of likely occurrences of Chinese tallow tree by mapping six of the nine Hyperion swaths and four of the nine land classes, but this attempt produced uncertain results that could not be fully evaluated for accuracy. Even though the final mapping showed promise in identifying likely Chinese tallow tree occurrences, the low percentage of occurrences hindered mapping performance and validation. Results of the mapping suggested that successful detection of Chinese tallow tree in the study area would require a spectral sensor similar to the Hyperion but with a higher ground-level spatial resolution.
Although the Hyperion-based spectral mapping did not provide the desired results, the associated field (ground and aerial) surveys did provide for a qualitative assessment of the overall Chinese tallow tree distribution within the study area. Ground and aerial surveys suggested that Chinese tallow tree occurrences were uncommon and were without an observed pattern in relation to proximity to the Toledo Bend Reservoir. Although uncommon and scattered, Chinese tallow trees and shrubs most commonly existed along forest edges, water edges, and fence lines, probably most in line with seed dispersal by birds. Chinese tallow trees were observed to be more densely dispersed within some scrublands and grasslands than were observed in pine, hardwood, and mixed forests.
Citation Information
Publication Year | 2013 |
---|---|
Title | Remote sensing survey of Chinese tallow tree in the Toledo Bend Reservoir area, Louisiana and Texas |
DOI | 10.3133/ofr20121215 |
Authors | Elijah W. Ramsey, Amina Rangoonwala, Terri Bannister, Yukihiro Suzuoki |
Publication Type | Report |
Publication Subtype | USGS Numbered Series |
Series Title | Open-File Report |
Series Number | 2012-1215 |
Index ID | ofr20121215 |
Record Source | USGS Publications Warehouse |
USGS Organization | National Wetlands Research Center; Northern Prairie Wildlife Research Center |