HR: 0800h
AN: H11A-1250    [Abstracts]
TI: Remote Characterization of Gravel Bars in Big Creek, Idaho
AU: * Streutker, D
EM: stredavi@isu.edu
AF: Idaho State University Boise Center Aerospace Lab, 12301 W. Explorer Dr. #102, Boise, ID 83713 United States
AU: Mundt, J
EM: mundjaco@isu.edu
AF: Idaho State University Boise Center Aerospace Lab, 12301 W. Explorer Dr. #102, Boise, ID 83713 United States
AU: Glenn, N
EM: glennanc@isu.edu
AF: Idaho State University Boise Center Aerospace Lab, 12301 W. Explorer Dr. #102, Boise, ID 83713 United States
AB: This study utilizes remote sensing and field validation techniques to characterize sediment size on exposed bars in Big Creek, Idaho. Hyperspectral imagery, consisting of 126 contiguous bands between 350 and 2500 nm and with a spatial resolution of 3 m, was collected over the wilderness study area of Big Creek, Idaho, in July of 2004. After the collection was completed, field crews digitally recorded the particle size distributions of 36 separate bars directly onto georegistered hyperspectral maps, thereby generating a coregistered dataset. Using a geographic information system (GIS), these distributions were intersected (on a pixel by pixel basis) with the reflectance intensity of the 2.10 μm band, and bivariate plots contrasting reflectance intensity with field-estimated particle size were generated. A weak but significant inverse relationship (r² of 0.70) was determined between particle size and reflectance intensity. Based on this relationship, it is hypothesized that remotely sensed imagery may be capable of discriminating particle size distributions on exposed in-stream sediment bars. To better characterize particle size distribution, scaled digital photographs were also collected by field crews at 363 separate locations across 27 sediment bars. The photographs were taken with a nadir-perspective from a height of approximately one meter with a field of view of roughly 1 m². Using photo-sieve and geostatistical methods, particle size distributions are constrained for each of the field photographs. Photo-sieve methods include virtual point frame and random sampling, while geostatistical methods include semivariogram analysis and modeling. Upon calibration of the field photographs for absolute sediment size distribution, a separability analysis of hyperspectral reflectance profiles from regions of variant sediment size classes provides the basis to develop spectral mapping methods capable of discriminating global grain size classes. These mapping techniques are then used to develop a general grain size map for an extended stretch (~25 km) of Big Creek. Field photographs withheld from classification training are then utilized for accuracy assessment.
DE: 1819 Geographic Information Systems (GIS)
DE: 1855 Remote sensing (1640)
DE: 1856 River channels (0483, 0744)
DE: 1862 Sediment transport (4558)
SC: Hydrology [H]
MN: Fall Meeting 2005