EVALUATION OF HYDICE SENSOR FOR MAPPING FLOODPLAIN
VEGETATION IN THE UPPER MISSISSIPPI RIVER SYSTEM

P. H. Gowda1, T. Owens1, J. G. Lyon2, and Y. Yin1
1Environmental Management Technical Center, BRD, USGS,
Onalaska, WI 54650, 2Department of Civil Engineering,
Ohio State University, Columbus, OH 43201.

Efficient inventory of natural systems such as forest,
wetland, and floodplain ecosystems is required for
monitoring resources. Remote sensing technologies
allow monitoring at various levels of detail. Several
multispectral sensors such as Landsat Multispectral
Scanner (MSS) and Thematic Mapper (TM) are being used
for this purpose, however, they lack narrow bandwidths
and higher spatial resolution required for more
precise mapping. 

The Hyperspectral Digital Imagery Collection
Experiment (HYDICE) sensor is a second generation,
"state-of-the-art," nadir-viewing, push broom, high
resolution airborne imaging spectroradiometer. This
sensor system was developed by the Hughes Danbury
Optical Systems in coordination with the Naval
Research Laboratory and funded by the U.S. Government.
The sensor was intended for various purposes such as
evaluations of vegetation, water quality, bathymetry,
and minerals. The spatial resolution varies from one
to 4 meters depending on the aircraft's altitude above
ground level, and the spectral resolution includes 210
contiguous bandwidths from the visible to shortwave
infrared (400-2500 nm). 

An evaluation of the HYDICE sensor was made for
identifying floodplain vegetation in the Upper
Mississippi River System. For this purpose, a 16-bit
hyperspectral image at 4-meter spatial resolution was
acquired for Potter's marsh area of Pool 13 in the
Upper Mississippi River System. Required ground truth
data was collected for understanding the spectral
behavior of floodplain features and for preparing the
land cover map. A preliminary spectral analysis was
made for those features by comparing their brightness
values for the 210 spectral bands. A set of suitable
bands was selected based on individual bands' ability
to discriminate floodplain features. Land cover maps
were developed for the Potter's marsh area using
unsupervised and supervised classification techniques.
Evaluation of the land cover maps indicated that high
quality floodplain vegetation maps can be developed
from hyperspectral images acquired using the HYDICE
sensor.

Keywords: mapping, floodplain vegetation, remote
sensing, HYDICE sensor


Name:          Prasanna H. Gowda Ph.D.
Address:  Associate Researcher
          Environmental Management Technical Center
          BRD, USGS
          Onalaska, WI 54650.
Phone:    (608)783-7550 ext. 22
Fax:      (608)783-8058
e-mail:   phg0@emtc.nbs.gov

Presentation format: Platform only