1.  SUBMIT TO:  SD102

2.  SESSION:  Imaging Spectrometery V,  Michael R. Descour and Silvia S.
Shen

3.  TITLE:  Materials Identification and Anomaly Detection with Simulated
MTI Data

4.  AUTHORS:  Brian R. Stallard and John G. Taylor
Sandia National Laboratories, MS 0980, Albuquerque, NM, 87185
Tel:  505-844-2631
FAX:  505-844-2057
brstall@sandia.gov

5.  PRESENTATION:  prefer oral presentation

6.  ABSTRACT:
	The Multispectral Thermal Imager (MTI) is a US Department of Energy
research and development satellite which will demonstrate and evaluate
multispectral and thermal imaging technology for nonproliferation treaty
monitoring and other national security and civilian applications.  MTI is
estimated to have a limited on-orbit lifetime of three years.  Therefore
data analysis tools need to be in place and initial experiments need to be
planned before launch.  Data simulation is critical to accomplishing this
goal.  Our approach is to use MODTRAN and 6S software together with ground
truth data to convert radiance to reflectance for data cubes collected by a
variety of airborne sensors.  Then a partially randomized atmosphere is
added to approximate what the satellite will see at an altitude of 550 km.
The data cubes created in this manner are studied to determine MTI's
potential for materials identification and anomaly detection.  MTI has a
number of unique features:  1) spectral coverage from visible through
long-wave infrared,  2) precise radiometric calibration,  3) relatively
small ground sampling distance, and  4) unique two-look capability.  We show
how these features can enhance MTI's capabilities in the present
application.  The results focus on how reliability of classification is
impacted by atmospheric effects, spatial resolution, spectral resolution,
and band selection.  We compare these results to capabilities of other
sensors. 

Sandia is a multiprogram laboratory operated by Sandia Corporation, a
Lockheed Martin Company, for the United States Department of Energy under
contract number DE-ACO4-94AL85000.

7. KEY WORDS:  multispectral imaging, classification, atmospheric correction

8. BIOGRAPHY:  Brian R. Stallard is a principal member of the technical
staff at Sandia National Laboratories.  He received a BSc from the
University of Calgary in 1976 and a PhD from Cornell University in 1982.
Both degrees were in physical chemistry.  His current specialty is remote
spectral imaging.  Brian has co-authored about 50 publications and holds two
patents.