137
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986
Shuttle imaging radar response from sand dunes and subsurface
rocks of Alashan Plateau in north-central China
Guo Huadong
Institute of Remote Sensing Application, Academia Sinica, Beijing, China
G.G.Schaber & C.S.Breed
US Geological Survey, Flagstaff, Ariz., USA
A.J.Lewis
ERSAL, Oregon State University, Corvallis, USA
ABSTRACT: SIR-A and SIR-B images of three sand dunes and two bedrock regions in alashan Pla
teau of north-central China have been studied and compared to Landsat imagery and field in
vestigation data. The results of the study show that radar illumination direction is an im
portant factor which results in the echo strength change for the same type and size of sand
dunes. The sand dunes in which the slipfaces are oriented to radar beam have a bright-point
response and the dunes in which the gentle slopes face the radar beam have a dark signature
on the radar image. Two Precarnbrian metamorphic rock bodies buried beneath the thin layer of
alluvial material or aeolian sand have been detected by using these two radar images.
1 INTRODUCTION
The Shuttle Imaging Radar-A and B (SIR-A and
SIR-B) carried on the Space Shuttle Columbia
in November 1981 and the Challenger in Octo
ber 1984 acquired images of part of Alashan
Plateau in north-central China. This very
dry plateau covers three deserts of the main
thirteen ones over the country. One of these
deserts, the Badain Jaran, contains some of
the tallest dunes in the world.
These SIR-A and SIR-B images are the only
orbital radar data acquired over China and
provide a great deal of information on these
sand dunes as well as subsurface geological
features. In this paper five areas are stu
died, including three ounes anu two aeolian
sand and bedrock regions (Fig.1). The pur
pose of this study is to understand the L-
band radar responses from sand dunes and
some subsurface features.
One of the authors (Guo Huadong) conducted
field work in several study areas covered by
radar imagery. The Landsat images compared
with the radar data were optimally processed
on the VAX 11/750 computer system of the U.S.
Geological Survey in Flagstaff, Arizona.
Fig. 1 Index and simplified geologic map
of the Alashan Plateau showing location
of the five study areas.I--Yamalik
Dunes; II--Yapulaishan area; III—Badain
Jaran Desert; IV—Shuangjinzhi study
area; V--Aertengaobao study area.
R--Bedrock
N--Neogene
Q--Quaternary
2 COMPUTER ENHANCEMENT TO THE SIR-a AND MSS
IMaGES
SIR-A is a L-band synthetic aperture radar
with HH polarization and is imaged witn op
tical recording method. The radar on ooard
recieves the information on phase and ampli
tude of the echo from the ground target,
which represents the change of the scatterring
strength. The information is recorded on the
data film and then processed with optical
relative processor to form image film.
The SIR-A image is processed with digital
enhancement technique in order to study the
radar scatterring characteristics from the
desert region more effectively; the Landsat
MSS CCT data is also processed with the same
technique to be used as the data for matching
analysis. In this study, the radar image
covering the test region is digitized to form
an image matrix consisting of 2439 by 46OO
pixels. The high pass filtering technique is
considered to be an important one in the pro
cessing of SIR-A data because it has the same
defect with other on-board radar images, that
is, there are too many speckles on the image,
which are easily confused with speckle-shaped
echoes from the dunes. The speckle noise may
be lowered by filtering the image.
A selection of the filtering window is im
portant to the processing. The window of 101
by 101 has been accepted a traditional one
in the filtering process for the MSS data.
But there is no such an optimum window for
the space radar image in this dune area. So,
four windows, 3 by 3, 5 by 5, 23 by 25 and
101 by 101 are chosen respectively. Anotner
kind of filtering, the divided filtering, is
also used. The back scatterring strength of
radar changes sharply with different back
ground formed by different ground objects.
Due to the higher dielectric constant as well
as the effect of corner reflector , the ob
jects such as bedrock, vegetation and buil
ding usually have stronger echoes; and the
echoes from the objects such as smootn ground,
water body and dry sand are much weaker. The
obvious difference between the DN values of
these two types of objects makes a high con
trast on the image. This is, of course, an
unfavourable factor to image interpretation.
The principle of the divided filtering is to
divide the original data by its low pass
