periment 12
periment 13
periment 14
periment 15
tual
>asurement
mem
cent
ed sensor)
yrizontally
f elevation
celeration
it received
der two or
elerometer
y is better
ent of the
horizontal
onger than
icle speed
e tire one
icle speed
n for a rear
individual
use, it is
produced
d Remote
rchives of
Henning Skriver
MAPPING BY AIRBORNE SYNTHETIC APERTURE RADAR (SAR)
Henning SKRIVER, Jesper SCHOU, and Wolfgang DIERKING
Department of Electromagnetic Systems
Technical University of Denmark, Denmark
a
KEY WORDS: Mapping, Synthetic Aperture Radar (SAR), Polarimetry, Remote Sensing, Land Use Classification,
Feature Extraction, Change Detection.
ABSTRACT
Using advanced SAR data from the Danish airborne SAR, EMISAR, a research project was initiated in 1998 as a
collaboration between the Technical University of Denmark and the Danish National Survey and Cadastre to investigate
the potential of using SAR data for production and updating of cartographic maps. The ultimate objective of the
research is to be able to perform quickly and efficiently compiling and updating of topographic maps using satellite and
airborne remote sensing data. The specific objectives for the project are to evaluate the potential of the polarimetric and
interferometric SAR to detect the necessary object classes to be used for the thematic map content, and to assess the
complementary information gained relative to the results of the traditional methods, and to evaluate the accuracy of
methods for retrieval of the necessary map information from SAR data. In the project methods have been developed
and studied for extraction of relevant information from the polarimetric and interferometric SAR data about objects like
roads, buildings, lakes, fields, forests, etc. A number of feature extraction and land-use classification methods have
been implemented and analyzed. Also, the very important issue to use the SAR for change detection as a means of fast,
frequent and cost-efficient updating of existing map information has been studied.
1 INTRODUCTION
Topographic maps containing both thematic information about objects (e.g. forests, agricultural areas, lakes, heath,
roads, and buildings), and elevation information for the land surface are used in numerous applications, both for
registration and planning purposes in municipalities, counties, government agencies, economic forecasters, and private
civil engineering companies, and for science applications, such as plant and animal ecology, mapping of wetlands,
hydrologic modeling, and general circulation modeling. The traditional cartography using traditional surveying and
aerial photogrammetry normally produces detailed maps with a relative long time interval between updating, i.e. up to 5
to 6 years. The different users have different requirements for the map content, but especially with the increasing
utilization of Geographic Information Systems (GIS) a more frequent updating process for the maps is of key
importance for many users.
The availability of high-resolution Earth observation data from spaceborne and airborne sensors both today and in the
future has focused the attention on the role that such sensors could play in speeding up and making more efficient the
process of compiling and revising topographic maps. Only sparse experience has been gathered using spaceborne SARs
(Synthetic Aperture Radars), such as those on ERS-1/2, and Radarsat, and using airborne SAR systems, such as the
Danish EMISAR. Plans are emerging both in Europe and in US for high-resolution spaceborne SARs using the small
satellite concept (e.g. the NASA LightSAR mission) as well as high-resolution optical satellites. The advantages of the
SAR sensors are their all-weather mapping capabilities (since in particular longer radar waves are not affected by cloud
covers), their lower sensitivity to atmospheric correction compared to optical sensors, and the fact that they provide
complementary information to optical data.
In the present paper, results from a research project using data from the advanced Danish airborne SAR system,
EMISAR, will be presented. The project is collaboration between the Technical University of Denmark and the Danish
National Survey and Cadastre with the purpose of studying the potential of using SAR data for production and updating
of cartographic maps. Methods for extraction of relevant map information from SAR images have been studied and
developed, and below results are discussed for methods related to land use classification, feature extraction and change
detection.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part Bl. Amsterdam 2000. 307