|
GPS & Photogrammetry: Digital
Photogrammetry |
Page 1 of 1
Errors in Digital Surface Models
Derived From Airborne Ifsar
David BRUCE
Senior Lecturer,
School of Natural and Built Environments,
University of South Australia
GPO Box 2471,
Adelaide, South Australia, Australia,
5001,
Tel: (61)-8-83021856 Fax:
(61)-8-83022252
Email: david.bruce@unisa,.edu.au
ABSTRACT
Bruce (1996) reported on errors and accuracy of
Digital Elevation Models derived from IFSAR
measurements from the NASA/JPL AIRSAR mission to
Australia in 1993. In this research the best
results were reported when gross errors from
poorly correlated returns were ignored, a global
rotation to the IFSAR DEM was applied via the
use of ground control points in the near and far
range and when vertical features (bushes, trees,
houses, etc) were ignored in the generation of
the statistics. During August 2000, an improved
version of the NASA/JPL TOPSAR interferometer
was flown over the same test sites as it was in
1993. This paper reports the results of
analysing the Digital Surface Model (DSM) from
the TOPSAR instrument by comparison with Digital
Ground Models (DGM) derived from medium scale
and large scale photogrammetric observations.
The same type of errors reported in Bruce (1996)
were again observed, but with the extent of the
errors diminished and with the magnitude of the
errors significantly improved. The effects of
multi-pathing of the backscattered signal off
the aircraft wing tips, which was just
noticeable in the 1993 data, was again observed
in the 2000 data. However, the processing of the
data from 2000 was able to model this phenomenon
using a FFT. In the current data the effects of
vertical features such as shrubs, trees,
buildings, etc. are very obvious. Literature
suggests a number of strategies to transform DSM
data into DGM (bald earth) data. Some (NOAA
(2004), Apogee – Nextmap (2004)) suggest the use
of a simultaneous P Band interferometer, which
enables canopy penetration in a forest
situations, but of course does not enable solid
objects such as buildings to be removed. Wang et
al (2001) propose a method of using local minima
and a hierarchical surface fitting method. A
similar approach was applied in the work
presented in this current paper, except that
firstly vertical features were discriminated via
the use of the co-registered polarmetric data.
For pixels mapped as vertical objects, a spatial
filtering technique was then applied, wherein
neighbourhood local minima were compared with
heights of central pixels. Neighbourhood size
was successively increased. Observing the trend
in height difference against the size of the
neighbourhood enabled a threshold to be
extracted which provided an estimate of the
height of the ground object. This technique
tended to fail when the slope of the terrain
around the subject pixel became excessive and
thus the method was required to be iterative,
using the apparent slope of the terrain in a
corrective term. References
- Apogee – Next Map, 2004, P-band - Seeing
through the forest despite the trees - website
http://www.apogee.com.au/products/nextmap/pband.html
(viewed June 04, 2004) Bruce, D.A., 1996, The
Accuracy of Topographic Shape Determination from
Interferometric SAR. In Proceedings of 8th
Australasian Remote Sensing Conference,
Canberra, 25-29 March, 1996.
- NOAA, 2004, Remote Sensing for Coastal
Management – IFSAR website
http://www.csc.noaa.gov/crs/rs_apps/sensors/ifsar.htm
(viewed June 12, 2004) Wang, Y., B. Mercer, V.
C. Tao, J. Sharma and S. Crawford (2001).
Automatic generation of bald earth digital
elevation models from digital surface models
created using airborne IFSAR. In: Proceedings of
2001 ASPRS Annual Conference, St. Louis,
Missouri, U. S. A., April 23-27, 2001. (Also at
http://www.intermaptechnologies.com/PDF_files/asprs2001_Intermap_E.pdf
)
Page
1 of 1
| | | | |
