The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
with the comparable SPOT HRS have been reached (Jacobsen
2004). Of course with the different GSD and different height to
base relation the absolute vertical accuracy based on SPOT
HRS cannot be as good like for Cartosat-1. Unquote.
Lehner et al 2008 report realizing a DSM with 10 m grid
posting over Mausanne test site with results as follows: For the
Jan stereo pair, the mean difference with ref. DEM: - 1.4 m; S.D.
of height differences w.r.t. ref. DEM: 3.8 m as computed at 4.82
million point locations. For the Feb. stereo pair, these mean and
SD are respectively -1.1 m and 3.53 m as computed at 6.14
million point locations. The reduction in number of points
matched for the Jan stereo pair is possibly attributed to the large
roll angle used in acquiring the same (-13.6 degrees) to achieve
overlap between the two stereo pairs. These results are without
any classification of landcover or terrain slope constraint. They
also report of achieving a 3.2 m SD by considering both stereo
pairs together in a combined multiple stereo processing mode.
Baltsavias et al 2007 report achieving a mean difference of
0.01 and 0.02 m respectively for the 10 m DSMs generated with
Jan and Feb stereo pairs with S.D. values 2.73 and 2.94 m
respectively after computing and reporting a 3-D bias values
between the respective DSMs and the reference DEM, which
are approximately 3 m, -2.5 m and 0.3 m in x, y and z directions
for both cases. Titarov 2008 reports an RMS error of 7.2 m with
a mean error of 0.8 m as computed at 6, 358, 422 points of the
10 m DSM derived over Mausanne test site while comparing
with the reference DEM. Gianinetto 2008 reports generating
DSM using the commercial off-the-shelf software ENVI V4.3
with different grid postings from 2.5 m to 90 m and concludes
that the results are meeting the Reference-3 D specifications and
better as compared to the DSM from SRTM source.
For the Warsaw stereo pair, Zych et al 2006 report an RMS
error of 1.26 m as measured at 25 GCP locations and evaluates
the DSM accuracy by comparing the results along three linear
profiles. They conclude that Cartosat-1 stereo data is good for
automated DTM generation (15-20 m) at least in flat areas with
vertical accuracy comparable, or even better than IKONOS.
Dowman & Mitchalis 2006 report achieving an RMS error of
6.65 m using 6 GCPs. Titarov 2007 & 2008 report achieving an
RMS error of 2.3 m with a mean error of 1.0 m as measured at 1,
985, 266 point locations of the 20 m grid DSM derived over
Warsaw test site.
Lehner et al 2008 report of realising a 5 m posting DSM over
Catalonia test site with a mean difference of -1.0 m and a SD of
3.05 m as measured at 7.08 million points while comparing with
the reference DEM. See Figures 1 and 2.
4.5 Orthoimage Quality:
Kay & Zielinski 2007 report: quote: Our tests show that it was
comparatively straightforward to produce reliable products, well
inside the expected performance of a modem satellite
instmment, from 2 to 3m RMSE1-D (i.e. in either Northing or
Easting directions) mainly using RPC bias method in LPS with
just 6 GCPs. Unquote. Lehner et al 2008 have generated
orthoimages using both Fore and Aft images using the DSM
generated and compared them to assess the residual sub-pixel
shifts between them as a measure of the DSM usability to
generate orthoimages. They conclude that CARTOSAT-1 stereo
imagery is well suited for the derivation of DSM and
orthoimages with about half pixel lateral and 1-2 pixel vertical
accuracy (la) in terrain with good pattern matching
characteristics and moderate slope angles using a few well-
defined ground control points. Armenakis & Beaulieu 2008
report achieving a planimetric accuracy of the Cartosat-1
orthoimage of 2 m with about -1 m bias in both x and y
directions. The road network DLG1985 was superimposed and
demonstrated a good matching between the image road features
and the vector road data.
4.6 Scale for Topographic Mapping:
J. Zych et al 2006 report that data from Cartosat-1 seems to be
very good and it can be used for topographic maps updating in
the scales 1: 10,000 (only on agricultural terrain) and
1:25,000. According to them, the Warsaw scene snow coverage
lowers its usefulness for acquiring information concerning land
use and cover. Nevertheless, the results suggest that the
Cartosat-1 satellite images may be useful for updating
topographic map contents and as a layer in miscellaneous GIS
systems because they provide high planimetric accuracy and
slightly worse interpretation capabilities due to recording only
in the visible range of the electromagnetic spectrum by means
of panchromatic sensors. Accordingly, the Cartosat orthoimage
might be used to updating 1:10,000 scale topographic maps. As
per B. Sadasiva Rao et al 2006, the geometric accuracy and
information potential of orthoimages and DEM provided by the
Cartosat-1 mission can be exploited for (i) updating 1:25,000
and 1:50,000 scale maps; (ii) making fresh topographic maps at
1:25,000 scale; (iii) making thematic maps at 1:10,000 scale;
and (iv) contouring at 10 m interval. Armenakis & Beaulieu
2008 report that concerning the topographic data acquisition, it
is expected that there would be no problems using Cartosat-1
data for planimetric data acquisition. The dynamic range is good,
making terrain features clearly visible and terrain morphologies
differentiable. The spatial resolution, at 2.5 m, enables
acquisition of planimetric data at CTI’s 1:50,000 scale within
the technical specifications (minimal dimension, various
features required, etc.). Titarov 2008 reports that the map
accuracy requirements differ depending on the regional
regulation, but in general the orthoimage accuracy achieved
corresponds to 1: 10,000 map scale. Nevertheless one should
keep in mind that it may be difficult to recognize on the images
all the objects that must be shown on the 1: 10,000 map.
5. COMPARISON WITH OTHER OPTICAL STEREO
SENSORS
Jacobsen, Crespi et al 2008 report: quote: In relation to other
satellites, the matching with Cartosat-1 models is extremely
successful. Unquote. Here for their conclusions they have used
two more Cartosat-1 stereo pairs over and above those provided
under C-SAP, viz. Warsaw and Mausanne. Also see Table-5
above adopted from Jacobsen 2007 and the remarks following
the same comparing with SPOT-HRS. A. Gruen 2008 concludes
that regarding the image quality, Cartosat-1 is better than
ALOS/PRISM but inferior to Spot-5 HRS or HRG. Their
remark w.r.t. SPOT-5 comparison is based on identifying
certain types of noise in the Rome data set after converting it
from 10-bits to 8-bits due to a constraint in their software. Also
to quote Jacobsen, Crespi et al 2008, from the very high
resolution optical space sensors, apart from being expensive,
only a limited number of stereo pairs, taken from the same orbit,
are available. The automatic matching of images taken with
significant time interval is very difficult and not leading to
satisfying results. J. Zych et al 2006 also report of the DSM
derived from Cartosat-1 being superior to those obtained from
IKONOS.