The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
1364
Our investigations on Cartosat-1 data focus on: a) radiometric
problems and preprocessing for image quality improvement, b)
RPC refinement and 3D point measurement accuracy potential,
and c) automatic DSM generation. The SAT-PP functions are
used for RPC refinement, and DSM generation in this study.
The sensor modeling is performed via Rational Polynomial
Coefficients (RPCs). Two different models are applied to
improve the given RPCs by using ground control points (GCPs).
The RPC correction models include two translational
parameters (RPC1) and 2D affine transformation (RPC2).
Different numbers and distribution of GCPs are used for the
investigations.
The investigations in this paper are part of the ISPRS-ISRO
Cartosat-1 Scientific Assessment Programme (C-SAP). IGP acts
as principal investigator for some test sites and has evaluated
data from other test sites. We have already reported on our
investigations with the Rome and Mausanne test sites in
Baltsavias et al. (2007). In this paper, we report on our
investigations in the Catalonia test site (official C-SAP test site
10), Spain, and Sakurajima test site, Japan. Other results
regarding the Catalonia test site are reported in Lehner et al.
(2007).
2. INPUT DATA
A Cartosat-1 stereo pair acquired over the Catalonia test site is
provided by ISRO within C-SAP (Test site 10). The GCP image
coordinates are measured at the DLR, Germany, using the Least
Squares Matching (LSM) method (Lehner et al., 2007). Some
details on the dataset are given in Table 1. The GCP distribution
over the Cartosat-1 Aft image can be seen in Figure 1. As
reference data, a DTM with a 15 m grid spacing and
orthoimages with scale 1:5000 were provided by the Institut
Cartografic de Catalunya (ICC), Spain for the whole testfield.
The date of primary data acquisition for the DTM is unknown;
this can result in differences in the comparison with the
Cartosat matching DSM due to temporal differences.
Imaging date
01 Feb 2006
Sun Azimuth
Aft: 157.16918°
Fore: 158.65601°
Sun Elevation
Aft: 31.65443°
Fore: 28.91865°
No. of GCPs
70
Point measurement
69 points are measured with LSM and
method
1 point is measured manually
Reference DTM
29 km x 25 km
- grid spacing
15 m
- accuracy (lo)
1.1
m
- height range
0 m-970 m
Table 1. Catalonia Cartosat-1 test dataset description.
The Sakurajima testfield is generated as a joint project of the
Kochi Institute of Technology and Kanazawa Institute of
Technology. 62 GCPs are provided by the Japan Association of
Remote Sensing (JARS) and 61 points could be measured on
the Cartosat-1 images (Figure 2). However, just a small area
about 4 km to 4 km of the volcano Sakurajima is given as a 5m
grid laser scanner reference dataset provided by Kokusai Kogyo
Co. Ltd, Japan (see Figure 7). Main characteristics of the test
site and the Cartosat-1 stereo pair are given in Table 2.
ALOS/PRISM images acquired over the same test site are
processed and the results are reported in Gruen et al. (2008).
Figure 1. GCP distribution over the Cartosat-1 Aft image of the
Catalonia test site
Figure 2. GCP distribution over the Cartosat-1 Aft image of the
Sakurajima test site
Imaging date
31 March 2006
Sun Azimuth
Aft: 139.17555°
Fore: 143.38021°
Sun Elevation
Aft: 59.43782°
Fore: 57.26630°
No. of GCPs
61
Point measurement
58 points are measured with LSM and
method
3 points are measured manually
Reference DSM
4 km x 4 km
- method
laser scanning
- grid spacing
5 m
- height range
300 m -
-1100 m
- remark
no GCP inside reference DSM
Table 2. Sakurajima Cartosat-1 test dataset description