International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 Intern 
maps, whose accuracy is not well known (e.g. see Table 1). order to make results more consistent we have used 4.3.2 
Control is also performed with a little set of points. Its obvious different photogrammetric software applications: 
that, in an accuracy analysis, control process reliability must be OrthoBase Pro and Ortho Engine. dk 
AS . To compare different DEM generated by several of oe 
This paper relies on accuracy and its reliability by using a set of datas source: TERRA-ASTER and SPOT-HRV data 
315 randomly distributed check points whose coordinates were versus cartographic data. 44 I 
determined by differential GPS (DGPS) techniques. A long 
observation time was used to guarantee an error margin of less 4 MATERIAL AND METHODS The a 
than 10 cm. sensor 
4.1 Area under study generz 
2. BACKGROUND ON DEM GENERATION FROM ; : 
TERRA-ASTER Work area is a 23 km x 28 km rectangle in the province of In ord 
Granada (Southern Spain) (Fig. 1). It is an area with a TERR 
Deriving DEM from stereoscopic satellite images is not new; complex topography: Steep slopes in the South and flat some 
however accuracy results and the method used to calculate surfaces in the North. Elevations are in the range 300-2800 m the d: 
error and reliability in DEM differ according to the literature with an average of 1060 m. depen 
revised. This variation may be due to the method used to coeffic 
estimate error in DEM as much in the number as in the source 
of check points used. We h 
optim: 
The Advanced Spaceborne Thermal Emission and Reflection distrib 
Radiometer (ASTER), on board the NASA’s TERRA satellite, size oi 
was launched in December 1999. TERRA-ASTER provides and c 
along-track near-IR stereoscopic images at 15 m resolution. ASTE 
TERRA-ASTER is a quite recent sensor; thus there is little 
research that analyzes the accuracy of DEM generated, mostly 
on simulated ASTER data (Welch et al, 1998; Abrams & 
[ossi 
  
  
  
Hook, 1995; Lang & Welch, 1999). The Algorithm Theoretical a 
Basis Document, ATBD, for ASTER Digital Elevation Models id 
(Lang & Welch, 1999) suggests that RMSE for Z values in 
ASTER DEM should be in the order of 10 to 50 m, but this is a 
too wide range to define the accuracy of a product. C 
  
  
There is little research focusing on possibilities in DEM 
generation with a variable elevation RMSE between 7 and 60 
m. Table I shows results of research about accuracy in DEM 
derived from ASTER images. Studies about ASTER data and 
DEM accuracy present similar problems to those 
aforementioned. 
Important questions such as the number of check points and the 
measure capture method are not standardized. Some authors do Che 
not even inform about control methods. 4.2 Image data 
  
Figure 1. Study area 
(SPOT image dropped over DEM). 
  
  
We have used two pairs of TERRA-ASTER scenes (15 m 
  
: RMSES ; pixel). These images were taken on on 22" August 2000. 
Date Reference im) Method 
2001 Toutin & 7.9 6 DGPS check points 4.3 Software 
Cheng 
2002 Kääbetal 18-60 Photogrammetrically derived DEM ASTER data were processed with Erdas Imagine and 
2002 Hirano ot 25 Different methods T Geomatica. We have used two photogrammetric programs: 
al * Erdas Imagine 8.5 with OrthoBASE Pro (Leica 
  
Geosystems). 
* Geomatica 8.2 with OrthoEngine (PCI Geomatics). 
Only variying characteristics are commented below. 
* Root Mean Square Error. 
? Method by which RMSE has been calculated. 
Table 1. Some works about ASTER-DEM accuracy 
Li 4.3.1 Erdas Imagine 8.5 with OrthoBase Pro 
determination. 
OrthoBASE Pro has a specific module to work with SPOT data, 
but ASTER is only supported by means a generic module 
introducing the values for angles, B/H ratio, etc. The DEM may 
be generated only as a vector structure, a Triangulated Irregular 
Network or TIN. 
3. OBJECTIVES 
This paper aims: 
. To verify the accuracy of DEM accuracy 
generated from stereoscopic TERRA-ASTER data. In 
560