The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part Bl. Beijing 2008 
In eq. (5), e, is the elevation difference between an InSAR 
DEM and a truth DEM and n is the number of points involved 
in the evaluation. 
3. DATA, TOOLS AND EXPERIMENT 
3.1 Data 
The SAR data are two pairs of ERS-1/2 tandem mode single 
look complex images. The first pair covers about 10 counties in 
northern Indiana, USA, where the terrain is relatively flat. The 
second pair covers about 10 counties in southern Indiana, USA, 
with more varied terrain. Both pairs were acquired in fall 1995 
(Figure 1). 
Figure 3 Orthophoto DEM overlaying Indiana counties 
figure 1 SAR images overlaying Indiana counties 
The USGS SRTM 3 arc second DEM are used as the reference 
DEM for the InSAR DEM registration and alignment. It has 
global coverage between 60 degrees N and 56 degrees S 
latitude. The vertical accuracy is +/-10 meters RMSE. The post 
spacing or GSD (ground sample distance) is about 90 meters. 
(Figure 2) 
All of those DEMs, including processed InSAR DEM, are 
reprojected into WGS84 UTM (Zone 16 North) with post 
spacing of 16 meters, and the vertical datum of WGS84 
ellipsoidal height in meter. 
Selecting GCPs on SAR images proved to be difficult. GCPs 
maintained for optical imagery are not visible on ERS SAR 
images unless comer reflectors are placed during acquisition or 
the target and background are large and have quite different 
backscatter. In this study, GCPs are derived from a combination 
of ortho imagery for planimetry and the associated DEM for 
elevation, with features also visible on the SAR images. 
3.2 Tools 
(Leica) ERDAS IMAGINE was used for InSAR DEM 
processing. GCPs are applied in the processing for the second 
workflow. 
Matlab program was developed for the registration of InSAR 
DEM and reference DEM, searching for the conjugate points, 
transforming and resampling InSAR DEM, and InSAR DEM 
evaluation against “truth DEM”. 
C program from “Introduction to Modem Photogrammetry” 
was used for deriving seven-parameter transformation equations 
from 3D conjugate points. (Mikhail et al., 2001) 
3.3 Experiment 
In this study, the primary experiment was to compare the 
accuracy of the refined InSAR DEM with no GCPs applied in 
InSAR processing, to the accuracy of the GCP assisted InSAR 
DEM. Both were evaluated against the independent, high 
resolution DEM. 
Two InSAR processes were performed for each InSAR pair: the 
first process had no GCPs applied and the second process had 
10 GCPs applied. Figure 4 displays the subsets of InSAR DEM 
with and without GCPs applied. 
Figure 2 SRTM 3” DEM overlaying Indiana counties 
The “truth DEM” was produced from the “Indiana 2005 
Statewide Orthophotography Project”, which includes both 
ortho imagery and a high resolution DEM (Orthophoto DEM). 
The DEM has 5-foot (~1.5m) post spacing and 6-foot (~1.8m) 
vertical accuracy at 95% confidence level. (Figure 3)