In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
507 
3. DEM EVALUATION STRATEGY 
In the project, initially, the common coordinate system was 
defined as UTM (universal transverse Mercator) for the whole 
of data sets and references. That’s why before the evaluation 
processes, the transformation of the coordinate systems of all of 
data sets into UTM suitable zone have been performed using 
program BLTRA. This program is a member of system BLUH 
(Bundle Block Adjustment Leibniz University Hannover), 
developed by Dr. Karsten Jacobsen, Institute of 
Photogrammetry and Geoinformation (IPI), Leibniz University 
of Hannover, Germany. 
After the transformation of all models into UTM suitable zone, 
for the evaluation analysis, a number of investigations have 
been performed using several modules of the software package 
BLUH. Table 3 shows these evaluation programs and their 
functions. 
Program 
Function 
RASCOR 
Filtering, analysis, correction and plot of a 
DEM 
DEMSHIFT 
Shifting of a DEM to another in X, Y, Z 
and scaling in Z 
DEMANAL 
Accuracy Analysis of DEM against a 
reference DEM 
MANI 
Manipulation of object coordinates, image 
orientations, IMU-data and pixel addresses, 
numbering the points 
BLCON 
Conversion of ground coordinates window 
function, reduction to equal distributed 
points, change of spacing 
DEMINT 
Computation of Z-value for points with 
given X and Y by interpolation of a raster- 
digital elevation model 
ZANAL 
Analysis of a DEM 
ZPROF 
Plot profiles 
UNDUL 
Calculation of Geoid Undulation 
HPSHOW 
Creation of aspects 
Table 3. Evaluation programs and their functions 
Besides, for the visualization, interpolation (by triangulation, 
moving surfaces etc.) and regular gridding of DSMs and DEMs, 
program LISA has been used. This program has been generated 
at the Institute of Photogrammetry and Geoinformation, Leibniz 
University of Hannover, Germany by Dr. Wilfred Linder. In 
program LISA, a height model and its height levels can be seen 
in a color scale accompanied minimum up to maximum heights 
at the same screen and different color palettes can be created 
and used for visualization. Using the optimum palette in every 
process, the details can be established clearly. All color coded 
versions of the height models have been created by this 
program. 
4. RESULTS AND DISCUSSION 
4.1 Shift of DSMs and Preparations 
In order to perform the correct accuracy analysis, the DSM 
which will be evaluated must have the same location as the 
reference DEM. For this the determination of shifts to the 
reference height model has to be determined. Initially, using 
program DEMSHIFT shifts against the reference DEMs are 
determined. In this operation, maximum accepted ÀZ is selected 
as 50m and based on the handling the number of iterations are 
preferred as 11. 
As the result of this first determination, the large shift values 
which cause large incorrect RMSZ values up to 14m have been 
seen and the radius of convergence for the shift adjustment was 
exceeded. Accordingly, for the elimination of large shifts the 
DEMs were pre-corrected by manual shift via point matching 
using program MANI. The points of corresponding location and 
their planimetric coordinates (X and Y) were selected from 
evaluated DSMs and reference DEMs. The coordinate 
differences were calculated and used for the rough shifts. 
Figures 9 shows the selected points at the corresponding 
locations for the pre-correction of the evaluated DSMs and 
reference DEMs in the test field. 
Figure 9. High resolution IKONOS PAN images 
The Table 4 shows the RMSZ values before and after the pre 
correction of the shifts, adjusted shift values by DEMSHIFT 
and the final RMSZ values. 
DEM 
Original 
RMSZ 
(m) 
RMSZ 
After 
Manual 
Shift(m) 
Shift by 
DEMSHIFT (m) 
Final 
RMSZ 
(m) 
AX 
AY 
IKONOS 
13.67 
7.44 
-11.38 
+8.03 
7.32 
TSX SL 
13.78 
8.44 
-7.93 
+21.4 
8.25 
Table 4. Adjusted shifts after pre-correction by MANI 
4.2 Accuracy Analysis of DSMs 
After shifting of DEMs, the accuracies of them have been 
analyzed in relation to the reference DEMs using program 
DEM ANAL. For the analysis by DEM ANAL, the maximal 
accepted DZ was limited as 50m and the maximal accepted 
tangent of terrain inclination was selected as 1.00. In the second 
iteration, shift and vertical scale were respected. These settings 
were made depending upon the characteristic of the test fields. 
Following Tables and Figures show the results of accuracy 
analysis for IKONOS and TSX SL DSMs against REFDEM. 
Nearly full area of test field is built-up area and the forest 
coverage is so less that’s why the accuracy analysis were 
performed just for the general area not separate for open and 
forest layers. 
DSM 
General 
SZ [m] 
NAP[%] 
IKONOS (3m) 
7.04+1.03 xtan(a) 
0.00 
TSX SL (3m) 
7.09+11.22 xtan(a) 
0.00 
Table 5. Adjusted shifts after pre-correction by MANI