4.5 Mapping using GIS 
For better visual interpretation of linear features in 
conjunction with their back ground texture, merging 
of pan data with that of LISS-3 is found helpful. 
Terrain mapping needs derivation of cultural 
features, elevation changes and thematic 
information from the derived DEM and Orthoimage. 
This task can be better achieved by using a GIS. 
Basic tools required for terrain mapping are slopes, 
surface area, volume, line of sight coverage, 
draping, perspective view etc., can be efficiently 
derived/obtained in the GIS environment. 
5.0 Early results from IRS-1C Data Evaluation 
for Cartographic purposes 
The first sets of cloud free stereo pairs from IRS- 
1C Panchromatic data were taken for generating 
DEM and Orthoimage. The data specifications are 
given in Table-2. The stereo pairs were generated at 
National Remote Sensing Agency  (NRSA), 
Hyderabad, using an operational software 
developed by SAC. The data has undergone only 
radiometric corrections, no geometric corrections 
are applied. Three mapsheet areas (47M/13/SE, 
56A/1/NW and 53K/13/SW) are considered. Since 
the overlap between stereo images is only 6096, full 
map sheet area is not available for any of these 
map sheets. Hence an area of 7.5" X 7.5" is 
selected nearer to the above map sheets (8096 of 
total map sheet area is covered), through corner co- 
ordinates. Ground control points are selected from 
each stereo pair, and they are digitised from 
available 1: 25,000 and 1: 50,000 scale mapsheets. 
Out of these four/five are used as control points and 
remaining five/six are used as check points. Table-3 
gives the RMS values of accuracy obtained from the 
model on check points. This clearly shows, the 
model accuracy depends on the input GCP 
accuracy. GCPs collected from 1: 25,000 scale map 
resulted a better DEM accuracy than the GCPs 
collected from 1: 50,000 scale, as expected. 
DEM and Orthoimages are generated for the map 
sheet areas using the software system described in 
the previous section and detailed evaluation is 
carried out. Results are shown in Table 4. This error 
is a result of all the errors namely, DEM error, 
interpolation error and model error. 
A qualitative evaluation of Orthoimage for set 1 is 
done by 
(a) overlaying all linear features of map, with the 
help of a tracing at 1:25000 scale, which almost 
sits onto the Orthoimage with the above accuracy. 
Further, Orthoimage showed the recent 
information like new roads/constructions and other 
elevation changes like quarrying etc., which can be 
used for updating. 
(b) Comparing the draping of (i) the Orthoimage 
over derived DEM from the stereo pair and (ii) 
Orthoimage over map derived DEM. Draping from 
map derived DEM showed blocky 
effect/discontinuties, because of lack of continuous 
information of elevation, whereas the other draping 
showed smooth variation of the terrain, since it has 
the continuous information of terrain elevations. The 
map DEM is generated by digitising the 10m 
contours from the 25000 scale map. 
(c) Comparison of elevation contours, showed a 
good match of peaks and valleys of elevation, from 
both DEMs derived from image and map. 
6.0 Conclusions and Recommendations 
The early results described in section 5.0 can be 
categorised in three classes: 
1. The reconstruction of terrain profile. This is 
related to the precision of height determination for 
various terrain conditions viz. highly undulating, 
moderately undulating and flat. It is shown that in 
case of moderately undulating terrain conditions 
with moderate contrast the terrain model can be 
automatically reconstructed to meet requirements of 
1:25,000 scale mapping. The experiments described 
in 5.0 were conducted with stereo pairs acquired 
with approx. 0.5 base to height ratio. It can be 
concluded that for regions having moderate 
radiometric contrast the elevation information 
derived from IRS-1C imagery is sufficient for 
1:25,000 scale mapping. For other types of terrain 
this conclusion comes from the capability of 
achieving any base to height ratio upto 1.1. 
It has been seen that the breaklines and breakpoints 
can be identified with high confidence in manual 
mode. Identification of breaklines and breakpoints in 
automatic mode is a subject of further research. 
827 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996