progressive sampling with the grid interval of 
250 metres. 
In the progressive sampling method the measur- 
ing mark is driven within a defined boundary 
from one sampling point to another sampling 
point and an initial sampling patch of nine points 
is created. If one of the second differeces in height 
is equal to or greater than the threshold value, 
second differences are calculated and similarly 
the values collected in the second run determine 
the need for the third run. After a maximum of 
three runs a sampling data of adequate density 
which may match the undulations of the terrain 
is generated. Keeping the hilly characterstics of 
the data set in view the data set was further 
upgraded by incorporating the significant relief 
features like ridges, drains and break lines. 
The DEM data set was then processed by DTM 
software wherein the data was converted to 
POINT file. It was subsequently changed to a TIN 
file, which contains the data of the vertices of 
slope triangles as X, Y, Z triplets. Finally from 
the TIN file profiles were extracted by converting 
to a POINT file (GRID file). The GRID file data 
was finally converted to graphics. 
The DTM surface (mesh of horizontal and verti- 
cal profiles) was then evaluated as given in the 
subsequent pragraphs. 
6. EVALUATION AND ANALYSIS 
6.1 In case of data set A, the large scale photog- 
raphy was used to exploit the change in the 
contrast in the imagery for interpretation of the 
details from the stereo model setup in the photo- 
grammetric workstation. The data captured in : 
respect of this data set was evaluated by compar- 
ing with ground truth. The plot of this data is 
enclosed as Annexure I About 9046 of the details 
agreed on evaluation. Details are brought out in 
para ( 7 ) under findings. 
In case of data set B, the output as indicated in 
Annexure II was not subject to ground truth. 
6.2 The DEM generated was evaluated by com- 
paring the DEM with the heights of some promi- 
nent points as inpretted from the existing map. 
The analysis is given in Annexure - II. 
7. FINDINGS 
7.1 The data in respect of the following features 
could br stereo-digitized with convenience by 
using stereo-SPOT imageries in isolation : 
1) Roads 
ii) Forests limits 
iii) River courses 
iv) Cultivation 
v) Village limits ( Blocks ) 
7.2 Reference to the photography helped in in- 
terpretation of the following details to a great 
extent : 
i) Cluster of isolated huts 
ii) Type of roads 
iii) Orchards 
iv) Depiction of roads in built up area 
7.3 RMSE value in height is about 11 meters. 
Keeping in view the sensitivity of the floating 
mark in stereo-imagery environment, it is recom- 
mended that the DEM be used to generate con- 
tours at 50 meters vertical interval. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
In 
$10! 
the 
bec 
sea 
me 
upc 
sar 
ries 
ma: 
at tl 
res 
cou 
ster 
rapl 
whi 
ima 
fror 
leve