Journal of Geographic Information System, 2010, 2, 85-92
doi:10.4236/jgis.2010.22013 Published Online April 2010 (http://www.SciRP.org/journal/jgis)
Copyright © 2010 SciRes. JGIS
Orthorectification and Digital Elevation Model (DEM)
Generation Using Cartosat-1 Satellite Stereo Pair in
Himalayan Terrain
Vivek Kumar Singh1, Prashant Kumar Champati Ray2, Ayyeum Perumal Thillai Jeyaseelan1
1Jharkhand Space Applications Center, Department of Information Technology,
Governm e nt of Jharkhand, Ra nc h, India
2Indian Institute of Remote Sensing (IIRS), Department of Space, Government of India,
Dehradun, Uttarakhand, India
E-mail: vivekearth@gmail.com
Abstract
High resolution data have high relief displacement in hilly terrains. Development of Digital Elevation model
helps to assess bio resources more accurately in such terrains. While estimating bio resources in the Himala-
yan hilly terrain using multispectral LISS-III data of 23 m spatial resolution, the need for orthorectifcation of
satellite data was necessary to correct for spatial distances due to high undulating slopes. Therefore, Cartosat
stereo pair based Digital Elevation Model (DEM) was generated using the Rational Polynomial Coefficients
(RPC) supplied along with the data products. By using the DEM orthorectification of LISS-III was created.
In order to evaluate the positional accuracy of ortho rectified LISS-III Ground control points were selected
using the Global Positioning System in differential GPS mode. As there is variation in the spatial distances
and height over few points, the GCP corrected DEM was used for ortho rectifcation of Cartosat PAN and
LISS-III data. This paper presents the procedure followed for ortho rectification and digital elevation model
generation using Cartosat stereo pair data. The result of the study indicated high spatial resolution stereo im-
ages helped generation of three dimensional mountainous regions more accurately which helps in estimating
the bio resources using multispectral LISS III data.
Keywords: DEM, Cartosat, Stereo Pair, Orthorectification, Himalaya
1. Introduction
Stereo imaging from space-borne platforms offers inf-
ormation about terrain elevation besides supplying spec-
tral reflectance of the scene. This greatly assists the
analysis and interpretation of images in terms of identi-
fying slopes, surface material, waterways, vegetation
growth etc. Applications like urban planning, agriculture,
defence etc., need to use Digital Elevation Model (DEM)
derived from stereo images, which is an important com-
ponent of geo-spatial data. With the launch of Cartosat-1,
ISRO’s first satellite with along track stereo capability in
May 2005 by PSLV-C6 vehicle, a new possibility has
emerged for remote sensing and Mapping communities.
The high-resolution stereo data beamed from twin cam-
eras onboard Cartosat-1 mission facilitates topographic
mapping up to 1:25,000 scale [1]. The primary advantage
of Cartosat-1 mission is seen as generation of Digital
Elevation Model (DEM) for production of Orthoimage
and 3D terrain visualization of large tracts of landmass at
fairly large scale. The 10 bits per pixel radiometric reso-
lution of Cartosat-1 sensors allows for improved discrimi-
nation of objects, which enhan ces the cart ogra phi c po ten-
tial of the sensor. The satellite has shown a very stable
attitude behavior, which in turn helps in realizing data
products with low internal disto rtion.
Cartosat-1 is the first operational remote sensing satel-
lite capable of providing in-orbit stereo images with 2.5 m
nadir resolution and 27 km swath. The two payloads viz.
PAN-Fore and PAN-Aft are designed with state-of-
the-art technologies in order to provide images of high
quality [1,2]. They are mounted in along track direction
with fixed tilts of +26 deg (Fore) and –5 deg (Aft) re-
specttively to provide along track stereo, each with ap-
proximately 2.5 m ground resolutions.
Satellite Photogrammetry techniques have been exte-
nsively used by the scientific community in deriving hig h
V. K. SINGH ET AL.
86
resolution DEM, Ortho image and terrain parameters such
as slope, aspect, contours, drainage etc. Digital Elevation
Model (DEM) has become an inevitable component in
most of the remote sensing applications viz. infrastruc-
ture development, watershed management and develop-
ment, hydro-geomorphology, urban morphology, disaster
management etc. Keeping these applicatio ns in view, the
current study aimed at exploitation of Cartosat-1 stereo
data for various applications.
Rational functions models (RFMs) have gained popul-
arity, with the recent advent of high resolution data sup-
plying Rational Polynomial Coefficients (RPCs) along
with stereo/mono data. Providing these coefficients along
with stereo data, instead of delivering the interior and
exterior orientation parameters and other properties re-
lated to physical Sensor, one can proceed to satellite
photogram metric processes which approximate the sen-
sor model itself. A detailed study of the RFMs for pho-
togrammetric processing has been carried out by Tao and
Hu [3]. Di [4] demonstrated different ways to improve
the geo-positioning accuracy of Ikonos stereo imagery by
either refining the vendor provided RF (Rational Func-
tion) coefficients, or refining the RF derived ground co-
ordinates. Poon [5] focuses on Digital Surface Model (DSM)
generation from high resolution satellite imagery (HRSI)
using different commercial of the shelf (COTS) packages.
They validated the stereo DEM with InSAR DEM for
different land forms. Nadeem [6] valida ted DEM ge ne ra te d
from Cartosat-1 stereo data.
Crespi [7] evaluated the DSM by comparing the h e i ght s
of several buildings and points on the road axis derived
from a large scale (1:2000) 3D map. Fracer and Hanley
[8] demonstrated the wide applicability of bias compen-
sated RPCs for high accuracy geo-positioning f ro m st ere o
HRSI for a mountainous terrain. Chen [9] compared geo-
metrical performance between rigorous sensor model
(RSM) and RFM in the sensor modeling of FORMO-
SAT-2 satellite image. Dabrowski [10] evaluated DEMs
generated with different numbers of GCPs from Carto-
sat-1 stereo data at large number of evenly distributed
check points. Similar attempts to evaluate the accuracy of
the DEM using different number of GCPs have been
made by Michalis and Dowman [11] and Rao [12].
2. Study Area
The study area in Chamoli district lies in the northeastern
part of Uttarakhand state (Figure 1). It is bounded by
North Latitude 29º55’00” & 31º03’45” and East Longi-
tude 79º02’39” & 80º03’29” and falls in Survey of India
toposheet nos. 53 O, M and N. The geographical area of
the district is 7820 km2. Chamoli district the sec- ond
largest district of Uttarakhand, is also important from
strategic point of view as it shares its northern boundary
with Tibet (China). Geologically the area be- long to the
Lesser Himalayas and lies in a tectonic fore deep. The
Lesser Himalayas are comprised of fanglomerates fol-
lowed by bedded quartzites, slates, phyllites and low-
grade schists. The rock types are ranging from green
schist to lower amphibolite facies. The main rock types
are schists, phyllites and quartzites.
Agriculture is the main occupation of the people. The
agricultural activities are restricted to river terraces, gen-
tle hill slopes and intermontane valleys. The major crops
are rice, wheat, potato, pulses, millets and seasonal vege-
tables. Forest cover (58.38%) is the main landuse. Ala-
knanda river, Ramganga River and their tributaries drain
the district. Prominent of the tributaries are Dh auli ganga,
Birhi ganga, Nandakini, Pindar etc. The main drainage
patterns are dendritic, sub-dendritic, trellis, sub-rectan-
gular and rectangular. The major rivers are Alaknanda,
Dhauli ganga, Pindar are of antecedent type, where the
drainage in the structurally disturbed area of subsequent
type.
The climate varies from Sub-tropical monsoon type
(mild winter, hot summer) to tropical upland type (mild
winter, dry winter, short warm summer). The northern,
northwestern, northeastern and western part of the dis-
trict is perennially under snow cover, here the climate is
sub-arctic type as the area is represented by lofty Hima-
layan Range. Severe winter and comparatively higher
rainfall are the characteristic features of the northern
part.
Larger part of the district is situated on the southern
slopes of the outer Himalayas, monsoon currents can
penetrate through trenched valleys, the rainfall reaches
its maximal in the monsoon season that spans betweens
June to September. Rainfall, spatially, is highly variable
depending upon the altitude. In the Lesser Himalayan
Zone (1000-3000 m) maximum rainfall occurs about 70
to 80% in southern half. August is the rainiest month.
Rainfall rapidly decreases after September and it is the
least in November. About 55 to 65% rainfall occurs in
the northern half in Central Himalayan Zone. About 17%
of the annual precipitation occurs in winter season.
Chamoli district comprises of high hills and mountains
with very narrow valleys, deep gorges having very high
gradient. The northern, northwestern, eastern and north-
eastern part of the district comprises Tethyan Himalaya
with snow covered throughout the year. Physiographi-
cally the catchment of Alaknanda River comes under
Gangotri-Badrinath-Kedarnath Complex (i.e. Himadri,
Greater Himalaya zone) shows Radial Drainage pattern.
The soils are natural, dynamic, heterogeneous, non-
renewable resource, which support plant and animal life.
The tract of Chamoli district consists of outward success-
sion of ridges viz; Greater Himalaya and Lesser Himalaya
of decreasing height. These hills posse very little level
land. The soils have developed from rocks like granite,
schist, gneiss, phyllites, shales, slate etc. under cool and
moist climate. Very steep to steep hills and Glacio-fluvial
Copyright © 2010 SciRes. JGIS
V. K. SINGH ET AL.
Copyright © 2010 SciRes. JGIS
87
Figure 1. Satellite image showing study area.
4. Data Processing and Methodology
valleys are dominantly occupied with very shallow to
moderately shallow excessively drained, sandy-skeletal
to loamy-sk eletal, neu tral to sli ghtly acidic w ith low av ai l-
able water capacity soils. They have been classified as
Lithic/Typic Cryorthents. These soils are in general un-
der sparse vegetation. The Lesser Himalayan range is
mainly composed of highly compressed and altered rocks
like granite, phyllites, quartzite etc. and a major part of it
is under forest. Intermittent sparse patchy terraced culti-
vation is also practiced on fairly steep hill slop es whereas
dry and wet cultivation are prevalent on the uplands and
low-lying valleys respectively. The broader valley slopes
dominantly have deep, well drained, fine-loamy, moder-
ately acidic and slightly st ony.
A standard methodology has been adopted for the generat-
ion of DEM and ortho Image as is shown in Figure 2. It
comprises of reconnaissance survey and DGPS survey,
establishment of reference station by network adjustment
with IGS stations , establishment of a sub reference station
with respect to reference station, establishment of GCPs
with respect to sub-reference station, stereo data analysis
using RPCs and updation of RPCs using GCPs, generation
of DEM and Ortho image generation from DEM, accuracy
assessment of DEM and ortho image, generation of DEM
by refining Rational Polynomial Coefficients (RPCs) with
different number and distribution of GCPs, validating the
DEM, generating the DEM, Ortho images at the best
check point RMSE.
3. Data Used
The field data of base and rover stations were processed
in DGPS mode using Leica’s Ski-Pro. The field recordings
were first transferred to the system and the points were
assigned as reference and rover accordingly. Single point
rocessing (SPP) was first done for the base points and
For Digital Elevation Model and ortho image generation
from Cartosat-1 satellite data following data sets were
used. The details of the Cartosat-1 data are given in Ta-
ble 1. p
V. K. SINGH ET AL.
Copyright © 2010 SciRes. JGIS
88
Table 1. Details of cartosat stereo pair used for DEM generation and orthorectification.
S. No. Aft Scenes Fore Scenes
1. Satellite ID = CARTOSAT-1 Satellite ID = CARTOSAT-1
2. Date of Pass = 09NOV2005 Date of Pass = 09NOV2005
3. Sensor = PAN_AFT Sensor = PAN_FORE
4. Path = 0534 Path = 0534
5. Row = 0258 Row = 0258
6. Resolution along = 2.5 m Resolution along = 2.5 m
7. Resolution across = 2.5 m Resolution across = 2.5 m
8. No. of scans = 120 00 No. of scans = 12000
9. No. of Pixels = 12 000 No. of Pixels = 12000
Figure 2. Flow chart methodology showing the detail proc-
edure for DEM generation from Cartosat-1 stereo pair and
orthorectification.
then the rover points were processed with respect to the
base. The points which could not be accurately resolved
were post-processed to remove ambiguities. However,
only seventeen points could be determined precisely.
The fore and aft scenes of Cartosat-1 data were used to
generate the Digital Elevation Model in Leica’s Photo-
grammetry Suite with OrthoBase Pro. The scenes were
provided with Rational Functions Coefficients (RPC).
These coefficients are used to specify the geometric model,
which defines the internal characteristics (i.e. internal ge-
ometry of the camera or sensor while capturing the im-
agery) and external parameters (i.e. original position and
orientation of the cam era or sensor).
The reference coordinate system is assigned a project-
ion in UTM with spheroid and datum as WGS84. A
block file is created in LPS and the two scenes added to
the frame (Figure 3). The chipping coefficients are di-
rectly taken from the RPC text files provided with Car-
tosat-1 data.
Pyramid layers, based on a binomial interpolation alg-
orithm and a Gaussian filter, were generated to preserve
image contents and save computation time. Internal ori-
entation is done to define the pixel coordinate positions
of the calibrated fiducial marks within each image of the
block. External orientation is done to define the position
and orientation of the perspective centre. If very precise
values (i.e. less than a meter) of exterior orientation are
imported, the aerial triangulation process can even ass-
igned a horizontal c ontrol as they were l ocated besi de clif f.
5. Results and Discussions
It was observed that while using only RPC information
for Cartosat-1 stereo data, the error in height was in the
range 124 to 286m. However, after use of GPS points
and triangulation adjustment, the Cartosat DEM (Figure
4) becomes smooth and the error in height was reduced
to 3 to 18m (Figure 5). It was found that accuracy of
contours generated from Cartosat-1 stereo data was very
accurate and close to ground height. Accuracy of DEM
and ortho image was improved by triangulation iteration
convergence option with less RMSE as shown in Tab le 2.
The ortho image of Cartosat aft and fore image by DEM
is shown in Figure 6. This Cartosat-1 stereo data can be
used for height information generation at 4 m contour
interval. The DEM generated from Cartosat-1 stereo data
will be very much useful for topographic analysis in the
field of water recourses, Landslide study, agriculture etc.
V. K. SINGH ET AL.
Copyright © 2010 SciRes. JGIS
89
Figure 3. Generation of DEM and Ortho image generation in LPS.
Table 2. Details of triangulation summary showing RMSE error.
Triangulation Summary
Triangulation Iteration Convergence: Yes
Total Image Unit-Weight RMSE: 0.2721570
Control Point RMSE Check Point RMSE
Ground X 0.0711773 (10) Ground X 7.4913788 (3)
Ground Y 0.3628044 (10) Ground Y 7.3844452 (3)
Ground Z 0.0930104 (10) Ground Z 6.8060341 (3)
Image X 4.6107092 (20) Image X 0.4999596 (3)
Image Y 8.0061207 (20) Image Y 2.8900647 (3)
The use of coarse resolution multispectral data of
LISS-III (23.5 m) in the hilly terrain may not give accu-
rate estimate of bio resource assessment due to high relief
displacement. Therefore, the integration of the coarse
resolution satellite data that of high-resolution satellite
data (Cartosat PAN) after orthorectified helped in im-
proving the accuracy of bio-resource assessment (Figure
7). The quality of orthorectification depends upon the
quality of DEM. Therefore, high-resolution DEM needs
to be used wherever possible. It is suggested that the loss
of information in stretched areas could be supplemented
with gro un d truth.
Digital Elevation Model generated from Cartosat-1
Stereo data could be improved with using more accurate
and well-distributed GCP’s for refining the rational func-
tion coefficients. Millimeter accuracy GCP’s can be col-
lected while using Geodetic Dual Frequency GPS in
relative mode, which can improve accuracy of stereo
model. It had also been observed that there was hardly
any effect of small cloud covers present on the images,
V. K. SINGH ET AL.
90
Figure 4. Digital elevation model generated from Cartosat-
1 satellite stereo pair.
during automatic conjugate point matching. Generally
speaking, we can a ffirm that the Cartosat -1 DSM’s accura cy
decreases as the number of GCPs used decreases, with in-
creasing ground sampling distance and with increasing
terrain slope. Moreover, the use of high quality GCPs is
fundamental to obtain good DSMs, filtering may help to
enhance the elevation accuracy and the generation method
used is fundamental for determining the final quality of
products.
6. Conclusions
The reason for the accuracy difference obtained using
only one against four (or more) GCPs is due to the use of
RPCs for image orientation. Data processed without any
GCPs mainly sh ow linear s yste matic errors a nd few G C P s
can be used to improve the positioning accuracy by fit-
ting the RFM calculated coordinates to the coordi- nates
of the additional GCPs; with one GCP is possible only to
correct for shifts while using more GCPs an addi- tional
transformation in the image space can be applied.
More in general, Cartosat-1 stereo images have proven
to be an excellent source of data for the production of
DSMs with a ground resolution of about 10 m. Even if
within the range of available high resolution optical re-
mote sensing satellites there are several units with a higher
geometric resolution than Cartosat-1, Cartosat-1 DSMs
can nevertheless be compared to similar models produced
from higher resolution input imagery.
Figure 5. Graph showing comparison in height from GPS, Cartosat-1 derived DEM using GPS observation and Cartosat-1
derived DEM using RPC.
Copyright © 2010 SciRes. JGIS
V. K. SINGH ET AL.91
Figure 6. Ortho image generated (Fore & Aft) from Cartosat-1 satellite Stereo pair.
Figure 7. Liss-III + pan merged orthorectifed data showing
sharp ridge and valley pr ofile s.
7. Acknowledgements
Authors thankfully acknowledge the constant encoura-
gement and support received from Mrs. Shefali Aggar-
wal, Head PRSD, IIRS & Prof. R. C. Lakhera, Head
Geosciences Division, IIRS. Thanks are also due to Dr.
R. D. Garg, Department of Civil Engineering, IIT Roor-
kee for hi s v a luable suggestion and guidance.
8. References
[1] P. K. Srivastava, K. Gopala, B. Srinivasan, T. P. Amitabh,
S. Trivedi and R. Nandakumar, “Cartosat-1 Data Prod-
ucts for Topographic Mapping,” ISPRS Commission-IV
International Symposium on Geospatial Databases for
Sustainable Development, Vol. 37, 2006, pp. 1357-1362.
[2] R. Nandakumar, T. P. Srinivasan, K. B. Gopala and P. K.
Srivastava, “Data Products for Cartosat-1,” ISG Newslet-
ter, Vol. 11, No. 2-3, 2005, pp. 18-24.
[3] C. V. Tao and Y. Hu, “A Comprehensive Study of the
Rational Function Model for Photogrammetric Process-
ing,” Photogrammetric Engineering & Remote Sensing,
Vol. 67, No. 12, 2001, pp. 1347-1357.
[4] K. Di, R. Ma and R. Li, “Rational Functions and Potential
for Rigorous Sensor Model Recovery,” Photogrammetric
Engineering & Remote Sensing, Vol. 69, No. 1, 2003, pp.
33-41.
[5] J. Poon, C. S. Fracer and C. Zhang, “Digital Surface
Models from High Resolution Satellite Imagery,” Photo-
grammetric Engineering & Remote Sensing, Vol. 73, No.
11, 2007, pp. 1225-1232.
[6] A. Nadeem, M. Anjum, A. Ritesh, P. Jayaprasad, S. K.
Pathan, Ajai, D. K. Singh and A. K. Singh, “Extraction
and Validation of Cartosat-1 DEM,” Journal of the In-
dian Society of Remote Sensing, Vol. 35, No. 2, 2007, pp.
121-127.
[7] M. Crespi, F. Barbato, L. D. Vendictis, R. Onori, D. Polic,
F. Volpe and X. Wang, “Orientation, Ortho-Rectification,
DSM Extraction and 3D City Modeling by Cartosat-1
Stereo Imagery: First Results of a Test over Rome,” The
International Archives of the Photogrammetry, Remote
Sensing and Spatial Information Sciences, Goa, Vol. 36,
2006, pp. 1020-1025.
[8] C. S. Fracer and H. B. Hanley, “Bias Compensated RPCs
for Sensor Orientation of High Resolution Satellite Im-
agery,” Photogrammetric Engineering & Remote Sensing,
Vol. 71, No. 8, 2005, pp. 909-915.
Copyright © 2010 SciRes. JGIS
V. K. SINGH ET AL.
92
[9] L. C. Chen, T. A. Teo and C. L. Liu, “The Geometrical
Comparison of RSM and RFM for FORMOSAT-2 Satel-
lite Images,” Photogrammetric Engineering & Remote
Sensing, Vol. 72, No. 7, 2006, pp. 573-579.
[10] R. Dabrowski, M. Kedzierski, W. Fedorowicz, R. Kac-
zynski and J. Zych, “Geo-Interpretation Capabilities and
Precision of an Ortho-Photomap Obtained from Cartosat
Images,” The International Archives of the Photogram-
metry, Remote Sensing and Spatial Information Sciences,
Vol. 36, 2006, pp. 1038-1040.
[11] P. Michalis and I. Dowman, “Sensor Model Evaluation
and DEM Generation for Cartosat-1,” The International
Archives of the Photogrammetry, Remote Sensing and
Spatial Information Sciences, Goa, Vol. 36, Part IVB,
2006, pp. 1009-1013.
[12] B. S. Rao, A. S. R. K. V. Murali Mohan, K. Kalyanara-
man and K. Radhakrishnan, “Evaluation of Cartosat-1
Stereo Data of Rome,” The International Archives of the
Photogrammetry, Remote Sensing and Spatial Informa-
tion Scie nces , Goa, Vol. 36, Part 4B, 2006, pp. 1026-1029.
Copyright © 2010 SciRes. JGIS