INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 2, No 3, 2012

INTERNATIONAL JOURNAL OF GEOMATICS AND GEOSCIENCES Volume 2, No 3, 2012 Copyright 2010 All rights reserved Integrated Publishing services Research article ISSN 0976 4380 Generation and evaluation of Cartosat -1 DEM for Chhota Shigri Glacier, Himalaya Centre of Studies in Resources Engineering, Indian Institute of Technology, Bombay Mumbai -400076, India pratimapandey@iitb.ac.in ABSTRACT A digital elevation model (DEM) is a simple representation of a surface in 3 dimensional way with height as the third dimension along with x and y in rectangular axes. DEM has wide applications in various areas like disaster management, hydrology and water management, geomorphology and in urban development. Valuable information about a terrain can be inferred by exploiting a DEM in proper way. Study of DEM becomes very useful for studying mountainous terrain such as Himalaya which is otherwise hard to access due to harsh weather and inaccessibility. Cartosat-1 or IRS P5 (Indian Remote Sensing Satellite) is a state-of-the-art remote sensing satellite built by ISRO which is mainly intended for cartographic applications. The satellite carries two panchromatic cameras which are capable of acquiring stereoscopic data along the orbital track. The high resolution stereo data have great potential to produce high-quality DEM. This paper discusses the generation of DEM from Cartosat -1 data for Chhota Shigri glacier (Himachal Pradesh, India). The DEM from Cartosat data was generated both using ground control points (GCPs) and without GCPs. The accuracy of both the DEM has been assessed. Keywords: Digital Elevation Model, Cartosat-1, Glacier, Chhota Shigri, Photogrammetry 1. Introduction A DEM is a representation of Earth surface with latitude, longitude and altitude i.e. X, Y horizontal coordinates and height Z. DEMs play an important tool for the analysis of glaciers and glaciated terrains (Bishop et al. 2001, Duncan et al. 1998, Etzelmüller & Sollid 1997, Paul et al. 2004, Sidjak & Wheate 1999). The launch of ISRO s Cartosat -1 satellite has opened a vast possibility for various areas like disaster management, hydrology and water management, geomorphology, urban development, map creation and resource management. Cartosat-1 is designed for cartography applications (Kocaman, 2008). The satellite is placed in the polar Sun Synchronous Orbit of 618 km from Earth. It has a payload consisting of two cameras - one near nadir looking aft (A) and the other forward looking fore (F) with a tilt of - 5 degree and +26 degree providing the real time stereo data along the track. These cameras are mounted with a fixed geometry which helps in collecting stereo coverage of the terrain at a fixed B/H ratio of 0.62. The swath covered by these high resolution PAN cameras is 30 km and their spatial resolution is 2.5 meters. A description of the Cartosat-1 mission is given in Krishnaswamy and Kalyanaraman (2002), Krishnaswamy (2002) (Kocaman et al., 2008). The stereo data from this satellite along with the Rational Polynomial Coefficients (RPC) can be used to generate DEM. The positional and height accuracies of Cartosat data products generated using RPC are in the range of 100m to 250m. The Rational Polynomial Coefficient (RPC) file contains the third degree polynomial coefficients that relate the image to the object space considering the imaging sensor geometry. These RPCs are sensor derived and terrain Submitted on January 2012 published on February 2012 704

independent (Rao et al.). Rational Polynomial satellite sensor models are simpler empirical mathematical models relating image space (line and column position) to latitude, longitude, and surface elevation. The name Rational Polynomial derives from the fact that the model is expressed as the ratio of two cubic polynomial expressions (Gopal Krishna et al, 2008). 2. Objectives This paper examines the generation of DEM from Cartosat -1 stereo data pair. To generate an accurate DEM from stereo data ground control points are required. In many practical applications the collection of GCPs are a major problem and for some of the region which are not accessible, it is very difficult to acquire GCPs manually. Himalayan region is one such area where the collection of GCPs is very difficult and expensive. In this study DEM has been generated using both GCPs and without GCPs and compared. 2.1 Study area The study was carried out on Chhota Shigri glacier (32 11'N - 32 17'N and 77 30 E - 77 32 E), Chandra valley, located in Lahaul-Spiti Valley, Himachal Pradesh. The Lahaul Spiti district comes under the Pir-Panjal range of Himalaya (figure 1). Chandra valley lies between an altitude ranges of 3300m to 6000m. The valley falls under the monsoon-arid transition zone and is climatically important as it is influenced by both Indian Summer Monsoon and Western disturbances. 2.2 Data used Figure 1: Location map of Chhota Shigri, Himachal Pradesh, India Cartosat-1 stereo data pair dated 28 th September 2008 were used to generate the DEM. 3. Methodology for DEM Generation Leica Photogrammetry suite (LPS) version 9.3 Software package was used to generate DEM from Cartosat-1 stereo data pairs. The software can be used to generate DEM, orthorectified image, editing of generated DEM, mosaic and image calibration. The software supports reading of data, manual or automatic GCP/tie points (TP) collection and geometric modeling of different satellites including RPC model and zero order, a first order and second order RPC polynomial adjustments (Gopal Krishna et al, 2008). The RPC method based on the block adjustment method developed by Grodecki and Dial was used for DEM generation. A block project file has to be created inside the software for the DEM generation defining the 705

geometric model as RPC model. The block project has assigned the horizontal and vertical coordinates with UTM projection and WGS 84 datum. The stereo pair images band a and band f were added to the frame. The interior and exterior orientations corresponding to the RPC files were carried out in frame editor. Interior orientation defines the internal geometry of a sensor, as it existed at the time of image capture and exterior orientation is the position and angular orientation of the sensor that captured the image. The software extracts sensor information from RPC file and carries out the interior and exterior orientation. The pyramid layers were computed and updated. The generation of DEM from stereo data needs geometric model and GCPs. A DEM prepared using GCPs is known as absolute DEM and that without using GCPs is known as relative DEM. In the absolute DEM the horizontal and vertical references systems are tied to geodetic coordinates whereas a relative DEM is a DEM with relative differences in position, scale, and rotation from the geodetic coordinates on the ground (horizontal reference system) and the mean sea level (vertical reference system) (from ENVI tutorial). The DEM was generated both using GCPs and without GCPs. The flow chart of complete method is shown in figure 2. 3.1 DEM generation with GCPs Figure 2: Flow chart of DEM generation methodology LPS 9.3 was used for DEM generation and orthorectified image generation from Cartosat-1 stereo data. The stereo data Band a and Band f overlap is around 90%. Since the study area is located in the rugged part of Himalaya, collection of GCPs around the area is not easily possible pertaining to its harsh weather and inaccessibility. The region is highly terraneous and difficult to reach. So well distributed control point collection was not possible. Ground control points were collected near and around the snout of the Chhota Shigri glacier. The points taken were from GPS having a vertical accuracy of approx 8m (as provided by the instrument) and having horizontal accuracy in the centimeter range. 5 ground control points were used in the refinement of RPC model and DEM generation and 2 points were used as check point to check the accuracy of the DEM generated. 706

The classical point measurement tool was used to add the GCPs and to generate tie points. Ground points were added to the images. Tie points were generated both manually and automatically for even distribution of them. Tie points are points which can be identified in the overlap area of stereo images and whose ground coordinates are not known. A total 59 tie points were generated and there location accuracy was checked. The tie points and GSPs distribution in Band a and Band f images are shown in figure 3. Figure 3: The distribution of tie point, control points and check points in Band a and Band b The triangulation was run after adding GCPs and tie points. Triangulation is used to improve the accuracy and refinement of RFM. Triangulation process established relation between images, sensor model and ground points (Krishnamurthy, 2008). The process was run to check the accuracy for GCPs and tie points. The triangulation error report for GCPs is shown in figure 4. Figure 4: Triangulation error report After running the triangulation the DSM was extracted with cell size of 10m. GCPs and tie points were used as seed vertices for DEM creation. This input enhances the relative position of the DEM (Krishnamurthy et al, 2008). The quality of extracted DEM based on the software is given in table below (table 1). 707

Table 1: quality of the extracted DEM Excellent % (1-0.85) 75.9704 % Good % (0.85-0.70) 23.0898 % Fair % (0.70-0.5) 0.9397 % Isolated % 0.9397 % Suspicious % 0.0000 % AFT (Band a) image was used for Orthorectification image because of its near nadir acquisition angle. Figure 5 shows the DEM and orthorectified image generated. Figure 5: Cartosat-1 generated DEM at 10m pixel size and the orthorectified image generated from Cartosat-1 3.2 DEM generation without GCPs DEM generated without using GCPs is known as the relative DEM. Practically it is not always possible to collect ground control points especially in the difficult and hilly terrain of Himalaya. Though the accuracy of the relative DEM extracted without GCPs are poor as it depends on the accuracy of satellite, however if the relative DEM produced has good relative accuracy and represents relative terrain relief well, it is very useful for applications in geology, geomorphology (R. Tateishi and A. Akutsu, 1992). The generation of relative DEM in LPS 9.3 follows the same procedure as shown in the flow chart (figure 3) except addition of GCPs. Tie points were generated both manually and automatically for even distribution of the points and triangulation was run with zero polynomial order for refinement of model. The triangulation result is shown in the figure 6. Figure 6: Triangulation error report 708

The software based quality of the extracted DEM is given in table 2. Table 1: quality of the extracted DEM Excellent % (1-0.85) 75.7173 % Good % (0.85-0.70) 23.3510 % Fair % (0.70-0.5) 0.9317 % Isolated % 0.0000 % Suspicious % 0.0000 % The DEM and the orthorectified images generated without using GCPs are shown in figure 7. Figure 7: Cartosat-1 generated DEM at 10m pixel size and the orthorectified image generated from Cartosat-1 without GCPs 4. Evaluation of DEM For absolute DEM, the rational function model was used to correct the Cartosat-1 image GCPs. 2 check points were used to verify the accuracy. The GCPs were acquired using a high performance GPS with about 50 cm horizontal accuracy and 8 m vertical accuracy. The GCPs were acquired with UTM projection and WGS 84 datum. A total 5 GCPs were used as control points to refine the model and 2 were used as check points for positional accuracy. The overall RMSE of the DSM was 12.4646 with absolute linear error (Absolute Linear Error 90) of 17.5015. The overall ground control points RMSE are 28.76 with Absolute Linear Error 90 of 42.0349. The ground RMSE of check points are 18.6480 (Mean Absolute Error: 14.9530, Absolute Linear Error 90: 26.0955). The vertical accuracy of DSM generated from Cartosat-1 stereo data is within 20-30 m with the use of GCPs. The evaluation of relative DSM was done with respect to the absolute DSM. The elevation profile of the relative DSM was compared with the absolute DSM. The profile was drawn both at high and low elevations and mean was calculated. From the evaluation it was found that the relative Cartosat-1 DSM overestimates the absolute DSM by a height of around 95m (figure 8). It is in agreement with the fact that the DSM generated from Cartosat-1stereo pairs without GCPs have vertical accuracy between 100-250m. 709

Figure 8: Profile of elevation comparison between DEM generated with GCPs and without GCPs 5. Conclusion The aim of the paper is to generate the DEM with stereo data for Himalayan terrain where the undulation is irregular. The DEM of Himalaya can be used to study the change in the elevation of glaciers which are again due to change in climate or due to tectonic activities. To study the elevation of Himalaya, DEM is the best source as direct method is very difficult due to harsh weather and rugged terrain of Himalaya. The DEM of the glacier can also be used to study the mass balance change in the glaciers and melt run off. This study attempts to generate a DEM of Himalaya using Cartosat -1 dataset which can be used for elevation change study and hence for mass balance change in the glacier. The DEM was prepared both using GCPs and without GCPs. Once, a DEM of Himalayan terrain is available, it can be used for various applications like landslide and hazards study, for tectonic activities study and for climatological studies. Acknowledgment Authors express their thanks to Dr. Sh. Ashwagosha Ganju, Director, SASE and Dr. Snehamani, Deputy Director, SASE, for providing support to carry out the field work. Authors are deeply thankful to Mr. G.G. Ponnurangam, CSRE, IIT Bombay for all his help. 6. References 1. Bishop, M.P., Bonk, R., Kamp, U. and Shroder, J.F., (2001), Topographic analysis and modeling for alpine glacier mapping. Polar Geography, 25, pp 182-201. 2. Duncan, C.C., Klein, A.J., Masek, J.G. and Isacks, B.L., (1998), Comparison of Late Pleistocene and modern glacier extents in central Nepal based on digital elevation data and satellite imagery. Quaternary Research, 49, pp 241-254. 3. Etzelmüller, B., and Sollid, J.L., (1997), Glacier geomorphometry an approach for analysing long-term glacier surface changes using grid-based digital elevation models. Annals of Glaciology, 24, pp 135-141. 4. Gopala Krishna, B., Amitabh, Srinivasan, T..P, Srivastava, P. K., (2008), Dem generation from high resolution multi-view data product. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B1. Beijing 2008, pp 1099-1102. 710

5. Kocaman, S., Wolff, K., Gruen, A., Baltsavias, E., (2008), Geometric validation of Cartosat-1 imagery.., proc. 21 st ISPRS Comgress, 3-11 July, 2008 Beijing. In: International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 37, Part B1-3, pp 1363-1368. 6. Krishnaswamy, M., (2002), Sensors and Platforms for High Resolution Imaging for Large Scale Mapping Applications - Indian Scenario. Indian Cartographer, DAPI-01, URL: http://www.incaindia.org/technicalpapers/02_dapi01.pdf (accssed 29 January, 2008). 7. Krishnaswamy, M., Kalyanaraman, S. (2002). Indian Remote Sensing Satellite Cartosat-1: Technical features and data products. Paper presented at Map Asia 2002 conference, 7-9 August, Bangkok, Thailand. Available at http://www.gisdevelopment.net/technology/rs/techrs023.htm (accessed 29 January, 2008). 8. Paul, F., Huggel, C. and Kääb, A., (2004), Combining satellite multispectral image data and a digital elevation model for mapping debris-covered glaciers. Remote Sensing of Environment. 89(4), pp 510-518. 9. Rao, C.V., Sathyanarayana,P., Jain, D.S. and Manjunath A.S., Topographic map updation using cartosat-1 data. 10. Sidjak, R.W., and Wheate, R.D., (1999), Glacier mapping of the Illecillewaet icefield, British Columbia, Canada, using Landsat TM and digital elevation data. International Journal of Remote Sensing, 20, pp 273-284. 11. Tateishia, R. and Akutsua, A., (1992), Relative DEM production from SPOT data without GCP, International Journal of Remote Sensing, 13, pp 2517-2530. 711