Cartosat –1 (IRS –P5) Stereo Data Processing – A Case Study of Dehradun Area


Anil Kumar
Indian Institute of Remote Sensing
Dehradun, India
anil@iirs.gov.in

ABSTRACT:
Generation of Digital Elevation Models (DEMs) from satellite stereo data had been an important field from last few decades, started with the launch of the first civilian remote sensing satellite. Digital Elevation Models (DEMs) are a type of raster GIS layer. Raster GIS represents the world as a regular arrangement of locations. In a DEM, each cell has a value corresponding to its elevation. There is always a question mark how quickly and accurately the DEM’s can be generated especially using satellite stereo data. In last two decades people had tried using across track satellite stereo data for generation of DEM’s using Satellite Photogrammetric approach. Some satellites are capable for acquiring stereo data using across track approach e.g., SPOT, IRS –1C/1D etc, and some can acquire stereo data using across as well as along track, from two different orbits e.g., IKONOS, QUIKBIRD etc. In both the cases two scenes of stereo pair were acquired from two different orbits, due to which there is always problems of time difference between two scenes of one stereo pair. Which can present difficulties while transferring Ground Control Points (GCP’s) in stereo model and during automatic image matching for DEM generation.

To over come the time difference problem between two stereo scenes the vibrant Indian remote sensing satellite programme had launched advanced satellite, known as IRS-P5 (Cartosat-1) on May 5, 2005 aims to provide data with higher resolution for cartographic purpose. This satellite has two identical panchromatic sensors kept on the platform with fore, +26o and aft, -5o tilt respectively. This system is capable to provide stereo data without any time difference between two scenes. This paper presents one case study for processing stereo data acquired from Cartosat-1 satellite, of Dehradun area, for generation of DEM as well as Ortho-Image. A small evaluation study was also done, and it was observed that DEM generated using stereo data from Cartosat-1 satellite was having accuracy 2 to 12 m.

Introduction:
In the area of Satellite based remote sensing in the past, the first generation satellite IRS-1A and 1B were designed, developed and launched successfully during 1988 and 1991 with multi-spectral cameras with spatial resolution of 72.5 m and 36 m. respectively. Subsequently, the second genera-tion remote sensing satellites IRS-1C and 1D with improved spatial resolutions of 70 m in multi-spectral and 5.8 m. in Panchromatic bands and a wide field sensor with 188m resolution and 800 Km. swath, have been developed and successfully launched in 1995 and 1997 respectively. These satellites have become the principal components in the National Natural Resource Management System and the data was used in various applications, viz., agriculture and soil, land form and land use studies, water resource, forestry, drought and flood monitoring, cartography, town planning and coastal zone monitoring. Especially IRS-1C/D data has been used for cartographic and town-planning application up to 1:10,000 scale. These satellites also provide stereo pairs of imageries to get height information to an accuracy of approximately 10 meters.

With the above scenario, India has a lead in the civilian remote sensing field in the world not only in terms of realization and launching of complex satellites with high, medium and coarse resolution cameras, but also in the application areas as well. In order to maintain this lead and also provide continuity of data to global users, Cartosat-1 with two improved fore and aft PAN cameras with better than 2.5 m. spatial resolution was launched on May 5, 2005. This work briefly presents the output generated from Cartosat –1 data of Dehradun and its surroundings.

Data used:
For evaluation of Cartosat –1 data, data of Dehradun district covering west part of it was used. Fore and Aft scenes of west part of Dehradun town, dated 2nd October 2005 was used. The detailed information of fore and aft scenes are mentioned in table 1.

Table 1: Information of Fore and Aft scenes


In this paper the outputs (DEM, Contours and Ortho Image) generated from Cartosat –1 data were compared with outputs (DEM, Contours and Ortho Image) generated using IRS – 1C stereo data. The details of the IRS – 1C stereo data used in this work are mentioned in table 2.

Table 2: Parameters and Specifications of IRS-1C PAN stereo data


Data Processing:
The Fore and Aft scenes of Cartosat –1 data were provided with Rational functions (RFs) coefficients. Rational Functions (RFs) have been applied in photogrammetry and remote sensing to represent the transformation between the image space and object space whenever the rigorous model is made unavailable intentionally or unintentionally. It attracts more attention now because high-resolution images are being released to users with only RF coefficients.

For refining the orbital parameters of Fore and Aft scenes of Cartosat-1, few GCP’s (Ground Control Points), 11 in number were used for evaluating the cartosat-1 data of Dehradun area. Ground Control Points for this area were collected using Geodetic Single Frequency GPS in Relative mode. The Accuracy of GCP’s acquired from Geodetic Single Frequency GPS in Relative mode are shown in table 3. While generating the stereo model, eight ground points were used as GCP’s and other three known points on the ground were used as Tie Points as shown in following figure 1;

Table 3: Accuracy of GCP’s acquired from Geodetic Single Frequency GPS in Relative mode



Figure 1: Ground Control Points and tie points’ location on the stereo model


The exterior orientation parameters were refined while using eight GCP’s through triangulation. After triangulation Digital Elevation Model (DEM) was generated. Using the generated DEM Ortho image was generated from Aft PAN image.

For evaluation of the DEM generated using Cartosat-1 stereo data of Dehradun area, known points on the ground, used as GCP’s as well as Tie points used in the stereo model were studied. The comparison of the ground height generated by DEM as well as ground height information collected by Geodetic Single Frequency GPS in Relative mode is shown in table 4.

Table 4: For evaluating height from DEM generated from Cartosat-1 Stereo data.


Digital Elevation Model and Ortho Image generated from Cartosat-1 stereo data of Dehradun area shown in figure 2 and 3.


Figure 2: Digital Elevation Model generated from Cartosat-1 stereo data of Dehradun area



Figure 3: Ortho Image generated from Cartosat-1 stereo data of Dehradun area


Comparison Study:
A comparative study of outputs generated using Cartosat –1 stereo data of Dehradun area has been done in this section with IRS – 1C stereo data of same area. From stereo data of Cartosat –1 as well as IRS – 1C satellite three types of outputs were generated, e.g., DEM, Contours and Ortho Image. First a visual comparison for all the three outputs (DEM, Contours and Ortho Image) was done, using small common area lying on these stereo data sets. For this analysis, contours of 4 m (contour interval = 3*?, were ? is error in height) were generated using Cartosat –1 stereo data. Contour of 4 m interval as well as 10 m (contour interval = 3*?, were ? is error in height) interval was generated using IRS – 1C stereo data. It was found that DEM as well as contours generated using Cartosat –1 stereo data, were very much close to the ground information as shown in figure 4, 5, 6. Figure 7, 8 and 9 were used for evaluating the contours generated from both the data sets.


Figure 4: Contours generated from Cartosat-1 stereo data of Dehradun area at 4 m interval



Figure 5: Contours generated from IRS – 1C stereo data of Dehradun area at 10 m interval



Figure 6: Contours generated from IRS – 1C stereo data of Dehradun area at 4 m interval



Figure 7: Ortho Image of contour area from Cartosat-1 stereo data of Dehradun area



Figure 8: DEM of contour area from Cartosat-1 stereo data of Dehradun area


Conclusions:
It was observed that while using only RPC information for Cartosat –1 stereo data, the error in height was in the range 100 to 200m (table 4). While using eight GCP’s for Cartosat –1 stereo data, the error was reduced to 2 to 13m (table 4). It was found that accuracy of contours generated from Cartosat –1 stereo data were very accurate and close to ground height. 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, agriculture etc. In the case of IRS –1C stereo data of the same area, there were lot of conjugate points hanging in the air, and giving the spicks impression on the DEM. These spicks impression was not observed on the DEM generated using Cartosat –1 stereo data. This concludes that DEM generated from Cartosat –1 stereo data, through automatic mode was very smooth (no spicks impression), and no manual editing is required.

It had been observed that accuracy of 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 function coefficients. Millimeter accuracy GCP’s can be collected 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, during automatic conjugate point matching.


Figure 9: DEM of contour area from IRS –1C stereo data of Dehradun area


Reference:

  1. Edward M., James B., McGlone J. C., Introduction to Modern Photogrammetry, Edition 2001, Jhon Wily & Sons (USA).
  2. Egels, Y., Kasser M., Digital Photogrammetry, Edition: 1, Published By: CRC Press.
  3. James R. S., Introduction to Geodesy: The History and Concepts of Modern Geodesy, Edition1997, Jhon Wily & Sons (USA).
  4. Mitsuharu T., Seiichi H., Shigetoshi I., Shunji M., 1994, DEM Generation using JERSI OPS Stereo Data, Asian Conference on Remote Sensing, November 17 - 23, Bangalore, India.
  5. Paul R Wolf, Bon A. DeWitt, and Elements of Photogrammetry with Applications in GIS, Edition 2000, Publisher: McGraw-Hill.
  6. Wilfried L., Digital Photogrammetry: Theory and Application, Edition 2003, Publisher: Springer.