GISdevelopment.net ---> AARS ---> ACRS 1995 ---> Poster Session 1

Differences of Sar Image Between JERS-L and ERS-L To Land Use/Land Cover Investigation

Anusom Chantanaroj, Kenji Otsuka, Keisuka Katsuta


Abstract
Since SAR data had been developed to use in remote sensing technology. Areas in South- West of Bangkok were selected to study the differences between JERS-1 and ERS-1 SAR images for land use/land cover application. It was observed that parameters influencing the backscatters were orientation.of the object, incident angle and object height. Building oriented parallel to flight direction backscatter were very strong while object height which less than around 30 cm. would be very low. Angle of roof pitch was also one of the parameters influencing the RADAR backscatters.

Introduction
The joint research between the Geographical Survey Institute of Japan ( GSI) and Department of Land Development of Thailand (DLD), funded by the Science and Technology Agency (STA) has been set up since 1992 to conduct researches on the development of remote - sensing technology, using satellite data, particularly using ERS-1 and JERS-1 data in order to contribute environmental monitoring in tropical area. In 1993 project areas were carried out in Kanto Plain, Japan and Bangkok, Thailand to develop Manual of SAR Image Interpretation. In ; 1995, SAR images from JERS-1 and ERS-1 of the South-west of Bangkok were selected to study the differences between the two image data for land use/land cover application. Field verification was carried out in February 1995 and data analysis was done in GSI, Japan.

Outlines Of The Procedure
  • Collecting data
  • Specify area for field investigation
  • Field survey and get some information in the area
  • Discussion and report
Methods Of Image Processing
  • Gray scale conversion from 16 bit to 8 bit data
  • Filtering
    • low-pass for noise removing
    • high-pass for image sharpening
Results
JERS-l SAR image acquired on 25 June 1992, path-row 127-277 and 127-278 resampling pitch 12.5 m, altitude 568 km, beam direction 35°, frequency 1275 MHz (L band) and ERS-1 SAR image acquired on 10 August 1993, path-row 10684-3321 and 10684-3339, resampling pitch 12.5 m, altitude 785 km, beam direction 23°, frequency 5.3 Ghz (C band), polarization VV, spatial resolution 30 m, swath 100 km were visually analyzed. Topographic maps on a scale of 1:50,000 and OPS image (IB 30 2R) of JERS-l of the same area were used in ground truth. From visual interpretation of SAR image and field observation, it was found that parameters which mainly influence differences of backscatters in SAR image from JERS-1 and ERS-1 are orientation of objects, vegetation height or object height and incidence angle.

Orientation of object is directly related to look direction. When the orientation of object is perpendicular to look direction, backscatters will be very strong. On the other hand backscatter will be very low when orientation of object is parallel to look direction.

Vegetation height or object height is one of the parameter which cause difference in backscatter. When object height is more than 30 cm backscatter in SAR image from ERS-1 will be strong while SAR image from JERS-1 will be very low backscatter. So weed and grass with 30 cm height can be observed by ERS-l better than JERS-l. Incidence angle has an effect in urban and built up land. In Thailand most of the angle of roof pitch are approximately 20 o , backscatter to ERS-1 is stronger than JERS-1.

Conclusion and Recommendation
There are tree main parameters influencing the backscatters from land use/land cover, they are orientation of object, vegetation height or object height and incidence angle. Urban area identification is easier ERS-1 SAR than JERS -1 image but some area it will be more clear from .JERS-1 SAR image if object orientation is perpendicular to look direction. In agriculture land, backscatter will be stronger in ERS-1 image than JERS-1 SAR image, especially when vegetation height is less than 30 cm.

To use only SAR data to identify land use type, the result is not satisfied. To get best result, it is recommended to use SAR data together with other data from passive remote sensing system and also Geographic Information System.

Acknowledgments
This project was funded by the Science and Technology Agency (ST A). This support was greatly appreciated.

References
  • Shunji Murai; Remote Sensing Note, JERS -1 1993
  • Rob Schuman; Microwave Remote Sensing, ESA, 1994.
  • B.N. Koopmans; Side looking Radar, ITC, 1991
  • Valairat Wanpiyarat; Synthetic Aperture Radar Image Interpretation annual Development Thailand Area GSI 1994 .
  • Wirat Thongs; Land use analysis by using remote sensing data, GSI 1993.