THREE DIMENSIONAL MODELING OF GEOLOGICAL MAPS WITH CARTOSAT1 (IRIS P5) SATELLITE IMAGES WITHOUT GROUND CONTROL POINTS

Several geological applications require slope and orientation parameters (dip and strike) of the geological features such as faults and strata for the 3-dimensional structural analysis. Conventional methods for the extraction of these parameters utilize the field observation and measurement which are cumbersome and time consuming. More recent methods employ satellite imageries to extract the dip and strike parameters. However, there is a main disadvantage associated with using the satellite images for this purpose and this is the low accuracy level achievable from small scale satellite images. Moreover, geometric correction of these images still requires considerable amount of field works to be carried out to measure the coordinates of several ground control points. The recent improvements in the high resolution satellite image technology, has provided a possibility of achieving higher level of accuracy for 3-dimensional ground surface modeling using GPS/INS/Star-tracker installed in the satellite. These equipments are used to generate precise attitude and altitude information of the sensor on-board the satellite. These information are converted to the so-called Rational Polynomials Coefficients (RPC,s) which may be utilized to extract precise 3-dimensional ground surface information without using ground control points. In this paper the stereo along track high resolution IRS P5 satellite images are used over a mountainous terrain and the so-called Rational Function Mathematical model is implemented to extract a precise 3-dimensioal digital surface data using the RPC information. Dip and strike parameters are then determined by surface fitting to the extracted 3-dimensional ground data followed by computing the surface normal vectors. The generated dip and strike angles which determined over several surface patches are compared with the dip and strike parameters determined from a large scale digital terrain model of the test area. The comparison indicates an overall consistency of better than 1 degree which is quite remarkable given the fact that no ground control point measurement and field observation is carried out. The results are also compared with the geological maps of the area with an overall discrepancy of 3 degrees. The larger discrepancy as compared with the former result is due the lower accuracy of the geological map.