This study investigated the accuracy of DEMs generated from KOMPSAT-5 SAR (Synthetic Aperture Radar) stereo imagery using three types of CPs (Control Points) for RPC (Rational Polynomial Coefficient) correction. The three CP types were: 1) DCPs (Digital-map Control Points) derived from digital maps, using their ground coordinates and corresponding image coordinates from the SAR imagery; 2) PCPs (Public Control Points), using their ground coordinates and image coordinates derived from an RD (Range-Doppler) model; and 3) ICESat-2 points, using their ground coordinates and image coordinates derived from an RD model.  Experimental results demonstrated that the highest accuracy was achieved using PCPs, with an RMSE of approximately 11m(flat terrain) to 15m(flat and mountainous terrain). The RMSE values for the ICESat-2 points and DCPs were 11m(flat terrain)  to  23m(f lat  and  mountainous  terrain)  and  15m  (f lat  terrain)  to  17m  (f lat  and mountainous  terrain),  respectively. DEM accuracy is anticipated to be further enhanced using stereo imagery acquired with minimal time differences and convergence angles under 30 degrees. ICESat-2 data is considered highly efficient in areas with limited CPs availability.



Keywords : KOMPSAT-5 SAR Stereo Imagery, Rational Polynomial Coefficient, Range-Doppler, Digital-map Control Points, Public Control Points, ICESat-2 points