| Abstract |
Representing the earth surface topography is necessary in many applications and uses. Long time and great efforts are needed to make this by the traditional ground instruments. The revolution of satellite mapping enabled obtaining Digital Elevation Models (DEMs) to wide areas of the earth surface. So, the representation of the earth surface became easy and saves time and effort. Unfortunately, the biggest problem of this way is the low accuracy of these DEMs in many places according to the nature of the earth surface and the absence of the data in other places (voids). So, the need to evaluate and enhance the performance and accuracy of these DEMs become necessary.
In this study, the accuracy of two global digital elevation models: Shuttle Radar Topography Mission (SRTM1) and ALOS-PALSAR (ALOS Phased Array type L-band Synthetic Aperture Radar) are evaluated by using ground orthometric and GPS heights. Data in two different test areas are used: A grid of 239 fixed points cover an area of 18.85 by 12.15 km in Toshka south of Egypt, about 55,000 Feddans and a grid of 2722 points in the southwest of Egypt cover an area of 210 by 120 km, about 6 million Feddans.
The evaluation process was made by comparing the ellipsoidal heights of the two sites points with their corresponding values in the used two DEMs and also comparing the orthometric heights of the first site points with their corresponding values in two DEMs.
Then enhancement process made by through four steps. The first step is converting the ellipsoidal heights of the used DEM to orthometric values by using (SGG-UGM 2) global geoid model instead of the used EGM96 to show the effect of undulation values on the DEMs values. This process was applied to the two DEMs in the first site (Toshka south of Egypt). The second step is shifting the model heights using one point in the middle of the area, where the value of this shift is the difference between the ground value and its corresponding value on the used DEM. This shift process is applied on the orthometric height values (in the first site) and once more on the ellipsoidal height values (in both areas). The third step is shifting using well distributed five points while, every point served an area with 4km radius, this method was made in two sites included in test area 2. The fourth step is applying first and second order polynomials by using the well-distributed five and seven control points respectively (in both test areas). Again, the shift process is applied on the orthometric height values (in the first site) and once more on the ellipsoidal height values (in both sites) using the well distributed five points. The obtained results showed that in most study cases SRTM1and ALOS-PALSAR both gave reasonable results for the geodetic heights and it is better to deal with the ellipsoidal heights of the GDEMs and they showed also that shifting process using one point is the best improving method among the other methods applied in this thesis.
|
