The purpose of this paper is to introduce a methodology for DEM generation from 1/100,000
topographiccontoursbased on multiple polynomial interpolations. Sinai Peninsula, in Egypt, was
chosen as a study area for its size and topographic diversity. The selected test area stretches north
south from gently-slope flat land in northern part to rugged mountainous region in Sothern part.
The proposed modeling scheme comprises five steps. In the first step, contours for Sinai
Peninsulawere manually digitized from 1:100,000 scale topographic maps. The obtained data is
about 940,624 points. As well, SRTM DEMs were downloaded and processed to be used for
comparison. In the second step, theSRTM DEMs were classified into three classes: flat; rolling;
and mountainous depicts slopes of less than 5%, from 5 to12% and greater than 12%
respectively. In the third step, three DEMs with a 30m horizontal resolution were derived from
the digitized contours using linear; quadratic; and cubic polynomial equations in a least-square
fashion.In the fourth step, flat, rolling and mountainous areas were extracted from linear,
quadratic and cubic-based DEMs respectively.A refined DEM was constructed by combining the
three extracted areas in one scene. Finally, the reconstructed DEMwere compared withvariety of
existing interpolation techniques. The comparison is based on four criteria: RMSE; systematic
errors; derived slope; and computational cost. Statistics show that for the reconstructed DEM, the
RMSE was 2.38m with no systematic errors. The derived slope and computational cost were
comparable to the well known interpolation technique, Kriging, with simpler implementation. |
