Optimizing the Global Digital Elevation Models (GDEMs) and accuracy of derived DEMs from GPS points for Iraq's mountainous areas

Abstract

Optimizing the combined horizontal and vertical accuracy of the well-known Global Digital Elevation Models (GDEMs) of various resolutions for each country and region especially in Iraq's mountainous areas is still questionable. All the three GDEMs, approximately, have the same vertical accuracy with the Root Mean Square (RMSE) of ±7.3 m, ±7.6 m and ±6.5 m via 12 fixed Ground Control Points (GCPs) for the Advanced Land Observation Satellite Phased Array L-band Synthetic Aperture Radar (ALOS PALSAR 12.5 m), the Shuttle Radar Topographic Mission (SRTM 30 m) and the TerraSAR-X (the name of twin satellites) add-on for Digital Elevation Measurement (TanDEM-X 90 m) GDEMs respectively. Moreover, the percentage of outliers that are greater or smaller than ±10 m detection of the height extraction from both the ALOS PALSAR and SRTM Digital Elevation Models (DEMs) contains 16.7% and for TanDEM-X was 25%. In this paper, the special DEM is derived using 2123 handheld GPS points for Sulaymaniyah Governorate, Kurdistan region, Iraq. The height extraction by discarding the outliers of 58% gives the RMSE of ±8.0 m in the case of adding geoid heights (N) to the ellipsoidal heights (h) via the Earth Gravitational Model 2008 (EGM2008) and ±5.6 m without adding N. It is expected that the derived DEMs will give more accurate results both horizontally and vertically in the mountainous areas when GPS observations are intensified. The horizontal accuracy is validated through extracting hierarchy stream types of the watershed map from the DEMs for higher than 100 pixels length. The ALOS PALSAR DEM extracted more numbers of stream orders than others. Finally, based on the criteria of RMSE, outlier detection, and the number of extracted stream orders, the ALOS PALSER DEM is regarded as the optimal GDEM in comparison with the close accuracy of both the TanDEM-X and SRTM DEM.