Google Earth's derived digital elevation model: a comparative assessment with Aster and SRTM data

Abstract

This paper presents a statistical analysis showing additional evidence that Digital Elevation Model (DEM) derived from Google Earth is commendable and has a good correlation with ASTER (Advanced Space-borne Thermal Emission and Reflection Radiometer) and SRTM (Shuttle Radar Topography Mission) elevation data. The accuracy of DEM elevation points from Google Earth was compared against that of DEMs from ASTER and SRTM for flat, hilly and mountainous sections of a pre-selected rural watershed. For each section, a total of 5,000 DEM elevation points were extracted as samples from each type of DEM data. The DEM data from Google Earth and SRTM for flat and hilly sections are strongly correlated with the R2 of 0.791 and 0.891 respectively. Even stronger correlation is shown for the mountainous section where the R2 values between Google Earth's DEM and ASTER's and between Google Earth's DEM and SRTM's DEMs are respectively 0.917 and 0.865. Further accuracy testing was carried out by utilising the DEM dataset to delineate Muar River's watershed boundary using ArcSWAT2009, a hydrological modelling software. The result shows that the percentage differences of the watershed size delineated from Google Earth's DEM compared to those derived from Department of Irrigation and Drainage's data (using 20m-contour topographic map), ASTER and SRTM data are 9.6%, 10.6%, and 7.6% respectively. It is therefore justified to conclude that the DEM derived from Google Earth is relatively as acceptable as DEMs from other sources.