Validation of Google Earth-based DEM using leveling data

Abstract

We generate a digital elevation model (DEM) from online open-source Google Earth imagery. The DEM generation steps from Google Earth are based on digitizing the track paths over the selected area. Heights are then updated over the digitized paths and interpolated to create the final DEM quality of the extracted DEM is assessed utilizing vertical accuracy. We use local leveling data to validate the newly-extracted DEM. The comparison between the leveling and DEM-based heights shows an offset of 2.8 m due to the systematic errors, the rootmean-square error (RMSE) of the differences is estimated to ± 5.47 m. We employ a 4-parameter fitting model to eliminate the systematic errors, which improves the RMSE to ± 4.63 m after removing the errors. The fitting model is also used to reform and refine the error distribution between the data sets of comparisons. Numerical investigations and analyses are included and illustrated.