The Use of Semi-Automated Method for Assessing the Horizontal Positional Accuracy of Google Earth imagery

Abstract

Google Earth imagery is frequently used in science, engineering, and other mapping applications. However, the company owning the tool announced that the data available in its geographical products is only approximate, so its accuracy is not officially documented. The Google Earth imagery in many areas around the world has been independently checked by scholars and third body parties. The estimated accuracies are found to largely vary depending on various factors but mainly due to, the imagery source or the image resolution. Positional accuracy testing methodology may also affect the assessment results. In processing, there should be many points around the tested area in order for the comparison to be more reliable. In this paper, the horizontal accuracy assessment was carried on the Google Earth imagery in Duhok city using the traces collected via GPS in Real Time Kinematic (RTK) technique. About 38 km of trajectory was collected for the two main roads in the selected area. Via semi-automated method, the points from RTK trajectory were compared to the corresponding extracted points from the centerline of the road network of Google Earth imagery. The nearest neighboring method through buildup algorithm was considered for comparison between both sets of data. Root Mean Square Error (RMSE) and maximum error were computed for horizontal positional coordinates and found to be 1.53m and 7.76m, respectively.