Investigating Amount of Sample Points Necessary for Accurate Topographic Representation of the Ground Truth
Wed, 04/29/2015 - 18:18
Abstract: Topographic or elevation data has many uses and applications especially when it is converted into a Digital Elevation Model (DEM). Such uses are erosion modeling, surface hydrology, watershed modelling geomorphology, land-sliding, agriculture and ecosystem modeling to list a few examples. This project intends to determine the amount of topographic data points that need to be collected in order to create an accurate model of the ground topography. To accomplish the objective, a topographic survey was conducted on a grid pattern, with a spacing of 7.5 feet between points regularly spaced over one acre. After the data were collected, varying percentages of the total amount of points collected were removed and the resulting digital elevation model (DEM) was compared to the ground truth DEM. When comparing accuracy of interpolated elevation across the entire DEM with a RMSE (root mean square error) it was found that using a subset of 25-30% of the entire data set were needed to create a model that did not significantly differ from the Ground Truth. The change in volume of the elevation surface compared to the Ground Truth results in a linear relationship, as more points are added the closed the change in volume is to zero. The P value derived from the T-test of the mean elevations of the trial DEM’s and the Ground truth, reflect the results from the change in volume, as more points are added the closer to the truth the DEM becomes.
Literary Rights: On
File Attachments: Final_Report_RGM.docx