Establish Empirical Estimation of Uncertainty


To better calibrate the output values of a FIS modeling DEM uncertainty empirical estimations of uncertainty can be established. This method is preferred over past practices of relying on factory reported uncertainty or expert judgement on the uncertainty for a surveying platform

Methods to Obtain Estimation

Different methods exist to establish an empirical estimation of uncertainty: finding coincident points, boot strapping exercises, use of unchanging surfaces, and use of unreasonable change areas in DoDs.

Coincident Points

This method can be used for surveying methods which have sufficient point density where x,y locations may have been surveyed multiple times. This often occurs with automated survey methods such as terrestrial laser scanners (TLS), multi-beam echo sounders (MBES), and structure from motion (SFM). When x,y locations are surveyed multiple times the z values at these locations can be differenced to obtain the uncertainty at the x,y locations. This is shown for MBES data in the figure below:

This method has been automated by the Coincident Points Tool in GCD 6.0. This tool is located under the Analysis -> Uncertainty Analysis  menu under the Point Cloud Based menu. Documentation for this tool can be found here.

TO DO: Bootstrapping

TO DO: Use of Unchanged Surfaces

TO DO:Use of Unreasonable Change Areas in DoDs

Demonstration of Applying Methods

Coincident Points

After running the Coincident Points Tool in GCD 6.0 a point feature class which contains a field for the uncertainty values for each point is created. By summarizing the values of the uncertainty field an understanding of the range of uncertainty values can be understood. For example a MBES survey from Slaughter Gulch produced the following range in uncertainty values.