Integration, validation and point spacing optimisation of digital elevation models

In recent years, collection and processing techniques for creating digital elevation models (DEMs; defined here as being surfaces composed of regular or irregular point data, without distinction) have advanced rapidly, allowing terrain to be represented with greater detail and accuracy. Accordingly, the amount of terrain data in existence proliferates as the resolution of sensors improves and acquisition costs lower. It is therefore becoming common for several models to cover any given area as higher accuracy or revised surveys are performed. Such coinciding data-sets allow improvement to the terrain depiction by integration techniques which merge and validate the individual DEMs. However, amalgamation of all available points is impractical, as errors and differences between the DEMs degrade the accuracy of the merged model, and the data volume may result in heavy oversampling. Consequently, this paper develops a DEM integration method by detailing the error budgets of individual and fused terrain models, then describing the merging procedure. Initially, comparison of the input surfaces is performed to isolate differences. Next, objects on the terrain surface are optionally removed before a surface matching algorithm is employed to detect and overcome systematic effects. Finally, the models are intelligently merged and the point distribution optimised to achieve an efficient surface representation. The methodology was used to integrate a dense airborne laser scanning DEM with sparser photogrammetric data encompassing a larger area. Surface matching proved to be vital, recovering a previously unidentifiable bias between the input DEMs, improving the accuracy of the fused surface