Aspects of Digital Elevation Model Determination

The paper describes aspects of research being undertaken on the determination and processing of digital elevation models. A 2-D Kalman filter has been developed based on dynamic and functional models for a grid DEM, to produce optimal estimates of terrain attributes of a DEM point using the relevant observations, and predictions derived from its orthogonal neighbouring points. A further method of filtering DEMs based on Wavelets has been developed to compare it with the efficiency of the Kalman filtering method. An intelligent DEM procedure under development comprises several components using image processing, image understanding and terrain modelling. These include the recognition of some buildings on the terrain and the correction of errors in the elevations caused by these buildings, recognition of areas of vegetation on the terrain surface, which will lead to poor quality image matching and hence erroneous elevations. In addition, the geomorphology of the terrain surface, based on the drainage patterns are being derived from the computed DEMs. Further, ERS1/2 Tandem SAR data has been tested as an appropriate data set for upgrading the national digital elevation database over Australia.