LAND COVER DEPENDENT DERIVATION OF DIGITAL SURFACE MODELS FROM AIRBORNE LASER SCANNING DATA

Abstract

In contrast to the algorithms for determining digital terrain models (DTMs) from airborne laser scanning (ALS) data, those for the generation of digital surface models (DSMs) are rather straightforward. A common way to derive DSMs is the determination of the highest point within a defined raster cell (DSMmax) or the determination of the DSM heights based on moving least squares interpolation e.g. moving planes (DSMmls). For low ALS point densities void pixels can occur in the DSMmax and for inclined smooth surfaces the DSMmax shows an artificial roughness mainly due to the irregular point distribution. These disadvantages of the DSMmax can be reduced by using an interpolated grid DSMmls. On the other hand, the DSMmls introduces smoothing effects along surface discontinuities e.g. building borders, tree tops, forest gaps, etc. In this paper a combined approach for the DSM generation is presented, which makes use of the strengths of both, the DSMmax and the DSMmls. The used algorithms are implemented in the scientific software package OPALS (Orientation and Processing of Airborne Laser Scanning data). The proposed processing chain for DSM determination is applied for four test sites in Austria with different land cover types. For these test sites ALS data with point densities between 5.5 and 60.0 echoes per m 2 are available. The derived DSMs are presented and compared to traditional DSMs. Especially, for deciduous forests the difference between the single DSMs can add up to several metres.