IWR / HGS MathComp

Duration: 2017–2021


Katharina Anders
Bernhard Höfle
Hubert Mara
FCGL – Forensic Computational Geometry Laboratory
Thomas Scharffenberger


Dr. Roderik Lindenbergh
Department of Geoscience & Remote Sensing, TU Delft (NL)


Research Project

Auto3Dscapes - Autonomous 3D Earth Observation of Dynamic Landscapes


Press release: Understanding the Spatial and Temporal Dimensions of Landscape Dynamics (March 2021)

All research news can be found in our GIScience News Blog. Follow project updates on ResearchGate and stay tuned on Twitter: #Auto3Dscapes.


Environmental monitoring is challenged by Earth surface processes of spatially and temporally dynamic character which permanently affect our physical and human environment. Autonomous terrestrial laser scanning systems (ATLS) have recently become available for permanently observing dynamic landscapes, enabling even hypertemporal (daily to sub-hourly) surveying of natural dynamics. These developments lead to increased demands in powerful methods for processing and analysis.

The strong spatiotemporal variability of Earth shaping processes poses significant challenges to current computational methods for 3D observation. Autonomous sensing generates an almost continuous time series of big volumes of 3D point clouds, which demands for improved strategies of capturing and analyzing Earth surface dynamics. As current 3D monitoring systems are working statically with pre-defined and fixed settings, the underlying challenge for improvement is to capture 3D data only when and where significant changes (potentially) occur, in order to keep data volumes low and make near real-time observation possible.


The main objective of the project is to develop a computational method for autonomous 3D Earth observation which interlinks the decoupled processes of surveying and data processing by integrating observed Earth surface dynamics as feedback. This approach enables immediate automatic adaptation of the 3D observation strategy (acquisition and analysis) to account for the specific spatiotemporal properties that have been obtained from the accumulating time series datasets. This dynamic adaptation requires automatic 3D geospatial data processing and analysis.


The dynamic methodological approach of Auto3Dscapes consists of three main components:

  • Process characterization through timeseries-based object detection and change analysis
  • Geographic forward modeling of observed change and laserscanning simulation in the virtual future landscape
  • Automatic adaptation of the acquisition and processing strategy
Related Projects
  • HELIOS++: Heidelberg LiDAR Operations Simulator.
  • GEODYNAMO4D: Tracing geographic dynamics on 4D Point Clouds.
  • Geomorph4D: Characterising multi-process geomorphic change.
  • AHK-4D: High-resolution and high-frequency monitoring of the rock glacier Äußeres Hochebenkar (AHK) in Austria.
Editor: Webmaster Team
Latest Revision: 2023-03-14
zum Seitenanfang/up