MorphoSAT - Automated geomorphological mapping based on satellite data

Abstract

Introducing the MorphoSAT project in a nutshell

Full text

Selected landforms / processes
MorphoSAT
Automated geomorphological mapping based on satellite data
Clemens EISANK1*, Frederic PETRINI-MONTEFERRI2, Gertraud MEISSL3, Sebastian D‘OLEIRE-OLTMANNS4,
Filippo VECCHIOTTI5, Volker WICHMANN2, Christian GEORGES2, Arben KOCIU5, Johann STÖTTER3
1 GRID-IT GmbH, Innsbruck, Austria | 2 Laserdata GmbH, Innsbruck, Austria | 3 Institute of Geography, University of Innsbruck, Austria |
4 Interfaculty Department of Geoinformatics – Z_GIS, University of Salzburg, Salzburg, Austria | 5 Geological Survey of Austria, Vienna, Austria
* Corresponding author
Dr. Clemens Eisank
GRID-IT GmbH
Innsbruck, Austria
eisank@grid-it.at
The research project MorphoSAT (Automated geomorphological mapping based on satellite data) is co-funded by the Austrian
Research Promotion Agency (FFG) through the Austrian Space Applications Programme (ASAP) under project number 859727.
TanDEM-X DEM data was kindly provided by the German Aerospace Center (DLR).
1| BACKGROUND & MOTIVATION
GIS-ready geomorphological maps are rare, but required by decision makers,
spatial planners, energy companies and others to support applications such as
natural hazards zonation, water management, land conservation, as well as
exploration and management of natural resources.
3| METHODS & INNOVATION
Recent (near-)global and consistent remote sensing datasets are used,
exploited, integrated and validated (TanDEM-X DEM, Sentinel-2)
Coupling of machine learning with object-based image analysis (OBIA) to
reach a high level of automation and interoperability of the mapping
Mapping of semantically-rich geomorphological features, i.e. landforms
such as alluvial fans and landslides, rather than of basic terrain elements
Standardization in automated geomorphological mapping is promoted
High usability of project results ensured by cooperation of industry,
public authorities and academia.
2| AIM
With the availability of new consistent near-global high-resolution satellite data
and products, the time is right to increase objectivity, applicability and
automation in the field of digital geomorphological mapping with the main aim
to produce consistent, objective and GIS-ready geomorphological maps at
regional/near-global scales.
4| CONCLUSION
The developed MorphoSAT geomorphological mapping system will be more
objective, interoperable and spatially independent than previous workflows,
and therefore, will enable large-scale production of semantically-rich
geomorphological information.
Download vCard
PROJECT TEAM ACKNOWLEDGEMENTS
STUDY AREA SATELLITE-BASED GEODATA LAYERS
TanDEM-X DEM (10m)
Geodata-
base
Derived terrain layers Copernicus layers Sentinel-2 image & indices
… … …
Training polygons
produced by experts
Multi-scale segmentation of input layers
0
0,2
0,4
0,6
0,8
1
Mean Std.Dev. Maximum Minimum
Value (stanardized)
Statistical feature
Slope
Curvature
Brightness
…
Layer
Expert descriptions
OPERATIONAL GEOMORPHOLOGICAL KNOWLEDGE MODELING
Prototypical data signatures
Structured models
Machine Learrning
based on layer properties
inside training polygons
Knowledge modeling
Operational
geomorphological
models
AUTOMATED EXTRACTION / CLASSIFICATION
Input layers
Layer weights
Classification features
Feature thresholds
Iterative
Object-Based
Image Analysis
VALIDATION
Landforms / processes
Expert validation
Quantitative assessment
Curvature
Slope
Coarse scale Focal scale Fine scale
Corine Land Cover Sentinel-2 RGB mosaic
© DLR, 2017