Geomorphic landform monitoring with raster and vector data cubes

Publikation: KonferenzbeitragAbstractPeer-reviewed

Abstract

Landscapes and geomorphic landforms are in constant change, where dynamic processes drive their evolution over time. Assessing these changes allows us to understand landscape patterns and interrelations. Further, monitoring the evolution of landforms related to natural hazards, like proglacial lakes, volcanic lava flows, landslides or gully erosion, is important for disaster risk prevention and mitigation. Advances in remote sensing techniques, such as data cubes, and the vast amounts of Earth observation data, allows the study of landscape dynamics. The gridded nature of data cubes facilitates the analysis of long time series of data at specific pixel locations. Despite their advantages, this type of queries over time ignore the spatial context of said pixel, focusing on a very limited portion of the area under study. Moving from a pixel to an object representation can improve the analysis of landscape dynamics, specially for geomorphological analyses, where landforms change their shape over time. Feature extraction techniques, such as object-based image analysis and deep learning, can aid with the detection of geomorphological features at different points in time. But once we transition from a coverage-based (array format) to a feature-based (vector format) data representation, the query and analysis advantages of a data cube are lost.

In this study, we explore the applicability of vector data cubes as a way to organise, analyse and visualise geomorphic landforms with changing geometries. The challenge lies on the changing shapes of the geomorphological features. At the implementation level, array-based and tabular representations are tested to build the data cubes. The approach is applied to volcanic lava flows as exemplary geomorphic landform. Further, the integration of vector data cube structures with raster data cubes are explored to aggregate information derived from Earth observation data over the landform geometries. The aggregation approach allows for further landform characterisation, matching the extracted data with consideration of the spatial and temporal properties of the landform. We believe that the usage of vector data cube representations could advance the spatio-temporal analysis and monitoring of landscapes and landforms, benefiting different disciplines related to the geosciences.
OriginalspracheEnglisch
DOIs
PublikationsstatusVeröffentlicht - 8 März 2024
VeranstaltungEGU General Assembly 2024 - Vienna, Österreich
Dauer: 14 Apr. 202419 Apr. 2024
https://www.egu24.eu/

Konferenz

KonferenzEGU General Assembly 2024
Land/GebietÖsterreich
OrtVienna
Zeitraum14/04/2419/04/24
Internetadresse

Systematik der Wissenschaftszweige 2012

  • 105 Geowissenschaften
  • 207 Umweltingenieurwesen, Angewandte Geowissenschaften

Dieses zitieren