Tracing snow cover dynamics using 4D level-sets on near-continuous terrestrial laser scanning time series

• Verfasst von: Arav, Reuma [VerfasserIn]
• Verfasst von: Anders, Katharina [VerfasserIn]
• Verfasst von: Pöppl, Florian [VerfasserIn]
• Verfasst von: Höfle, Bernhard [VerfasserIn]
• Verfasst von: Pfeifer, Norbert [VerfasserIn]
• Titel: Tracing snow cover dynamics using 4D level-sets on near-continuous terrestrial laser scanning time series
• Verf.angabe: Reuma Arav, Katharina Anders, Florian Poeppl, Bernhard Hoefle, Norbert Pfeifer
• E-Jahr: 2023
• Jahr: February 2023
• Umfang: 1 S.
• Fussnoten: Gesehen am 11.04.2024
• Titel Quelle: Enthalten in: Workshop on Land Ice and Snow (10 : 2023 : Bern)Remote sensing of the cryosphere
• Ort Quelle: Bern : Universität Bern, 2023
• Jahr Quelle: 2023
• Band/Heft Quelle: (2023) vom: Feb., Seite 126
• DOI: doi:10.13140/RG.2.2.27561.06243
• Datenträger: Online-Ressource
• Sprache: eng
• K10plus-PPN: 1885632614

Abstract

Accurate monitoring of snow cover dynamics contributes to the understanding of the changing cryosphere, enables the characterization of hydro-logical processes, and can help managing natural hazards. However, the high-mountain topography and the indistinctness of snow upon the surface turn such monitoring to a challenge. Furthermore, processes which occur at different rates and at various spatial extents, like deposition through snow drift, cannot be recognized at a single point in time. Therefore, they cannot be easily separated, quantified, and characterized. Recent years have seen an increase in near-continuous 3D observations that capture Earth surface dynamics at a local landscape scale. Such measurements enable advanced detection of surface changes. For example, by identifying similar change histories, surface modifications through time can be delineated. This way, changes that originate from different processes but coincide spatially and/or temporally - are still extracted. Such methods benefit from the combination of the temporal domain with the spatial one. However, they do not consider the spatiotemporal variability of surface activities. In this - study, we propose to trace snow cover dynamics in near-continuous terrestrial laser scans by utilizing a level set framework. We formulate the extraction of changes as a minimization problem where the zero level-set represents their borders. Using a fully automated 3D region-based level-set approach, we extract entities that share similar change rates - at a specific point in time directly from the point cloud. This way, we identify snow cover dynamics in 4D which may be indistinguishable in a single scan, but have a distinct signature through time. We demonstrate the proposed method on hourly - terrestrial laser scans of snow cover acquired at the Zugspitze (Schneeferner) in Germany over five days in April 2018. These scans feature snow melt and compaction, avalanches, and anthropogenic modifications (e.g., snow farming). Results show - that using the proposed method, we are able to trace different phenomena with a variety of spatial and temporal properties (e.g., timing, duration, magnitude, change rate, spatial scale). These are - extracted simultaneously across multiple epochs, depending on their duration and in a single analysis process. The focus on geometric properties through time suggests that we are unbound to predefined types or numbers of surface processes. Thus, the proposed method paves the way to a general monitoring which can capture a large variety of surface dynamics in complex cryospheric scenes through time.