Logo des Repositoriums

Auflistung nach Schlagwort "forest management"

1 - 2 von 2
  • Konferenzbeitrag
    On the Realisation of a Workflow for Continuous Earth Observation of Forest Dynamics: A Performance Engineering Challenge
    (Softwaretechnik-Trends Band 44, Heft 4, 2024) Herbst, Nikolas; Jaggy, Niklas; Dingel, David; Linke, David; Kuenzer, Claudia; Kounev, Samuel
    Up-to-date data on forest dynamics is vital for forest management and understanding their impact on biodiversity and climate change mitigation. In this context, remote sensing has emerged as promising solution, especially for large-scale forest-related scenarios. We are implementing a satellite data processing workflow to continuously feed a mobile application: The goal is to provide timely and targeted information on forest dynamics, local disturbance events, and biodiversity changes for the entire Bavaria region. The engineering, automating and scaling of such a data-intense and distributed processing workflow, from high-volume satellite data time series to mobile applications, poses a variety of performance engineering challenges. We showcase first measurements of individual workflow task: The resource demanding disturbance detection executed in a Python Dask HPC environment contrasted against the resource-bound, optimized DBSCAN clustering of approximately 22 million points into (currently) multiple 100k disturbance events in the mobile application back-end.
  • Konferenzbeitrag
    On the Realisation of a Workflow for Continuous Earth Observation of Forest Dynamics: A Performance Engineering Challenge
    (Softwaretechnik-Trends Band 44, Heft 4, 2024) Herbst, Nikolas; Jaggy, Niklas; Dingel, David; Linke, David; Kuenzer, Claudia; Kounev, Samuel
    Up-to-date data on forest dynamics is vital for forest management and understanding their impact on biodiversity and climate change mitigation. In this context, remote sensing has emerged as promising solution, especially for large-scale forest-related scenarios. We are implementing a satellite data processing workflow to continuously feed a mobile application: The goal is to provide timely and targeted information on forest dynamics, local disturbance events, and biodiversity changes for the entire Bavaria region. The engineering, automating and scaling of such a data-intense and distributed processing workflow, from high-volume satellite data time series to mobile applications, poses a variety of performance engineering challenges. We showcase first measurements of individual workflow task: The resource demanding disturbance detection executed in a Python Dask HPC environment contrasted against the resource-bound, optimized DBSCAN clustering of approximately 22 million points into (currently) multiple 100k disturbance events in the mobile application back-end.