it_team4_FOREst 5 December 2024New Research Advances Understanding of Storm Damage in European Forests
A recent study published by IT4Forest Stefan Reder with Melissa Kruse, Luis Miranda and Nicole Albert addresses the significant biomass losses caused by severe storm events in European forests. While direct storm damage is a well-known issue, this research highlights the additional impacts of biotic, abiotic, and market-related factors that exacerbate the situation, posing serious challenges for forest and nature conservation.
The study emphasises the importance of accurately assessing the extent of storm damage to facilitate effective salvage operations. Traditional methods and satellite-based remote sensing have limitations, primarily in delineating affected areas and estimating the volume of damaged timber. Although aerial remote sensing can identify the spatial distribution of tree trunks with a detection rate of up to 92%, it falls short in quantifying the volume of individual trunks.
To fill this gap, the researchers employed innovative UAV (Unmanned Aerial Vehicle) technology combined with a hybrid heuristic and deep learning approach. This methodology not only captures the spatial distribution of tree trunks but also estimates their volume, even reconstructing occluded parts of the stems. The deep learning model was trained using high-resolution orthomosaics, where 760 trunks were meticulously outlined and labeled.
The results are promising, with an average stem detection rate of 92.6%, a classification error rate of just 1.6%, and a reconstruction error rate of 0.2%. When quantifying stem volume, the study reported a relative bias of -4.35% for 560 reference stems across 15 investigation sites. However, results varied by site, influenced by weather conditions during flights, stand parameters, and the specific wind-throw patterns.
This innovative approach not only enhances the management of wind-throw areas but also contributes to monitoring biomass and carbon cycles. By improving the understanding of wind-throw dynamics, the research supports the development of sustainable forest management strategies, ultimately aiding in the conservation of European forests in the face of increasing storm events.
Suggested Citation:
Stefan Reder, Melissa Kruse, Luis Miranda and Nicole Albert: Unveiling windthrown trees: Detection and quantification of windthrown tree stems on UAV orthomosaics based on UNet and a heuristic stem reconstruction. Forest Ecology and Management 578, License
DOI:10.1016/j.foreco.2024.122411
The presented research was part of WINMOL, a shared project of the Eberswalde University for Sustainable Development and the Thünen Institute of Forest Ecosystem Eberswalde, funded by the FNR Fachagentur für Nachwachsende Rohstoffe e.V. and the Federal Ministry of Food and Agriculture, Germany (BMEL) (FNR funding number: 2220NR024A)