SYSSIFOSS - Synthetic structural remote sensing data for improved forest inventory models
News
Research news can be found in our GIScience News Blog and on X/Twitter: #SYSSIFOSS.
New paper: Check out our latest paper where we investigate convolutional neural network (CNN)-based aboveground biomass prediction using cross-sections through airborne laser scanning point clouds of forests. We used a 3D version of the VGG16 CNN with initial weights transferred from pre-training on the ImageNet dataset and further tested additional pre-training on virtual laser scanning data of synthetic forest stands, simulated with HELIOS++.
- Schäfer, J., Winiwarter, L., Weiser, H., Höfle, B., Schmidtlein, S., Novotný, J., Krok, G., Stereńczak, K., Hollaus, M. & Fassnacht, F.E. (2024): CNN-based transfer learning for forest aboveground biomass prediction from ALS point cloud tomography. European Journal of Remote Sensing, pp. 1-18. DOI: 10.1080/22797254.2024.2396932.
Motivation
Airborne light detection and ranging (LiDAR) data provides reliable information on forest structure. Related forest inventory approaches recently evolved into operational tools. Today, further optimization of existing approaches is pursued to ensure high data quality of the inventory information and cost-efficiency over varied environmental and silvicultural conditions. Synthetic LiDAR data has been suggested as useful tool to better understand the interactions between forest canopies and LiDAR acquisitions and hence as a key instrument for identifying further optimization potential since it enables to create remote sensing datasets that cover the full variability of the investigated environmental, silvicultural and technical parameters. However, so far synthetic LiDAR data has either been simulated with a very high level of detail and for small areas or with simplistic approaches for larger areas.
Creating synthetic airborne LiDAR data in HELIOS++
Objective
SYSSIFOSS is a joint project between the Institute of Geography and Geoecology (IFGG) of the Karlsruhe Institute of Technology (KIT) and the 3DGeo Research Group of Heidelberg University.
In this project we suggest a new approach to create synthetic LiDAR data by combining the outputs of an established forest growth simulator with a to-be-created database of species-specific model trees extracted from real LiDAR point clouds. This approach will result in inventory information at the single tree level and a matching 3D forest structure for large areas. The 3D forest structure will serve as input to HELIOS++ (Heidelberg LiDAR Operations Simulator), a LiDAR ray-tracing tool with which accurate LiDAR acquisitions can be simulated. Based on the HELIOS++ simulations, we will on the one hand conduct a sensitivity analysis (considering e.g., field inventory design, field plot size, statistical model, LiDAR acquisition settings, etc.) to identify the most important factors influencing LiDAR based forest inventories and thereby identify optimization potentials. On the other hand, we will examine the potential of the created synthetic data to minimize the amount of field-collected reference data. The latter will be realized by developing a look-up table like approach where synthetic data matching the local conditions of the area for which real LiDAR data is available are used to calibrate models which can directly be applied to the real LiDAR dataset. The project will focus on central European forests, but the concepts developed in the project are applicable to forests worldwide.
Previous Work
In a recent study Fassnacht et al. (2018) found that synthetic remote sensing datasets are a powerful approach for testing models in the absence of suitable large field data and to optimize workflows, model choices and even data collection. HELIOS++ has been applied in a forestry context in several studies for example to investigate the effect of LiDAR platform and flight planning (incl. UAV-LiDAR) for understory forest structure modeling (Hämmerle et al. 2017). This proof-of-concept study concluded the suitability of HELIOS++ for geometry-based LiDAR studies. Moreover, the 3DGeo group developed LVISA (LiDAR Vegetation Investigation and Signature Analysis System (Koenig et al. 2013), which builds the basis for a model tree online database that is required in this study.
Project Videos
Related Projects
- HELIOS++: Heidelberg LiDAR Operations Simulator.
- pytreedb
- VirtuaLearn3D:Virtual Laser Scanning for Machine Learning Algorithms in Geographic 3D Point Cloud Analysis.
- Landcover monitoring with emphasis on vegetation under the climatic change pressure using multitemporal and multisource remote sensing data fusion.
Publications
- Schäfer, J., Winiwarter, L., Weiser, H., Höfle, B., Schmidtlein, S., Novotný, J., Krok, G., Stereńczak, K., Hollaus, M. & Fassnacht, F.E. (2024): CNN-based transfer learning for forest aboveground biomass prediction from ALS point cloud tomography. European Journal of Remote Sensing, pp. 1-18. DOI: 10.1080/22797254.2024.2396932.
- Schäfer, J., Winiwarter, L., Weiser, H., Novotný, J., Höfle, B., Schmidtlein, S., Henniger, H., Krok, G., Stereńczak, K. & Fassnacht, F.E. (2023): Assessing the potential of synthetic and ex situ airborne laser scanning and ground plot data to train forest biomass models. Forestry: An International Journal of Forest Research. Vol. 2023, pp. 1-19. DOI: 10.1093/forestry/cpad061.
- Schäfer, J., Weiser, H., Winiwarter, L., Höfle, B., Schmidtlein, S. & Fassnacht, F.E. (2023): Generating synthetic laser scanning data of forests by combining forest inventory information, a tree point cloud database and an open-source laser scanning simulator. Forestry: An International Journal of Forest Research. Vol. 2023, pp. 1-19. DOI: 10.1093/forestry/cpad006.
- Weiser, H., Schäfer, J., Winiwarter, L., Krašovec, N., Fassnacht, F.E. & Höfle, B. (2022): Individual tree point clouds and tree measurements from multi-platform laser scanning in German forests. Earth System Science Data. Vol. 14 (7), pp. 2989-3012. DOI: 10.5194/essd-14-2989-2022.
- Weiser, H., Schäfer, J., Winiwarter, L., Krašovec, N., Seitz, C., Schimka, M., Anders, K., Baete, D., Braz, A.S., Brand, J., Debroize, D., Kuss, P., Martin, L.L., Mayer, A., Schrempp, T., Schwarz, L.-M., Ulrich, V., Fassnacht, F.E. & Höfle, B. (2022): Terrestrial, UAV-borne, and airborne laser scanning point clouds of central European forest plots, Germany, with extracted individual trees and manual forest inventory measurements. PANGAEA. DOI: 10.1594/PANGAEA.942856.
- Weiser, H., Winiwarter, L., Schäfer, J., Fassnacht, F.E. & Höfle, B. (2022): Airborne laser scanning (ALS) point clouds with full-waveform (FWF) data of central European forest plots, Germany. PANGAEA. DOI: 10.1594/PANGAEA.947038.
- Winiwarter, L., Esmorís Pena, A., Weiser, H., Anders, K., Martínez Sanchez, J., Searle, M., Höfle, B. (2022): Virtual laser scanning with HELIOS++: A novel take on ray tracing-based simulation of topographic full-waveform 3D laser scanning. Remote Sensing of Environment. Vol. 269. DOI: 10.1016/j.rse.2021.112772.
- Weiser, H., Winiwarter, L., Anders, K., Fassnacht, F.E. & Höfle, B. (2021): Opaque voxel-based tree models for virtual laser scanning in forestry applications. Remote Sensing of Environment. Vol. 265, pp. 112641. DOI: 10.1016/j.rse.2021.112641.
- Fassnacht, F.E., Schäfer, J., Weiser, H., Winiwarter, L., Krasovec, N., Latifi, H. & Höfle, B. (2021): Presenting the GeForse approach to create synthetic LiDAR data from simulated forest stands to optimize forest inventories. In: EGU General Assembly 2021. Vol. EGU21 (EGU21-9197), pp. 1-2. DOI: 10.5194/egusphere-egu21-9197.
- Weiser, H., Winiwarter, L., Schäfer, J., Fassnacht, F.E., Anders, K., Esmoris Pena, A.M. & Höfle, B. (2021): Virtual laser scanning (VLS) in forestry – Investigating appropriate 3D forest representations for LiDAR simulations with HELIOS++. In: EGU General Assembly 2021. Vol. EGU21 (EGU21-9178), pp. 1-2. DOI: 10.5194/egusphere-egu21-9178.
- Schäfer, J., Faßnacht, F.E., Höfle, B. & Weiser, H.(2019): Das SYSSIFOSS-Projekt: Synthetische 3D-Fernerkundungsdaten für verbesserte Waldinventurmodelle.In: 2. Symposium zur angewandten Satellitenerdbeoachtung, Cologne, Germany, pp.1-1.