University of Twente Student Theses
Intercomparison of methods for estimating Leaf Angle Distribution with terrestrial LiDAR for broadleaf tree species
Mutugi Murithi, Chris (2023) Intercomparison of methods for estimating Leaf Angle Distribution with terrestrial LiDAR for broadleaf tree species.
|
PDF
3MB |
| Abstract: | Monitoring changes in biodiversity and understanding their impact on society is crucial for conservation efforts and ecosystem management. Essential Biodiversity Variables (EBVs) have been established as a standardised framework to comprehend and track biodiversity changes, supporting these efforts. One key EBV that provides valuable information on ecosystem structure and function is the Leaf Area Index (LAI), a critical biophysical parameter of vegetation. The estimation of LAI using indirect methods is highly dependent on the leaf angle distribution (LAD). However, accurately measuring LAD has been challenging, leading to the limited characterisation of this parameter. It is often simplified using predefined mathematical functions, overlooking its actual variation. In recent years, Terrestrial Laser Scanning (TLS) instruments have emerged as valuable tools for acquiring detailed measurements of canopy structure. Various methods have been proposed to extract LAD information from TLS data, but the intercomparison of these methods is yet to be reported. This study addressed this gap by testing and comparing available TLS-based algorithms for estimating LAD using real tree data obtained from terrestrial Light Detection and Ranging (LiDAR) scans. The performance of these TLS-based algorithms was also evaluated against the established levelled digital photography (LDP) approach, and simulated LiDAR data from synthetic 3D tree models was used for further analysis. The inter-comparison results indicated that, while there were some notable variations, most TLS-based algorithms did not exhibit significant differences in their LAD estimates for real trees when compared to other algorithms. Among the TLS-based algorithms, the Bailey and Mahaffee (2017), Point Cloud Library (PCL), Vicari et al. (2019), and Zheng and Moskal (2012) algorithms demonstrated better performance than the Liu et al. (2019) and Stovall et al. (2021) algorithms, which performed poorly in their LAD estimation. Furthermore, the use of synthetic scans revealed that TLS algorithms showed better performance in estimating LADs that had more leaf area facing the scanner's direction. The study highlighted that the algorithms that used merged point clouds performed better than their single-scan counterparts. Overall, TLS offered a more comprehensive representation of the canopy structure, capturing the entire canopy height profile and overcoming the limitations of the LDP approach. Benchmark datasets, evaluation protocols and availability of algorithms shall promote fairness, reproducibility, and the advancement of LAD estimation techniques by enabling researchers to identify strengths, weaknesses, and further areas for improvement in their algorithms. |
| Item Type: | Essay (Master) |
| Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
| Subject: | 38 earth sciences |
| Programme: | Geoinformation Science and Earth Observation MSc (75014) |
| Link to this item: | https://purl.utwente.nl/essays/97923 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page