Staff Highlights: Prof. Tiangang YIN
Forests, which cover 31 % of Earth’s land surface, play a vital role in the global climate system, carbon sequestration, and energy cycles. Monitoring forest ecosystems is therefore essential for understanding their dynamics and guiding sustainable management practices.
Prof. Tiangang Yin, from the Department of Land Surveying and Geo-informatics (LSGI), leads the 3-D Sensing, Modeling, and Data Intelligence (3MSI) Lab (https://www.3msi.net/). By leveraging multi-platform lidar technology and big data techniques, the team focuses on accurate 3D reconstruction of vegetation and enhanced retrieval of forest biophysical variables from diverse remote sensing data. Through the integration of advanced radiative-transfer models (RTMs) and artificial intelligence (AI) methods, they aim to develop a comprehensive Forest Digital Twin (FDT) system capable of delivering large-scale, time-series realistic forest representations, satellite observations, and energy-cycle simulations.
Topic 1: Large-Scale and High-Fidelity 3D Forest Reconstruction Using Airborne Lidar Data
The 3D structure of forests is critical for understanding complex ecological processes and conducting forest inventories, thereby supporting sustainable forest management. By providing 3D information, lidar technology is an ideal tool for reconstructing forest structures. Compared to terrestrial lidar, airborne laser scanning (ALS) can efficiently characterize extensive forested areas, making large-scale 3D reconstruction feasible.
To achieve this, we developed an ALS-driven, large-scale forest 3D reconstruction workflow (LS-PVlad), capable of producing high-resolution (2 m) voxelized 3D scenes covering up to 11,000 ha. Using this workflow, we created the FoScenes product, which comprises multiple high-fidelity scenes of various forest sites across North America. When combined with 3D RTM (e.g., DART and DART-EB, https://dart.omp.eu/index.php#/), FoScenes supports multi-scale remote sensing simulation and sensitivity analysis, demonstrating strong potential for enhancing global forest parameter retrieval and advancing ecological research.
Topic 2: Retrieving Forest Canopy Surface by Integrating Stereophotogrammetry into 3D Radiative Transfer Model
Canopy Height Model (CHM), as a critical parameter characterizing forest canopy’s structures, plays a vital role in forest inventory management, carbon sink quantification, biodiversity assessment, and etc. Spaceborne photogrammetry has emerged as a crucial approach for acquiring very-high-resolution (VHR) CHM data. However, the acquisition of high-quality VHR stereopairs data remains constrained by multiple factors including satellite revisit cycles and sensor's observational angles, etc.
Our team proposed a novel stereopair simulation pipeline integrating PVlad-derived 3D LAD forest scenes with a 3D radiative transfer model (i.e., DART). Through simulating extensive VHR satellite stereopairs, this approach enables systematic sensitivity analysis of key factors affecting the estimation of forest canopy height. This research direction holds significant guidance value for future satellite mission design and the development of forest CHM retrieving algorithms.
Topic 3: Monitoring Urban Trees Using Multi-Source Lidar Data
Urban trees play a crucial role in human well-being by improving air quality, reducing the urban heat island effect, and enhancing living environments. Therefore, large-scale, long-term, and precise monitoring of key metrics—such as tree count, leaf area, and health—has become vital for urban management. Traditional manual surveys struggle with efficiency, coverage, and real-time needs, highlighting the potential of lidar combined with multisource data for accurate monitoring.
Our team has collected long-term lidar data from hundreds of urban trees in Hong Kong, creating a unique database in collaboration with local authorities. We focus on single-tree point cloud processing, including branch/leaf classification, occlusion completion, 3D reconstruction, and biomass/leaf area density estimating, using deep learning techniques trained on both empirical and simulated data.
References:
Yin, T., Cook, B. D. & Morton, D. C., 2022. Three-dimensional estimation of deciduous forest canopy structure and leaf area using multi-directional, leaf-on and leaf-off airborne lidar data. Agricultural and Forest Meteorology. 314, 108781. https://doi.org/10.1016/j.agrformet.2021.108781.
Yin, T., Montesano, P. M., Cook, B. D., Chavanon, E., Neigh, C. S. R., Shean, D., Peng, D., Lauret, N., Mkaouar, A., Morton, D. C., Regaieg, O., Zhen, Z. & Gastellu-Etchegorry, J.-P., 2023. Modeling forest canopy surface retrievals using very high-resolution spaceborne stereogrammetry: (I) methods and comparisons with actual data. Remote Sensing of Environment. 298, 113825. https://doi.org/10.1016/j.rse.2023.113825.
Yin, T., Montesano, P. M., Cook, B. D., Chavanon, E., Neigh, C. S. R., Shean, D., Peng, D., Lauret, N., Mkaouar, A., Regaieg, O., Zhen, Z., Qin, R., Gastellu-Etchegorry, J.-P. & Morton, D. C., 2023. Modeling forest canopy surface retrievals using very high-resolution spaceborne stereogrammetry: (II) optimizing acquisition configurations. Remote Sensing of Environment. 298, 113824. https://doi.org/10.1016/j.rse.2023.113824.
Wei, S. S., Yin, T. G., Dissegna, M. A., Whittle, A. J., Ow, G. L. F., Yusof, M. L. M., Lauret, N. & Gastellu-Etchegorry, J. P., 2020. An assessment study of three indirect methods for estimating leaf area density and leaf area index of individual trees. Agricultural and Forest Meteorology. 292, 108101. https://doi.org/10.1016/j.agrformet.2020.108101.
