Detailed 3D reconstruction of trees from TLS for radiative transfer modelling applications

0570662 - ÚVGZ 2024 DE eng A - Abstrakt
Janoutová, Růžena - Homolová, Lucie - Malenovský, Z. - Novotný, Jan - Navrátilová, Barbora
Detailed 3D reconstruction of trees from TLS for radiative transfer modelling applications.
Book of Abstracts FSPM2020. Hannover: Institute of Horticultural Production Systems, 2020. s. 135-136.
[9th International Conference on Functional-Structural Plant Models (FSPM2020). 05.10.2022-09.10.2022, Hannover]
Grant CEP: GA MŠMT(CZ) LTC20055
Institucionální podpora: RVO:86652079
Klíčová slova: 3d tree model * european beech * norway spruce * white peppermint * tls
Obor OECD: Remote sensing
https://www.fspm2023.net/boa-fspm2020

Dynamic changes in forest ecosystems caused by climatic changes are being more frequently assessed using remote sensing (RS) based methods which allow extensive monitoring in a relatively short time. Radiative transfer models (RTM) provide a physical link between canopy biochemical and structural properties and canopy reflectance, therefore they are often used to interpret RS data. RTM that are capable implementing detailed 3D objects can precisely simulate interactions between canopy elements and incident light in various parts of the electromagnetic spectra.

In this study we present a method for 3D reconstruction of forest tree species and implementation in 3D canopy RTM Discrete Anisotropic Radiative Transfer model (DART, Gastellu-Etchegorry et al., 2017). We selected three forest species, Norway spruce (Picea abies) and European beech (Fagus sylvatica) as representatives of European productive forests and White peppermint (Eucalyptus pulchella) as a common forest specie in Tasmania. All these species have their specific structural features. European beech represents deciduous forest species. In sense of structural characteristics there are simple shaped leaves of the same age that are distributed typically in spherical leaf angle distribution (LAD). Norway spruce is an evergreen coniferous specie. Compared to broadleaved species the tree structure is more complex and it is challenging to reconstruct it up to the level of single needles. Moreover, the presence of needle shoots with different age requires to separate the shoots in 3D tree model into two different age categories at least. White peppermint is structurally different from common European species. It is a broadleaf, but evergreen specie therefore it is necessary to separate leaves into different age categories. Moreover, it has sparse crown architecture with leaves predominantly distributed according to erectophile LAD.

We design an algorithm for reconstruction of 3D tree model based on terrestrial laser scanning (TLS) data and field measured characteristics as LAD, percentage of current and older needles or leaves etc., which we are not able to derive precisely from TLS data in close forest formation. The tree reconstruction algorithm was primarily designed for Norway spruce (Janoutová et al., 2019) and here it is expanded to broadleaf beech and peppermint trees. The algorithm contains four main steps: i) geometrical transformation of the point cloud to fit site-specific tree dimensions, ii) reconstruction of wooden skeleton from TLS data, iii) distribution of shoot or leaf positions within the tree crown, and iv) separation of the positions into two age categories. The last step is applicable only for spruce and peppermint trees.

The reconstructed 3D models of the tree species were used to build up small forested scenes in the DART model. This contribution summarizes canopy reflectance simulations, its angular behavior and comparison with spectral signatures extracted from airborne imaging spectroscopy data.

Trvalý link: https://hdl.handle.net/11104/0341989