Calders, K. | bioRxiv Science

bioRxiv2022

TLS2trees: a scalable tree segmentation pipeline for TLS data

Above Ground Biomass (AGB) is an important metric used to quantify the mass of carbon stored in terrestrial ecosystems. For forests, this is routinely estimated at the plot scale (typically [≥]1 ha) using inventory measurements and allometry. In recent years, Terrestrial Laser Scanning (TLS) has appeared as a disruptive technology that can generate a more accurate assessment of tree and plot scale AGB; however, operationalising TLS methods has had to overcome a number of challenges. One such challenge is the segmentation of individual trees from plot level point clouds that are required to estimate woody volume, this is often done manually (e.g. with interactive point cloud editing software) and can be very time consuming. Here we present TLS2trees, an automated processing pipeline and set of Python command line tools that aims to redress this processing bottleneck. TLS2trees consists of existing and new methods and is specifically designed to be horizontally scalable. The processing pipeline is demonstrated across 10 plots of 7 forest types; from open savanna to dense tropical rainforest, where a total of 10,557 trees are segmented. TLS2trees segmented trees are compared to 1,281 manually segmented trees. Results indicate that TLS2trees performs well, particularly for larger trees (i.e. the cohort of largest trees that comprise 50% of total plot volume), where plot-wise tree volume bias is {+/-}0.4 m3 and %RMSE is ~60%. To facilitate improvements to the presented methods as well as modification for other laser scanning modes (e.g. mobile and UAV laser scanning), TLS2trees is a free and open-source software (FOSS).

ecology↗

bioRxiv2022

Tree segmentation in airborne laser scanning data is only accurate for canopy trees

Individual tree segmentation from airborne laser scanning data is a longstanding and important challenge in forest remote sensing. There are a number of segmentation algorithms but robust intercomparison studies are rare due to the difficulty of obtaining reliable reference data. Here we provide a benchmark data set for temperate and tropical broadleaf forests generated from labelled terrestrial laser scanning data. We compare the performance of four widely used tree segmentation algorithms against this benchmark data set. All algorithms achieved reasonable accuracy for the canopy trees, but very low accuracy for the understory trees. The point cloud based algorithm AMS3D (Adaptive Mean Shift 3D) had the highest overall accuracy, closely followed by the 2D raster based region growing algorithm Dalponte2016+. This result was consistent across both forest types. This study emphasises the need to assess tree segmentation algorithms directly using benchmark data. We provide the first openly available benchmark data set for tropical forests and we hope future studies will extend this work to other regions.

ecology↗

bioRxiv2021

SimpleForest - a comprehensive tool for 3d reconstruction of trees from forest plot point clouds

The here-on presented SimpleForest is written in C++ and published under GPL v3. As input data SimpleForest utilizes forestry scenes recorded as terrestrial laser scan clouds. SimpleForest provides a fully automated pipeline to model the ground as a digital terrain model, then segment the vegetation and finally build quantitative structure models of trees (QSMs) consisting of up to thousands of topologically ordered cylinders. These QSMs allow us to calculate traditional forestry metrics such as diameter at breast height, but also volume and other structural metrics that are hard to measure in the field. Our volume evaluation on three data sets with destructive volumes show high prediction qualities with concordance correlation coefficient CCC [Formula] of 0.91 (0.87), 0.94 (0.92) and 0.97 (0.93) for each data set respectively. We combine two common assumptions in plant modeling "The sum of cross sectional areas after a branch junction equals the one before the branch junction" (Pipe Model Theory) and "Twigs are self-similar" (West, Brown and Enquist model). As even sized twigs correspond to even sized cross sectional areas for twigs we define the Reverse Pipe Radius Branchorder (RPRB) as the square root of the number of supported twigs. The prediction model radius = B0 * RPRB relies only on correct topological information and can be used to detect and correct overestimated cylinders. In QSM building the necessity to handle overestimated cylinders is well known. The RPRB correction performs better with a CCC [Formula] of 0.97 (0.93) than former published ones 0.80 (0.88) and 0.86 (0.85) in our validation. We encourage forest ecologists to analyze output parameters such as the GrowthVolume published in earlier works, but also other parameters such as the GrowthLength, VesselVolume and RPRB which we define in this manuscript. Upload statementSelf-uploaded pre-print for peer-review submitted manuscript. The manuscript was submitted on 26th of July 2021 to Plos Computational Biology: I, Jan Hackenberg uploaded this manuscript because the automated journal upload was rejected for the following reason: Thank you for considering posting your manuscript "SimpleForest - a comprehensive tool for 3d reconstruction of tree from forest plot point clouds." as a preprint. Your manuscript does not meet bioRxivs criteria and therefore we will not be sending it for posting as a preprint. For more information about our checks, see link. We have noted that it contains material that is potentially subject to copyright. In particular, screenshot in Figure 1. Preprints posted to bioRxiv following submission to PLOS journals are done so under the CC BY license. To avoid a potential breach of the copyright that applies to the material listed above, we are unable to make the manuscript publicly available. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=62 SRC="FIGDIR/small/454344v1_fig1.gif" ALT="Figure 1"> View larger version (11K): org.highwire.dtl.DTLVardef@1094b17org.highwire.dtl.DTLVardef@120d7a3org.highwire.dtl.DTLVardef@12d2fbcorg.highwire.dtl.DTLVardef@1990a4d_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOFig 1.C_FLOATNO Submission system screenshot. C_FIG Please note that this decision does not affect the editorial process at PLOS Computational Biology. Your manuscript is being separately assessed with regards to sending for peer review. From section Abstract on, the pdf you see is same as submitted one.

plant biology↗