open access publication

Article, 2024

Deep point cloud regression for above-ground forest biomass estimation from airborne LiDAR

Remote Sensing of Environment, ISSN 1879-0704, 0034-4257, Volume 302, Page 113968, 10.1016/j.rse.2023.113968

Contributors

Oehmcke, Stefan 0000-0002-0240-1559 (Corresponding author) [1] Li, Lei (Corresponding author) [1] Trepekli, Katerina 0000-0002-9040-4409 [1] Revenga, Jaime Caballer 0000-0002-9330-6572 [1] Nord-Larsen, Thomas 0000-0002-5341-6435 [1] Gieseke, Fabian Cristian 0000-0001-7093-5803 [1] [2] Igel, Christian [1]

Affiliations

  1. [1] University of Copenhagen
    2. [NORA names: KU University of Copenhagen; University; Denmark; Europe, EU; Nordic; OECD];
  2. [2] University of Münster
    3. [NORA names: Germany; Europe, EU; OECD]

Abstract

Quantifying forest biomass stocks and their dynamics is important for implementing effective climate change mitigation measures by aiding local forest management, studying processes driving af-, re-, and deforestation, and improving the accuracy of carbon accounting. Owing to the 3-dimensional nature of forest structure, remote sensing using airborne LiDAR can be used to perform these measurements of vegetation structure at large scale. Harnessing the full dimensionality of the data, we present deep learning systems predicting wood volume and above ground biomass (AGB) directly from the full LiDAR point cloud and compare results to state-of-the-art approaches operating on basic statistics of the point clouds. For this purpose, we devise different neural network architectures for point cloud regression and evaluate them on remote sensing data of areas for which AGB estimates have been obtained from field measurements in the Danish national forest inventory. Our adaptation of Minkowski convolutional neural networks for regression give the best results. The deep neural networks produce significantly more accurate wood volume, AGB, and carbon stock estimates compared to state-of-the-art approaches. In contrast to other methods, the proposed deep learning approach does not require a digital terrain model and is robust to artifacts along the boundaries of the evaluated areas, which we demonstrate for the case where trees protrude into the area from the outside. We expect this finding to have a strong impact on LiDAR-based analyses of biomass dynamics.

Keywords

AGB, AGB estimation, Danish national forest inventory, LiDAR point clouds, LiDAR-based, LiDAR-based analysis, National Forest Inventory, Re, accounts, accuracy, accuracy of carbon accounting, adaptation, airborne LiDAR, approach, architecture, area, artifacts, biomass, biomass dynamics, biomass estimation, biomass stocks, boundaries, carbon, carbon accounting, carbon stock estimation, cases, climate change mitigation measures, cloud, compare results, convolutional neural network, data, deep learning approach, deep learning system, deep neural networks, deforestation, digital terrain model, dimensionality, dynamics, estimation, evaluation area, field, field measurements, findings, forest biomass estimation, forest biomass stocks, forest inventory, forest management, forest structure, ground, ground biomass, impact, inventory, learning approach, learning system, lidar, local forest management, management, measurements, measures of vegetation structure, method, mitigation measures, model, network, network architecture, neural network, neural network architecture, outsiders, point, point clouds, process, regression, remote sensing, remote sensing data, results, robust to artifacts, sensing, sensing data, state-of-the-art, state-of-the-art approaches, stock, stock estimates, structure, system, terrain model, trees, vegetation structure, volume, wood, wood volume

Funders

  • Danish National Research Foundation
  • European Commission
  • The Velux Foundations

Data Provider: Digital Science

LINKS
-

Matching Records in NORA

SUBJECTS
+

METRICS
+