Synthesizing Disparate LiDAR and Satellite Datasets through Deep Learning to Generate Wall-to-Wall Regional Inventories for the Complex, Mixed-Species Forests of the Eastern United States

Abstract

Light detection and ranging (LiDAR) has become a commonly-used tool for generating remotely-sensed forest inventories. However, LiDAR-derived forest inventories have remained uncommon at a regional scale due to varying parameters among LiDAR data acquisitions and the availability of sufficient calibration data. Here, we present a model using a 3-D convolutional neural network (CNN), a form of deep learning capable of scanning a LiDAR point cloud, combined with coincident satellite data (spectral, phenology, and disturbance history). We compared this approach to traditional modeling used for making forest predictions from LiDAR data (height metrics and random forest) and found that the CNN had consistently lower uncertainty. We then applied the CNN to public data over six New England states in the USA, generating maps of 14 forest attributes at a 10 m resolution over 85% of the region. Aboveground biomass estimates produced a root mean square error of 36 Mg ha−1 (44%) and were within the 97.5% confidence of independent county-level estimates for 33 of 38 or 86.8% of the counties examined. CNN predictions for stem density and percentage of conifer attributes were moderately successful, while predictions for detailed species groupings were less successful. The approach shows promise for improving the prediction of forest attributes from regional LiDAR data and for combining disparate LiDAR datasets into a common framework for large-scale estimation.

Fine-resolution landscape-scale biomass mapping using a spatiotemporal patchwork of LiDAR coverages

“…Our model accuracy against the 20% testing partition of our model dataset was favorably comparable to previous LiDAR-AGB mapping studies (Huang et al 2019;Nilsson et al 2017;Ayrey et al 2021;Hauglin et al 2021). Using a set of FIA-developed methods Menlove and Healey 2020) we further demonstrated a strong agreement between our map-based estimates and FIA-derived estimates.…”

“…Huang et al (2019) pooled by ecoregion. Both Ayrey et al (2021) and Hauglin et al (2021) pooled all coverages but used a convolutional neural network and a mixed-effects model respectively, with differing protocols for inventory plot selection.…”

Multi-Layer Modeling of Dense Vegetation from Aerial LiDAR Scans

“…Lang et al [24] estimate the canopy height at global scale from satellite LiDAR data. Ayrey et al [4] use a 3D CNN model propose alternatively to explore the potential of using finegrained 3D annotations to model the vegetation multi-layer structure using a deep network.…”

“…We can then produce height maps for all layers, which we transform into watertight meshes. These outputs are useful for downstream applications such as biomass, carbon stock, fire fuel estimation [9,13], soil illumination [19], or vegetation parameters extraction for the forest inventory [4]. The contributions of this paper are as follows:…”

A Novel Carbon Stocking Estimation Through Continuous Catalog Learning

“…In recent years, significant progress has been made in this field of machine learning-based feature extraction and AGB estimation [16,37,18]. One of the most promising approaches in this field is the use of deep supervised learning, evolving into a widely adopted approach in forestry research in general, [38,39,40], and in AGB estimation in particular [41,42]. Deep learning (DL), as a class of machine learning algorithms, gained popularity due to its ability to automatically extract features from raw data, leading to state-of-the-art performance in various fields such as computer vision, [43] requiring large amounts of labeled data.…”