Trees as Gaussians: Large-Scale Individual Tree Mapping

2025-08-29

AI Summary

Existing global tree mapping products are largely limited to binary canopy cover or mean canopy height, hindering accurate individual-tree detection. To address this, we propose a deep learning framework based on Gaussian kernel modeling, representing each tree crown as a 2D Gaussian distribution. Leveraging PlanetScope 3-m satellite imagery and airborne LiDAR data, we generate billions of self-supervised training samples, enabling, for the first time, globally scalable detection of large individual trees. Our method unifies forested and non-forested areas under a single model and outputs both crown center coordinates and binary canopy cover maps. Validation against airborne LiDAR ground truth yields an R² of 0.81 for canopy cover estimation, demonstrating strong cross-biome generalizability and compatibility with fine-tuning. This work overcomes critical modeling and scalability bottlenecks in global individual-tree monitoring.

Technology Category

• Machine Learning: Kernel Methods
• Computer Vision: Remote Sensing / Geospatial AI
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Trees are key components of the terrestrial biosphere, playing vital roles in ecosystem function, climate regulation, and the bioeconomy. However, large-scale monitoring of individual trees remains limited by inadequate modelling. Available global products have focused on binary tree cover or canopy height, which do not explicitely identify trees at individual level. In this study, we present a deep learning approach for detecting large individual trees in 3-m resolution PlanetScope imagery at a global scale. We simulate tree crowns with Gaussian kernels of scalable size, allowing the extraction of crown centers and the generation of binary tree cover maps. Training is based on billions of points automatically extracted from airborne lidar data, enabling the model to successfully identify trees both inside and outside forests. We compare against existing tree cover maps and airborne lidar with state-of-the-art performance (fractional cover R$^2 = 0.81$ against aerial lidar), report balanced detection metrics across biomes, and demonstrate how detection can be further improved through fine-tuning with manual labels. Our method offers a scalable framework for global, high-resolution tree monitoring, and is adaptable to future satellite missions offering improved imagery.

Problem

• Large-scale individual tree mapping remains limited by inadequate modeling
• Global products lack explicit identification of trees at individual level
• Deep learning approach detects individual trees in global satellite imagery

Innovation

• Deep learning for global tree detection
• Gaussian kernels simulate tree crowns
• Training with airborne lidar data