MRA: Disentangling cross-scale influences on tree species, traits, and diversity from individual trees to continental scales

MRA：理清从个体树木到大陆尺度对树种、性状和多样性的跨尺度影响

• 批准号: 1926542
• 负责人: Ethan White
• 金额: $ 121.52万
• 依托单位: University of Florida
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-01 至 2024-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1926542&HistoricalAwards=false
• 关键词: MRA; Disentangling; cross; scale; influences

Trees are essential to ecosystems. They store carbon, reduce erosion, and serve as habitat for other species. The factors influencing trees, and the spatial scales at which they are managed, range from an individual tree to entire continents. Since there are approximately three trillion trees in the world collecting data on every tree over large areas is impossible using traditional methods. Therefore, it is necessary to use new technology to measure and describe individual trees over large geographic areas. This research will address this fundamental challenge by combining high resolution remote sensing data with field data on trees. Together, the remote sensing and field data will be used to understand what influences the number of trees, their size, where different species occur, and how this changes from spatial scales of local parks to the entire United States. This project will also make it easier for other scientists to study trees over large areas by developing software, producing data products, and providing training and collaboration opportunities for working with these novel datasets. This will help drive rapid advances in the cross-scale understanding of tree ecology with broad applications in forestry, management, and fundamental scientific understanding.This project combines National Ecological Observatory Network (NEON) data from airborne remote sensing and field data collection. These data will be used to develop machine learning based approaches to identify, measure, and characterize to species all of the canopy trees located within each forested NEON site. This will yield data on approximately 50 million individual trees at about 40 sites across the United States. These data from NEON will be combined with data from the US Forest Service Forest Inventory and Analysis Project, which samples millions of trees at over 100,000 locations across the United States. These combined data will be used to develop joint models of the distribution, abundance, and structural traits of trees, that explicitly incorporate the concept of scale. These models will be used to understand how the processes influencing tree distribution and traits change across scales by comparing the importance of different factors at scales ranging from a few meters, where individual trees directly interact, to the entire United States, where large gradients in climate and land use are important. This research will address three broad questions in ecology: 1) what processes govern species distribution and abundance at different scales and how do they interact? 2) how are landscape and regional process of species coexistence connected to local biodiversity? 3) how do changes in the processes influencing tree traits across scales impact estimates of biomass and carbon storage?This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

树木对于生态系统至关重要。它们储存碳，减少侵蚀，并作为其他物种的栖息地。影响树木的因素及其管理的空间尺度范围从一棵树到整个大陆。由于世界上大约有三万亿棵树，使用传统方法不可能收集大面积每棵树的数据。因此，有必要使用新技术来测量和描述大地理区域内的个体树木。这项研究将通过将高分辨率遥感数据与树木的实地数据相结合来解决这一基本挑战。遥感和现场数据将共同用于了解影响树木数量、大小、不同物种出现的地点，以及这种影响从当地公园的空间尺度到整个美国的变化。该项目还将通过开发软件、生产数据产品以及提供使用这些新颖数据集的培训和协作机会，使其他科学家更容易地研究大面积的树木。这将有助于推动对树木生态学的跨尺度理解的快速进步，并在林业、管理和基础科学理解方面具有广泛的应用。该项目结合了来自机载遥感和现场数据收集的国家生态观测网络（NEON）数据。这些数据将用于开发基于机器学习的方法，以识别、测量和表征位于每个 NEON 森林站点内的所有树冠树木的物种。这将产生美国约 40 个地点约 5000 万棵树的数据。来自 NEON 的这些数据将与美国林务局森林清查和分析项目的数据相结合，该项目对美国 100,000 多个地点的数百万棵树木进行了采样。这些组合数据将用于开发树木的分布、丰度和结构特征的联合模型，该模型明确纳入了规模的概念。这些模型将用于了解影响树木分布和性状的过程如何在不同尺度上变化，通过比较不同因素的重要性，范围从几米（个体树木直接相互作用）到整个美国（气候梯度较大）和土地利用很重要。这项研究将解决生态学中的三个广泛问题：1）哪些过程控制着不同尺度的物种分布和丰度以及它们如何相互作用？ 2）物种共存的景观和区域过程如何与当地生物多样性相关？ 3) 跨尺度影响树木性状的过程变化如何影响生物量和碳储存的估计？该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Training Data for the NeonTreeEvaluation Benchmark

NeonTreeEvaluation 基准的训练数据

• DOI: 10.5281/zenodo.3459803
• 发表时间: 2020-01
• 期刊: Zenodo
• 作者: Weinstein, Ben;Marconi, Sergio;White, Ethan
• 通讯作者: White, Ethan

DeepForest: A Python package for RGB deep learning tree crown delineation

DeepForest：用于 RGB 深度学习树冠描绘的 Python 包

• DOI: 10.1111/2041-210x.13472
• 发表时间: 2020-07-08
• 期刊: bioRxiv
• 作者: Ben. G. Weinstein;S. Marconi;M. Aubry;Grégoire Laurent Vincent;Henry Senyondo;E. White
• 通讯作者: E. White

Injecting Domain Knowledge Into Deep Neural Networks for Tree Crown Delineation

将领域知识注入深度神经网络以进行树冠描绘

• DOI: 10.1109/tgrs.2022.3216622
• 发表时间: 2024-09-14
• 期刊: IEEE Transactions on Geoscience and Remote Sensing
• 影响因子: 8.2
• 作者: Ira Harmon;S. Marconi;Ben Weinstein;Sarah J. Graves;D. Wang;A. Zare;Stephanie A. Bohlman;Aditya Singh;Ethan White
• 通讯作者: Ethan White

Estimating individual‐level plant traits at scale

大规模估计个体水平的植物性状

• DOI: 10.1002/eap.2300
• 发表时间: 2021-06
• 期刊: Ecological Applications
• 影响因子: 5
• 作者: Marconi, Sergio;Graves, Sarah J.;Weinstein, Ben G.;Bohlman, Stephanie;White, Ethan P.
• 通讯作者: White, Ethan P.

Capturing long‐tailed individual tree diversity using an airborne imaging and a multi‐temporal hierarchical model

使用机载成像和多时态分层模型捕获长尾个体树木多样性

• DOI: 10.1002/rse2.335
• 发表时间: 2023-05
• 期刊: Remote Sensing in Ecology and Conservation
• 影响因子: 5.5
• 作者: Weinstein, Ben G.;Marconi, Sergio;Graves, Sarah J.;Zare, Alina;Singh, Aditya;Bohlman, Stephanie A.;Magee, Lukas;Johnson, Daniel J.;Townsend, Phillip A.;White, Ethan P.
• 通讯作者: White, Ethan P.