NEON Tree Species Predictions

Individual Tree Predictions for 80 million trees in the National Ecological Observatory Network For site abbreviations see: https://www.neonscience.org/field-sites/explore-field-sites For each site, there is a .zip and .csv. The .zip is a set 1km .shp tiles. The .csv is all trees in a single file. Please see the manuscript for detailed methods. Summary We use the DeepForest python package to predict individual crown location in the RGB camera mosaic (Weinstein et al. 2020a). Tree crowns with less than 3m maximum height in the LiDAR derived canopy height model are removed. At this stage in the workflow each individual tree has a unique ID, predicted crown location, crown area and confidence score from the DeepForest tree detection model. Following individual tree detection, we classify each individual as Alive or Dead based on the appearance in the RGB data. Since NEON captures airborne data during the leaf-on season, any standing tree with no leaf cover was annotated as 'dead'. During prediction, the location of each predicted crown is cropped and passed to the Alive-Dead model for labeling as each Alive (0) or Dead (1) with a confidence score for each class. To classify each tree crown to species we use the multi-temporal hierarchical model in Weinstein et al. 2023. Using the best trained model for each site we predict all available areas within the NEON AOP footprint that have overlapping RGB data for crown prediction and hyperspectral data for species prediction. The predicted species label confidence score, as well labels from the higher levels are included in the shapefile. Column Name Definition Geometry A four pointed bounding box location in utm coordinates. indiv_id A unique crown identifier that combines the year, site and geoindex of the NEON airborne tile (e.g. 732000_4707000) is the utm coordinate of the top left of the tile. sci_name The full latin name of predicted species aligned with NEON's taxonomic nomenclature. ens_score The confidence score of the species prediction. This score is the output of the multi-temporal model for the ensemble hierarchical model. bleaf_taxa Highest predicted category for the broadleaf model bleaf_score The confidence score for the broadleaf taxa submodel oak_taxa Highest predicted category for the oak model dead_label A two class alive/dead classification based on the RGB data. 0=Alive/1=Dead. dead_score The confidence score of the Alive/Dead prediction. site_id The four letter code for the NEON site. See https://www.neonscience.org/field-sites/explore-field-sites for site locations. conif_taxa Highest predicted category for the conifer model conif_score The confidence score for the conifer taxa submodel dom_taxa Highest predicted category for the dominant taxa mode submodel dom_score The confidence score for the dominant taxa submodel

针对国家生态观测站网络（National Ecological Observatory Network, NEON）内8000万棵树木的单木预测数据集。站点缩写查询方式详见：https://www.neonscience.org/field-sites/explore-field-sites。每个站点对应一个.zip压缩包与一个.csv数据文件：.zip压缩包内含多幅1km分辨率的.shp矢量瓦片；.csv文件为整合所有树木数据的单文件。详细研究方法请参阅相关手稿。

研究概述
本研究借助DeepForest Python工具包，对RGB相机拼接影像中的单木冠幅位置进行预测（Weinstein等人，2020a）。我们将移除激光雷达（LiDAR）反演冠层高度模型中最大高度低于3米的树冠。在此流程阶段，每棵单木均拥有唯一标识符、预测冠幅位置、冠幅面积以及来自DeepForest树木检测模型的置信度得分。
完成单木检测后，我们基于RGB影像的外观特征将每棵单木划分为存活或死亡两类。由于NEON的机载数据采集于叶片生长期，所有无叶片覆盖的立木均被标注为死亡。预测阶段中，每个预测冠幅的区域会被裁剪并输入存活-死亡分类模型，以0代表存活、1代表死亡的方式生成类别标签，并输出每个类别的置信度得分。
为完成树木冠幅的物种分类任务，我们采用Weinstein等人2023年提出的多时相层级分类模型。针对每个站点使用最优训练得到的模型，对NEON机载观测平台（Airborne Observation Platform, AOP）覆盖范围内所有同时具备可用于冠幅预测的重叠RGB影像，以及可用于物种预测的高光谱数据的区域开展预测。该矢量瓦片文件中包含预测物种标签的置信度得分，以及层级模型各上层分类的标签。

字段定义
Geometry：以UTM坐标系表示的四角边界框位置。
indiv_id：唯一冠幅标识符，由NEON机载瓦片的年份、站点与地理索引组合而成（例如732000_4707000为该瓦片左上角的UTM坐标）。
sci_name：与NEON分类命名体系对齐的预测物种完整拉丁学名。
ens_score：物种预测的置信度得分，该值为多时相层级集成模型的输出结果。
bleaf_taxa：阔叶树分类子模型的最高预测类别。
bleaf_score：阔叶树分类子模型的置信度得分。
oak_taxa：栎树分类子模型的最高预测类别。
dead_label：基于RGB影像的二分类存活/死亡标签，0代表存活、1代表死亡。
dead_score：存活/死亡分类预测的置信度得分。
site_id：NEON站点的四字母代码，站点位置查询详见https://www.neonscience.org/field-sites/explore-field-sites。
conif_taxa：针叶树分类子模型的最高预测类别。
conif_score：针叶树分类子模型的置信度得分。
dom_taxa：优势类群模式子模型的最高预测类别。
dom_score：优势类群模式子模型的置信度得分。

搜集汇总

背景与挑战

背景概述

该数据集包含美国国家生态观测网络8000万棵树的预测数据，使用深度学习模型预测每棵树的树种、存活状态及相应置信度，数据格式包括shapefile和CSV，涵盖树木位置、树种信息等多维度数据。