Global Forest Mask for 2023 at 10 m Resolution from Multi-Sensor Satellite Imagery

This repository contains the global forest mask derived from the global maps of canopy height (CH) and canopy cover (CC) described by: Weber, M.; Beneke, C.; Wheeler, C. Unified Deep Learning Model for Global Prediction of Aboveground Biomass, Canopy Height, and Cover from High-Resolution, Multi-Sensor Satellite Imagery. Remote Sens. 2025, 17, 1594. https://doi.org/10.3390/rs17091594 The source variables for this derived product are available at 10.5281/zenodo.15269923. This dataset consists of GeoTIFF files covering a latitude range from 57° S to 67° N in splits of 3° x 3° per file. Each file contains a single band corresponding to the binary output of the operation fm = [(CH - CH_sd) > 5 m] & [(CC - CC_sd) > 10%] corresponding to the widely accepted definition of forest by the UN Food and Agriculture Organization (FAO) [1]. We subtract one standard deviation from the source variables, as estimated by the model, in order to compensate for the slight over-estimation at low values of CH and CC resulting in a more conservative classification of forest. Further details are given in the publication mentioned above. In addition, an alpha band is included indicating the valid pixels. Non-valid pixels are masked based on the following conditions: is water or is urban build up We recommend applying this alpha mask in addition to the data mask. The full dataset can also be retrieved from a public S3 bucket on AWS as a Requester-Pays service. Note that no transfer costs are incurred if downloading to an AWS resource within the same region (us-west-2). For further details on data transfer costs we refer to the AWS documentation. We encourage users to create their own AWS account (if not already existing) and transfer individual files within the same region by: aws s3 cp s3://eda-appsci-open-access/forestmask/2023/earthdaily_forestmask_{lon}_{lat}-[forest, alpha].tif DESTINATION_PATH --request-payer requester or the full dataset by: aws s3 sync s3://eda-appsci-open-access/forestmask/2023/ DESTINATION_PATH --request-payer requester A complete list of files avaialable in the S3 bucket is provided by filelist.txt. [1] Food and Agriculture Organization. (2000, November 2). FRA 2000 on definitions of forest and forest change (FRA Working Paper No. 33). Forest Resources Assessment Programme. Rome. Retrieved from FAO website https://www.fao.org/4/ad665e/ad665e00.htm本仓库包含基于冠层高度（CH, Canopy Height）和冠层覆盖度（CC, Canopy Cover）全球地图生成的全球森林掩膜数据，相关说明如下： Weber, M.; Beneke, C.; Wheeler, C. 提出的统一深度学习模型，用于从高分辨率多传感器卫星影像全球预测地上生物量、冠层高度与覆盖度[发表于《遥感》2025年，第17卷，第1594号文章，https://doi.org/10.3390/rs17091594]。 该衍生数据集的源变量可通过10.5281/zenodo.15269923获取。本数据集由GeoTIFF文件组成，覆盖南纬57°至北纬67°的范围，每个文件包含3°×3°的分幅数据。每个文件仅包含单波段数据，对应如下运算的二值输出： fm = [(CH - CH_sd) > 5 m] & [(CC - CC_sd) > 10%] 该结果符合联合国粮食及农业组织（Food and Agriculture Organization, FAO）[1]广泛认可的森林定义。为补偿模型在冠层高度和冠层覆盖度低值区域的轻微高估问题，我们对源变量减去其模型估计得到的一个标准差，以得到更为保守的森林分类结果。更多细节详见上述发表的论文。 此外，数据中包含alpha波段用于标记有效像素。无效像素将基于以下条件进行掩膜： 1. 为水体区域 2. 为城市建成区 我们建议除数据自带的掩膜外，额外使用该alpha掩膜。 完整数据集也可通过AWS的公共S3存储桶以"请求者付费"（Requester-Pays）服务获取。请注意，若在同一区域（us-west-2）内的AWS资源上下载数据，则不会产生传输费用。有关数据传输费用的更多详情，请参考AWS官方文档。我们建议用户创建个人AWS账户（若无账户），并通过以下命令在同一区域内传输单个文件： aws s3 cp s3://eda-appsci-open-access/forestmask/2023/earthdaily_forestmask_{lon}_{lat}-[forest, alpha].tif DESTINATION_PATH --request-payer requester 或传输完整数据集： aws s3 sync s3://eda-appsci-open-access/forestmask/2023/ DESTINATION_PATH --request-payer requester S3存储桶中可用的所有文件列表可通过filelist.txt获取。 [1] 联合国粮食及农业组织. 2000年11月2日发布的《FRA 2000森林与森林变化定义》（FRA工作论文第33号）. 森林资源评估计划. 罗马. 检索自FAO官网 https://www.fao.org/4/ad665e/ad665e00.htm