OpenspaceGlobal_全球169个特大城市城市开放空间数据集

城市开放空间（Urban Open Space, UOS）在人类社会与经济活动中扮演着重要角色，但全球主要城市仍缺乏系统的UOS测绘数据。为此，本研究采用深度学习驱动的小样本人工标注技术，绘制了全球169个特大城市（常住人口≥300万）建成区2021年度1.19米分辨率的UOS空间分布图（即OpenspaceGlobal数据集）。该数据集包含五大类城市开放空间：1.公园与绿地2.户外运动场所3.交通空间4.水体空间5.背景区域
OpenspaceGlobal数据集有助于深入解析全球特大城市人工地表空间格局，支撑城市宜居性评估、空间规划优化及可持续发展决策等应用。本数据集遵循非商业用途共享协议，允许在规范引用前提下免费用于科研与科普传播。

Urban Open Space (UOS) plays a pivotal role in human social and economic activities. However, systematic UOS mapping data remains unavailable for major cities across the globe.

To address this research gap, this study employed deep learning-driven few-shot manual annotation techniques to generate a 1.19-meter resolution spatial distribution map of UOS in the built-up areas of 169 megacities (with permanent population ≥ 3 million) worldwide in 2021, namely the OpenspaceGlobal dataset.

This dataset covers five major categories of urban open spaces: 1. Parks and green spaces; 2. Outdoor sports venues; 3. Transportation spaces; 4. Water bodies; 5. Background areas.

The OpenspaceGlobal dataset facilitates in-depth analysis of the spatial patterns of anthropogenic surfaces in global megacities, supporting applications such as urban livability assessment, spatial planning optimization, and sustainable development decision-making.

This dataset follows a non-commercial use sharing agreement, allowing free use for scientific research and science popularization under the premise of standardized citation.

搜集汇总

背景与挑战

背景概述

OpenspaceGlobal是全球首个针对169个特大城市（常住人口≥300万）的城市开放空间数据集，发布于2021年，提供1.19米高分辨率栅格数据，涵盖公园绿地、户外运动场所等五类开放空间。该数据集基于深度学习和小样本标注技术生成，旨在支持城市宜居性评估、空间规划优化等应用，遵循非商业共享协议供科研使用。