Pilot Dataset for UK Dynamic World Land-Use Modelling with GLDAS Climate Features

This archive contains a pilot dataset for a UK land-use modelling workflow using Google Dynamic World and GLDAS climate data. The dataset includes sampled 1 km Dynamic World GeoTIFF patches across urban, middle, and rural land-use strata for 2015-2025, a flattened CSV representation for spatial/CNN-style experiments, and a compact feature table for Multiple Linear Regression and Random Forest baselines. The compact table includes Dynamic World land-cover fractions and annual mean temperature features where available. This pilot dataset is intended to support early-stage development and testing of machine-learning workflows for land-use, urban heat, and climate-adaptation modelling.本数据集为依托谷歌动态世界（Google Dynamic World）与GLDAS气候数据构建的英国土地利用建模工作流试点数据集。数据集包含2015-2025年间覆盖城市、中等、农村三类土地利用层级的1km分辨率谷歌动态世界GeoTIFF采样斑块；同时提供适用于空间/卷积神经网络（Convolutional Neural Network, CNN）类实验的扁平化CSV格式数据，以及用于多元线性回归与随机森林基准测试的精简特征表。该精简特征表在数据可用的前提下，包含谷歌动态世界的土地覆盖占比与年平均气温特征。本试点数据集旨在支撑土地利用、城市热环境与气候适应建模相关机器学习工作流的早期开发与测试工作。