2019年全国农业塑料大棚遥感分类数据集

本数据集为依托Google Earth Engine（GEE）云计算平台，基于Sentinel-2遥感影像和随机森林模型，生成的30m空间分辨率2019年全国农业塑料大棚空间分布专题数据。首先通过野外调查和目视解译进行地面样本采集，并随机划分为训练集和测试集；然后进行地物光谱、纹理等特征提取，从而构建多维特征空间；最后构建随机森林分类模型，并利用训练好的模型对全国遥感影像进行并行计算，从而得到全国农业塑料大棚分类数据。精度测试表明，该数据集的平均分类精度为87.45%，能够正确反映农业塑料大棚在全国的空间分布情况。本数据集为第一个公开发布的全国农业塑料大棚空间分布专题数据，可为相关领域的科研人员提供数据参考。该数据集2024、2025连续2年入选“中国高影响力数据论文”。

This dataset is a thematic spatial distribution dataset of agricultural plastic greenhouses across China in 2019 with 30 m spatial resolution, generated based on Sentinel-2 remote sensing images and the random forest model via the Google Earth Engine (GEE) cloud computing platform. First, ground samples were collected through field surveys and visual interpretation, and randomly divided into training and test sets; then, spectral and textural features of ground objects were extracted to construct a multi-dimensional feature space; finally, a random forest classification model was built, and parallel computing was performed on the national remote sensing images using the trained model to obtain the classified data of agricultural plastic greenhouses across China. Accuracy tests show that the average classification accuracy of this dataset is 87.45%, which can accurately reflect the spatial distribution of agricultural plastic greenhouses nationwide. This dataset is the first publicly released thematic spatial distribution dataset of national agricultural plastic greenhouses, providing data references for researchers in related fields. This dataset has been selected for China's High-Impact Data Papers for two consecutive years, 2024 and 2025.

搜集汇总

背景与挑战

背景概述

该数据集是基于2019年Sentinel-2遥感影像，利用Google Earth Engine平台和随机森林模型生成的全国农业塑料大棚空间分布专题数据，空间分辨率为30米。数据生成过程结合实地调查和视觉解译样本，平均分类精度达87.45%，能准确反映全国大棚分布。作为首个公开发布的此类数据集，它连续两年被评为'中国高影响力数据论文'，为相关领域科研人员提供重要数据参考。