2019年全国农业塑料大棚遥感分类数据集

A dataset of remote sensing-based classification for agricultural plastic greenhouses in China in 2019

本数据集为依托Google Earth Engine(GEE)云计算平台,基于10米空间分辨率的Sentinel-2遥感影像和随机森林模型,所生成的2019年全国农业塑料大棚空间分布专题数据。具体而言,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。此外,为了更好对全国大棚分布数据进行可视化,本文同时计算了全国5公里格网内的大棚面积占比。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。

This paper provides a dataset of remote sensing-based classification for agricultural plastic greenhouses in China in 2019.Based on Google Earth Engine(GEE)cloud computing platform,the dataset is derived from the Sentinel-2 remote sensing images and random forest classification model with a spatial resolution of 10 meters.Specifically,we collected ground-truth samples through field survey and visual interpretation,and then divided the samples into a training set and a test set.Then,we carried out feature extraction,such as the extraction of spectral and texture,so as to construct a multi-dimensional feature space.Finally,we utilized the trained random forest model to classify national-scale remote sensing images through parallel computing to acquire the spatial distribution of China’s agricultural plastic greenhouses.The accuracy evaluation shows an average classification accuracy of 87.45%,which indicates that the proposed dataset can accurately reflect the spatial distribution of agricultural plastic greenhouses across China.In addition,in order to better visualize the nation-wide greenhouse distribution data,in this paper we calculated the proportion of the greenhouse area in the 5km grid as well.Above all,this is the first dataset with publicly released thematic data on the spatial distribution of China’s agricultural plastic greenhouses.It can be used for data references by scientific researchers in related fields.