阿克苏-阿拉尔区域指阿克苏地区与阿拉尔市，位于新疆中部。作者基于Google Earth Engine (GEE) 平台，利用2020年Landsat 8、Sentinel-2和MOD13Q1遥感影像数据，采用随机森林方法对阿克苏地区与阿拉尔市棉花种植区域进行提取，并进行分类处理，得到该区域的棉花分布数据集（2020）。作者对各县域影像总体分类精度都在0.9以上，Kappa系数都在0.8以上。该数据集内容包括：（1）研究区棉花种植分布，空间分辨率为250 m；（2）样本点数据。该数据集存储为.tif、.shp格式，由17个数据文件组成，数据量为385 KB（压缩为1个文件，134 KB）。
", "ds_source": "利用 2020 年 Landsat8、Sentinel-2 和 MOD13Q1 遥感影像数据。
", "ds_process_way": "本研究所用的数据基于 GEE 平台。首先，以阿克苏-阿拉尔区域各县域耕地范围的矢量图为边界，地理坐标系为 GCS_WGS_1984，利用 2020 年棉花生长期内 Sentinel-2 和Landsat8 高空间分辨率遥感影像，通过目视解译获取各区域棉花和非棉花样本点。将获取的各县域的样本点数据存储为.shp 文件。研究采用随机森林（RF）作为分类器，因为随机森林在处理大量训练样本和高维数据时效率较高，对于训练样本的容错能力强。RF 模型由多个分类树构成。在训练 RF 模型时，使用训练样本总数的 2/3 构建每颗决策树，剩余的训练样本用于验证每颗决策树的分类结果。分类时，随机森林中的每颗决策树获得各自的分类结果，在通过最大投票法获得 RF 最终的分类结果。其中将 Sentinel2 NDVI 数据和MOD13Q1 EVI 数据作为随机森林分类的特征值。搭建随机森林分类器，按县域进行分类，得到各县域棉花种植分布。并在 GEE 平台根据 connectedPixelCount 方法去除了小斑块的影响。最终，得到了阿克苏-阿拉尔区域的棉花空间分布。
", "ds_quality": "各县域验证结果表明总体分类精度都在 0.9 以上，Kappa 系数都在 0.8 以上。其中精度最高的是温宿县，总体分类精度高达 0.99，kappa 精度高达 0.97，精度最低的是柯坪县，总体分类精度 0.94，kappa 精度为 0.83。
", "ds_acq_start_time": null, "ds_acq_end_time": null, "ds_acq_place": "", "ds_acq_lon_east": null, "ds_acq_lat_south": null, "ds_acq_lon_west": null, "ds_acq_lat_north": null, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "login-access", "ds_total_size": 1786385, "ds_files_count": 2, "ds_format": ".tif和.shp数据格式", "ds_space_res": "无", "ds_time_res": "无", "ds_coordinate": "WGS84", "ds_projection": "WGS84", "ds_thumbnail": "b1b94570-81d9-4ceb-9cba-e7d159cc16ee.jpg", "ds_thumb_from": 0, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "", "ds_from_station": null, "organization_id": "a5877b42-96ea-4f13-af7e-246f355413d6", "doi_value": "", "subject_codes": [ "170.45" ], "quality_level": 1, "publish_time": "2025-07-04 20:22:35", "first_publish_time": null, "last_updated": "2025-07-04 20:22:35", "protected": false, "protected_to": null, "lang": "zh", "cstr": null, "license": null, "extra": null, "files_shape": [ { "name": "Aksu_Alaer_Cotton_2020.rar", "size": 137549, "is_dir": false }, { "name": "Datapaper_Aksu_Alaer_Cotton_2020.pdf", "size": 1648836, "is_dir": false } ], "features": null, "data_level": 0, "i18n": { "en": { "title": "Cotton Distribution Data Set (2020) Based on Multi-source Remote Sensing Images in Aksu-Alar Region", "ds_abstract": "Aksu-Alar Region refers to Aksu Prefecture and Alar City, located in central Xinjiang. Based on the Google Earth Engine (GEE) platform, the author used 2020 Landsat 8, Sentinel-2 and MOD13Q1 remote sensing image data to extract and classify the cotton planting areas in Aksu Prefecture and Alar City using random forest method to obtain the cotton distribution data set in this area (2020). The author's overall classification accuracy of images from each county is above 0.9, and the Kappa coefficient is above 0.8. The data set includes: (1) cotton planting distribution in the study area, with a spatial resolution of 250 m;(2) sample point data. The data set is stored in.tif and.shp formats and consists of 17 data files with a data volume of 385 KB (compressed into 1 file, 134 KB).
", "ds_source": "Using 2020 Landsat8, Sentinel-2 and MOD13Q1 remote sensing image data.
", "ds_process_way": "The data used in this study is based on the GEE platform. First, taking the vector map of the cultivated land range of each county in the Aksu-Aar region as the boundary, and the geographical coordinate system is GCS_WGS_1984, using the high spatial resolution remote sensing images of Sentinel-2 and Landsat8 during the cotton growing season in 2020, cotton and non-cotton sample points in each region were obtained through visual interpretation. Store the obtained sample point data for each county as a.shp file. The random forest (RF) is used as a classifier in the study because it is efficient when processing a large number of training samples and high-dimensional data, and has strong fault tolerance for training samples. The RF model consists of multiple classification trees. When training the RF model, 2/3 of the total number of training samples are used to build each decision tree, and the remaining training samples are used to verify the classification results of each decision tree. During classification, each decision tree in the random forest obtains its own classification result, and the final classification result of RF is obtained through the maximum voting method. Among them, Sentinel2 NDVI data and MOD13Q1 EVI data are used as characteristic values of random forest classification. Build a random forest classifier and classify it by county to obtain the cotton planting distribution in each county. The impact of small plaques was removed based on the connectedPixelCount method on the GEE platform. Finally, the spatial distribution of cotton in the Aksu-Aar region was obtained.
", "ds_quality": "The verification results of each county show that the overall classification accuracy is above 0.9, and the Kappa coefficient is above 0.8. Among them, the one with the highest accuracy is Wensu County, with the overall classification accuracy as high as 0.99 and kappa accuracy as high as 0.97. The one with the lowest accuracy is Keping County, with the overall classification accuracy of 0.94 and kappa accuracy of 0.83.
", "ds_format": ".tif and.shp data formats", "ds_projection": "WGS84", "ds_space_res": "no", "ds_time_res": "no" } }, "license_type": null, "doi_reg_from": "reg_local", "cstr_reg_from": "reg_local", "doi_not_reg_reason": null, "cstr_not_reg_reason": null, "is_paper_in_submitting": false, "ds_topic_tags": [ "棉花", "随机森林" ], "ds_subject_tags": [ "地理学" ], "ds_class_tags": [], "ds_locus_tags": [ "阿克苏地区", "阿拉尔市" ], "ds_time_tags": [ 2020 ], "ds_contributors": [ "张萍", "范敬龙" ], "ds_meta_authors": [ "张萍" ], "ds_managers": [ "李锦" ], "category": "土地利用/土地覆被" }