| 引用本文: | 张亮亮,包康,徐大伟,闫瑞瑞,刘贵河,刘贵波,刘忠宽,王国良,孙娟,史莹华,辛晓平.基于多时相Landsat 8 OLI的苜蓿高精度提取方法研究[J].中国农业资源与区划,2026,47(3):164~174 |
| |
|
| |
|
|
| 本文已被：浏览 388次 下载 11次 |
码上扫一扫！ |
| 基于多时相Landsat 8 OLI的苜蓿高精度提取方法研究 |
|
张亮亮1,2,3,4，包康2,3,4，徐大伟2,3,4，闫瑞瑞2,3,4，刘贵河5，刘贵波6,7，刘忠宽8,9，王国良10,11，孙娟12，史莹华13，辛晓平2,3,4
|
|
1成都理工大学地理与规划学院，四川成都 610059;2北方干旱半干旱耕地高效利用全国重点实验室，北京 100081;3呼伦贝尔草原生态系统国家野外科学观测研究站，内蒙古呼伦贝尔 021000;4中国农业科学院农业资源与农业区划研究所，北京 100081;5河北北方学院动物科技学院，张家口 075000;6河北省农林科学院旱作农业研究所，衡水 053000;7国家牧草产业技术体系衡水综合试验站，河北衡水 053000;8河北省农林科学院农业资源环境研究所，石家庄 050051;9国家牧草产业技术体系沧州综合试验站，河北石家庄 050051;10山东省农业科学院休闲农业研究所，济南 250100;11国家牧草产业技术体系东营综合试验站，山东济南 250100;12青岛农业大学草业学院，山东青岛 266109;13河南农业大学动物科技学院，郑州 450046
|
|
| 摘要: |
| 目的 苜蓿作为人工草地中重要的优质饲草，已被纳入国家粮食安全战略的重要参考指标，在畜牧业可持续发展和粮食安全保障体系中发挥着不可替代的作用。及时准确地获取苜蓿种植面积及空间分布，分析其动态变化，为苜蓿产业的科学管理及发展提供理论依据。方法 文章研究以华北平原（河北、河南、山东）为例，利用2022年3月1日至11月1日期间的16景Landsat 8 OLI遥感图像，计算归一化植被指数（NDVI），构建时间序列NDVI数据。结合苜蓿与其他农作物的物候特征地理差异，采用NDVI灰度级阈值分类法和随机森林分类法进行苜蓿人工草地信息提取，并通过实地调研数据验证结果精度。结果 （1）2022年河北省苜蓿面积315.70 km2，占华北平原苜蓿总面积的69.44%，主要集中在中部和南部地区，尤其是平原和河谷地带的潮土、砂壤土、淤土及盐碱土地区。（2）河南省苜蓿面积70.82 km2，95%以上的苜蓿分布于黄河滩岸的冲积土、淤土地区。（3）山东省苜蓿面积68.14 km2，零星状分布于沙壤土地区。（4）NDVI灰度级阈值分类法和随机森林分类法的精度分别为93.48%和97.83%。结论 利用多时相Landsat 8 OLI图像，采取NDVI灰度级阈值分类法和随机森林分类法，并以土地利用数据和实地调研数据作为辅助，可以实现对苜蓿人工草地的准确提取。 |
| 关键词: 苜蓿 遥感 阈值法 随机森林分类 精度评价 |
| DOI：10.7621/cjarrp.1005-9121.20260313 |
| 分类号:P642.5 |
| 基金项目:现代农业产业技术体系建设专项资金（CARS-34）；国家重点研发计划“天然草原智能放牧与草畜精准管控关键技术”（2021YFD1300500）；国家自然科学基金“呼伦贝尔草原生态系统碳水耦合及其放牧响应机制”（32130070） |
|
| HIGH-PRECISION EXTRACTION OF ALFALFA BASED ON MULTI-TEMPORAL LANDSAT 8 OLI IMAGERY |
|
Zhang Liangliang1,2,3,4, Bao Kang2,3,4, Xu Dawei2,3,4, Yan Ruirui2,3,4, Liu Guihe5, Liu Guibo6,7, Liu Zhongkuan8,9, Wang Guoliang10,11, Sun Juan12, Shi Yinghua13, Xin Xiaoping2,3,4
|
|
1School of Geography and Planning, Chengdu University of Technology, Chengdu 610059, Sichuan, China;2State Key Laboratory of Efficient Utilization of Arable Land in China, Beijing 100081, China;3National Hulunber Grassland Ecosystem Observation and Research Station, HulunBuir 021000, Inner Mongolia, China;4Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China;5College of Animal Science and Technology, Hebei North University, Zhangjiakou 075000, Hebei, China;6Dry-Land Farming Institute of Hebei Academy of Agricultural and Forestry Sciences, Hengshui 053000, Hebei, China;7Hengshui Comprehensive Experimental Station of National Forage Industry Technology System, Hengshui 053000, Hebei, China;8Institute of Agricultural Resources and Environment, Hebei Academy of Agricultural and Forestry Sciences, Shijiazhuang 050051, Hebei, China;9Cangzhou Comprehensive Experimental Station of National Forage Industry Technology System, Shijiazhuang 050051, Hebei, China;10Institute of Leisure Agriculture, Shandong Academy of Agricultural Sciences, Jinan 250100, Shandong, China;11Dongying Comprehensive Experimental Station of National Forage Industry Technology System, Jinan 250100, Shandong, China;12College of Grassland Sciences, Qingdao Agricultural University, Qingdao 266109, Shandong, China;13College of Animal Science and Technology, Henan Agricultural University, Zhengzhou 450046, Henan, China
|
| Abstract: |
| As a premium forage in artificial grasslands, alfalfa has been integrated into the critical reference indicators of the national food security strategy. It plays an indispensable role in the sustainable development of animal husbandry and the food security system. To obtain timely and accurate information on the planting area and spatial distribution of alfalfa, analyze its dynamic changes, and provide a theoretical basis for the scientific management and development of the alfalfa industry, this study utilized 16 Landsat 8 OLI remote sensing images from March 1 to November 1, 2022, covering the North China Plain (Hebei, Henan, and Shandong). A Normalized Difference Vegetation Index (NDVI) time-series dataset was constructed. And based on the geographical differences in the phenological characteristics of alfalfa compared to other crops, alfalfa artificial grassland information was extracted using NDVI gray-level threshold classification and random forest classification, and the accuracy of the results was verified through field investigation data. The results were presented as follows. (1) The alfalfa planting area in Hebei province in 2022 reached 315.70 km2, accounting for 69.44% of the total alfalfa area in the North China Plain. The alfalfa was primarily concentrated in the central and southern regions, particularly in the plains and valleys with tidal soil, sandy loam, warp soil and saline-alkali soils. In Henan province, alfalfa covered an area of 70.82 km2, with more than 95% distributed in the alluvial soil and warp soil areas along the Yellow River banks. In Shandong province, alfalfa covered an area of 68.14 km2, and was scattered across sandy loam regions. The accuracy of the NDVI gray-level threshold classification and random forest classification was 93.48% and 97.83%, respectively. In summary, this study demonstrates that combining multi-temporal Landsat 8 OLI imagery with NDVI gray-level threshold classification and random forest classification provides more accurate extraction results for alfalfa distribution. |
| Key words: alfalfa remote sensing threshold method random forest classification precision evaluation |
|
|
|
|
|
|
