安徽农学通报 >
2025 , Vol. 31 >Issue 4: 113 - 118
DOI: https://doi.org/10.16377/j.cnki.issn1007-7731.2025.04.023
基于随机森林特征的岳西县茶园信息提取
|
牛海庆(2003—),男,安徽定远人,从事资源环境遥感研究。 |
Copy editor: 胡立萍
收稿日期: 2024-10-22
网络出版日期: 2025-02-28
基金资助
宿州学院省级/国家级大学生创新创业训练计划项目(S202310379136)
宿州学院省级/国家级大学生创新创业训练计划项目(202310379042)
皖北3S技术应用研究中心(2021XJPT12)
安徽省合作实践教育基地项目(2023xqhz065)
安徽省高等学校自然科学研究重点项目(2024AH051818)
宿州学院科研发展基金项目(2021fzjj23)
宿州学院质量工程结余经费项目(szxy2023jyjf61)
Information extraction of tea garden in Yuexi County based on random forest features
Received date: 2024-10-22
Online published: 2025-02-28
本研究以安徽岳西县为研究区,多时相(10、12和5月)的Landsat 8/9 OLI影像为数据源,利用精确度评价证明随机森林算法的可行性,利用ENVI 5.3和ArcGIS 10.8软件分析7种典型地物(耕地、水体、建筑物、灌木、道路、林地和茶园)的时序光谱特征与NDVI变化,然后利用随机森林算法提取研究茶园信息进行茶园时空特征变化分析。结果表明,随机森林算法提取茶园信息效果较好,精度较高。2022年10月、2022年12月和2023年5月随机森林法提取茶园信息的总体精度分别为99.3%、98.2%和99.5%,Kappa系数分别为0.972、0.934和0.981;NIR、SWIR1和SWIR2是茶园的特征波段,茶园的NDVI值高于其他植被的NDVI值,可作为茶园与其他植被区分的重要节点。随机森林法结果表明,研究区2022年10月、2022年12月和2023年5月茶园种植面积分别为130.59、126.79和137.47 km2,其时空特征变化明显。综上,随机森林算法提取茶园信息效果较好,研究为随机森林算法在茶园信息提取中的应用提供参考。
关键词: 茶园信息提取; Landsat 8/9 OLI; 光谱特征; 随机森林算法
牛海庆 , 陈丽 , 刘爽 , 方刚 . 基于随机森林特征的岳西县茶园信息提取[J]. 安徽农学通报, 2025 , 31(4) : 113 -118 . DOI: 10.16377/j.cnki.issn1007-7731.2025.04.023
Yuexi County of Anhui Province was selected as the research area, and multi-temporal Landsat 8/9 OLI images (October, December and May) were used as the data source to prove the feasibility of the random forest algorithm by accuracy evaluation. ENVI 5.3 and ArcGIS 10.8 software were used to analyze the temporal spectral features and NDVI changes of 7 typical land features (cultivated land, water body, buildings, shrub, road, forest land, and tea garden), and then random forest algorithm was used to extract tea garden information in research area to analyze the spatio-temporal characteristics changes of tea gardens. The results showed that the random forest algorithm had good effect and high precision in extracting tea garden information in research area. The overall accuracy of tea garden information extraction by random forest method in October 2022, December 2022, and May 2023 were 99.3%, 98.2%, and 99.5%, respectively, and the Kappa coefficients were 0.972, 0.934, and 0.981, respectively. NIR, SWIR1 and SWIR2 were the characteristic bands of tea gardens, and the NDVI value of tea gardens was higher than that of other vegetation, which can be used as an important node to distinguish tea gardens from other vegetation. Random forest method calculation showed that the planting area of tea gardens in research area in October 2022, December 2022, and May 2023 was 130.59, 126.79, and 137.47 km2, respectively, and their spatio-temporal characteristics changed significantly. In summary, random forest algorithm has a good effect on extracting tea garden information, and this study provides a reference for the application of random forest algorithm in tea garden information extraction.
| 1 |
蒋永泉.基于多源遥感数据茶园识别及其识别精度尺度效应研究[D]. 昆明:云南师范大学,2021.
|
| 2 |
侯莹坤.作为日常生活实践的茶:从茶课与社会记忆探究茶的文化意义[J].遗产,2024(1):335-349,376.
|
| 3 |
孔玉屏. 我国茶产业增长特点的实证研究[D]. 杭州:浙江大学,2006.
|
| 4 |
袁观杰. 基于后向散射时空特征的夏玉米种植范围提取研究:以河南省为例[D]. 郑州:郑州大学,2021.
|
| 5 |
梁晨欣,黄启厅,王思,等. 基于多时相遥感植被指数的柑橘果园识别[J]. 农业工程学报,2021,37(24):168-176.
|
| 6 |
王斌,何丙辉,林娜,等. 基于随机森林特征选择的茶园遥感提取[J]. 吉林大学学报(工学版),2022,52(7):1719-1732.
|
| 7 |
钱瑞,徐伟恒,黄邵东,等. 应用GF-5高光谱遥感影像提取山区茶园[J]. 光谱学与光谱分析,2023,43(11):3591-3598.
|
| 8 |
李艳,张帆. 基于Sentinel-2影像的厦门市茶园遥感提取[J]. 亚热带植物科学,2023,52(4):327-335.
|
| 9 |
林卉,陈杰,王长海. 遥感原理与应用[M]. 徐州:中国矿业大学出版社,2022:296-297.
|
| 10 |
|
| 11 |
赵晓晴,王萍,荆林海,等. 时序Sentinel-2A影像光谱特征的茶园提取应用[J]. 测绘科学,2020,45(6):80-88.
|
| 12 |
张修远,刘修国. 基于随机森林算法的高维模糊分类研究[J]. 国土资源遥感,2014,26(2):87-92.
|
| 13 |
方匡南,吴见彬,朱建平,等. 随机森林方法研究综述[J]. 统计与信息论坛,2011,26(3):32-38.
|
| 14 |
黄艳红. 基于多源遥感数据的茶园面积提取[D]. 杭州:浙江大学,2020.
|
/
| 〈 |
|
〉 |