English

应用科学学报 >

2022 , Vol. 40 >Issue 6: 941 - 952

DOI: https://doi.org/10.3969/j.issn.0255-8297.2022.06.005

信号与信息处理

基于半监督回归的高光谱土壤重金属质量浓度反演

展开
  • 1. 中南大学 地球科学与信息物理学院, 湖南 长沙 410083;
    2. 中南大学 有色金属成矿预测与地质环境监测教育部重点实验室, 湖南 长沙 410083;
    3. 湖南省科学技术信息研究所, 湖南 长沙 410001

收稿日期: 2022-03-16

  网络出版日期: 2022-12-03

基金资助

国家重点研发计划(No.2018YFB0504500);内蒙古自治区科技计划(No.2022YFSJ0014)资助

收起

Hyperspectral Inversion of Soil Heavy Metal Mass Concentration Based on Semi-supervised Regression

Expand
  • 1. School of Geoscience and Info-physics, Center South University, Changsha 410083, Hunan, China;
    2. Key Laboratory of Metallogenic Prediction of Nonferrous Metals and Geological Environment Monitoring, Ministry of Education, Center South University, Changsha 410083, Hunan, China;
    3. Hunan Institute of Science and Technology Information, Changsha 410001, Hunan, China

Received date: 2022-03-16

  Online published: 2022-12-03

Fold

摘要

针对如何利用少量有标记样本和大量无标记样本训练出鲁棒性的土壤重金属质量浓度反演模型的问题，以土壤中重金属镉（Cd）为研究对象，选取4个不同地区（衡阳-郴州，原平-保定）的光谱数据分两组进行实验验证。在通过迁移成分分析方法缩小不同区域的光谱分布差异后，提出一种基于半监督回归的高光谱土壤重金属质量浓度反演模型。实验结果显示，与传统的全监督建模方法相比，在第1组衡阳-郴州的实验中，所提的半监督方法能够将可决系数R2提升至0.75，相对分析误差(relative predictive deviation,RPD)提升至2.15；在第2组原平-保定的实验中，R2提升至0.70，RPD提升至1.61。实验表明，在较少标记样本情况下，通过引入大量的未标记样本进行半监督回归分析可有效提升模型反演精度。

关键词： 高光谱遥感; 半监督回归; 迁移成分分析; 土壤重金属质量浓度反演

本文引用格式

毛耿旋, 涂彦, 崔文博, 陶超 . 基于半监督回归的高光谱土壤重金属质量浓度反演[J]. 应用科学学报, 2022 , 40(6) : 941 -952 . DOI: 10.3969/j.issn.0255-8297.2022.06.005

Abstract

Aiming at training a robust inversion model of soil heavy metal mass concentration with a small number of labeled samples and a large number of unlabeled samples, we took cadmium (Cd) in soil as research object, and experimentally verified the model by using two groups of the spectral data of four different regions (Hengyang-Chenzhou, Yuanping-Baoding). A hyperspectral retrieval model of soil heavy metal mass concentration based on semi-supervised regression was proposed after reducing the spectral distribution differences of different regions by means of transfer component analysis. Experimental results show that compared with traditional fully supervised modeling method, in the group of Hengyang-Chenzhou, the semi-supervised method proposed in this paper can improve the determination coefficient R2 to 0.75 and relative percent difference (RPD) to 2.15; In the group of Yuanping-Baoding, R2 increases to 0.70, and RPD increases to 1.61. The experiments show that the model inversion accuracy can be effectively improved by introducing a large number of unlabeled samples to semi-supervised regression analysis in the situations of few labeled samples.

Key words： hyperspectral remote sensing; semi-supervised regression; transfer component analysis; inversion of soil heavy metal mass concentration

参考文献

[1] 石荣杰, 潘贤章, 王昌昆, 等. 污染土壤对脐橙叶片镉含量影响的光谱预测[J]. 光谱学与光谱分析, 2015, 35(11):3140-3145. Shi R J, Pan X Z, Wang C K, et al. Prediction of cadmium content in the leaves of navel orange in heavy metal contaminated soil using VIS-NIR reflectance spectroscopy[J]. Spectroscopy and Spectral Analysis, 2015, 35(11):3140-3145. (in Chinese)
[2] Wang F, Gao J, Zha Y. Hyperspectral sensing of heavy metals in soil and vegetation:feasibility and challenges[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 136(2):73-84.
[3] Wong S C, Li X D, Zhang G, et al. Heavy metals in agricultural soils of the Pearl River Delta, South China[J]. Environmental Pollution, 2002, 119(1):33-44.
[4] 吴明珠, 李小梅, 沙晋明. 亚热带土壤铬元素的高光谱响应和反演模型[J]. 光谱学与光谱分析, 2014, 34(6):1660-1666. Wu M Z, Li X M, Sha J M. Spectral inversion models for prediction of total chromium content in subtropical soil[J]. Spectroscopy and Spectral Analysis, 2014, 34(6):1660-1666. (in Chinese)
[5] 叶明亮, 杨梦丽, 刘纯宇, 等. 高光谱遥感在土壤重金属污染监测中的应用[J]. 环境监测管理与技术, 2018, 30(6):1-5. Ye M L, Yang M L, Liu C Y, et al. Application of hyperspectral remote sensing in monitoring soil heavy metal pollution[J]. The Administration and Technique of Environmental Monitoring, 2018, 30(6):1-5. (in Chinese)
[6] 陶超, 崔文博, 王亚晋, 等. 可迁移的土壤重金属污染高光谱定性分类方法研究[J]. 光谱学与光谱分析, 2019, 39(8):2602-2607. Tao C, Cui W B, Wang Y J, et al. Soil heavy metal qualitative classification model based on hyperspectral measurements and transfer learning[J]. Spectroscopy and Spectral Analysis, 2019, 39(8):2602-2607. (in Chinese)
[7] Shi T Z, Chen Y Y, Liu Y L, et al. Visible and near-infrared reflectance spectroscopyan alternative for monitoring soil contamination by heavy metals[J]. Journal of Hazardous Materials, 2014, 265:166-176.
[8] Liu H Z, Shi T Z, Chen Y Y, et al. Improving spectral estimation of soil organic carbon content through semi-supervised regression[J]. Remote Sensing, 2017, 9(1):29.
[9] 乔晓英, 马少阳, 候会芳, 等. 矿区植物重金属污染的高光谱特性及其反演模型[J]. 安全与环境学报, 2018, 18(1):335-341. Qiao X Y, Ma S Y, Hou H F, et al. Hyper-spectral features of heavy metal pollutants in vegetables and their inversion model in the mining areas[J]. Journal of Safety and Environment, 2018, 18(1):335-341. (in Chinese)
[10] Liu Z H, Lu Y, Peng Y P, et al. Estimation of soil heavy metal content using hyperspectral data[J]. Remote Sensing, 2019, 11(12):1464.
[11] 贺军亮, 张淑媛, 查勇, 等. 高光谱遥感反演土壤重金属含量研究进展[J]. 遥感技术与应用, 2015(3):407-412. He J L, Zhang S J, Zha Y, et al. Review of retrieving soil heavy metal content by hyperspectral remote sensing[J]. Remote Sensing Technology and Application, 2015, 30(3):407-412. (in Chinese)
[12] 沈文娟, 蒋超群, 侍昊, 等. 土壤重金属污染遥感监测研究进展[J]. 遥感信息, 2014(6):112-117. Shen W J, Jang C Q, Chi H, et al. Progress in soil heavy metal pollution monitoring via remote sensing technology[J]. Remote Sensing Information, 2014(6):112-117. (in Chinese)
[13] 郭云开, 刘宁, 刘磊, 等. 土壤Cu含量高光谱反演的BP神经网络模型[J]. 测绘科学, 2018, 43(1):135-139. Guo Y K, Liu N, Liu L, et al. Hyperspectral inversion of soil Cu content based on BP neural network model[J]. Science of Surveying and Mapping, 2018, 43(1):135-139. (in Chinese)
[14] 赵玉玲, 杨楠楠, 张海霞, 等. 基于高光谱的邯郸市土壤重金属统计估算模型研究[J]. 生态环境学报, 2020, 29(4):819-826. Zhao Y L, Yang N N, Zhang H X, et al. Study on the statistical estimation model of soil heavy metals in Handan City based on hyperspectral[J]. Ecology and Environment Sciences, 2020, 29(4):819-826. (in Chinese)
[15] Chen T, Chang Q R, Clevers J G P W, et al. Rapid identification of soil cadmium pollution risk at regional scale based on visible and near-infrared spectroscopy[J]. Environmental Pollution, 2015, 206(11):217-226.
[16] Sun W C, Zhang X. Estimating soil zinc concentrations using reflectance spectroscopy[J]. International Journal of Applied Earth Observation and Geoinformation, 2017, 58(6):126-133.
[17] Sun W C, Zhang X, Sun X J, et al. Predicting nickel concentration in soil using reflectance spectroscopy associated with organic matter and clay minerals[J]. Geoderma, 2018, 327(10):25-35.
[18] Cheng H, Shen R L, Chen Y Y, et al. Estimating heavy metal mass concentrations in suburban soils with reflectance spectroscopy[J]. Geoderma, 2019, 336(9):59-67.
[19] Hong Y S, Shen R L, Cheng H, et al. Cadmium concentration estimation in peri-urban agricultural soils:using reflectance spectroscopy, soil auxiliary information, or a combination of both?[J]. Geoderma, 2019, 354(11):113875.
[20] Lü M Q, Li Y F, Chen L, et al. Air quality estimation by exploiting terrain features and multi-view transfer semi-supervised regression[J]. Information Sciences, 2019, 483(5):82-95.
[21] Zhang Y W, Liu W Y, Huan Y, et al. Online dynamic total transfer capability estimation using cotraining-style semi-supervised regression[J]. IEEE Access, 2020, 8(8):94054-94064.
[22] Khan F M, Liu Q H. Transduction of semi-supervised regression targets in survival analysis for medical prognosis[C]//2011 IEEE International Conference on Data Mining, 2011:1018-1025.
[23] 付馨, 赵艳玲, 李建华, 等. 高光谱遥感土壤重金属污染研究综述[J]. 中国矿业, 2013, 22(1):65-68. Fu X, Zhao Y L, Li J H, et al. Research on hyper-spectral remote sensing in heavy metal pollution soil[J]. China Mining Magazine, 2013, 22(1):65-68. (in Chinese)
[24] 史舟. 土壤地面高光谱遥感原理与方法[M]. 北京:科学出版社, 2014.
[25] Zhou Z H, Li M. Semi-supervised regression with cotraining-style algorithms[J]. IEEE Educational Activities Department, 2007, 19(11):1479-1493.
[26] Zhou X Y, Belkin M. Semi-supervised learning[J]. Academic Press Library in Signal Processing, 2014, 1(2):1239-1269.
[27] 刘建伟, 刘媛, 罗雄麟. 半监督学习方法[J]. 计算机学报, 2015, 38(8):1592-1617. Liu J W, Liu Y, Luo X L. Semi-supervised and ensemble learning:a review[J]. Chinese Journal of Computers, 2015, 38(8):1592-1617. (in Chinese)
[28] 蔡毅, 朱秀芳, 孙章丽, 等. 半监督集成学习综述[J]. 计算机科学, 2017, 44(1):7-13. Cai Y, Zhu X F, Sun Z L, et al. Semi-supervised and ensemble learning:a review[J]. China Computer Science, 2017, 44(1):7-13. (in Chinese)
[29] Tao C, Wang Y J, Cui W B, et al. A transferable spectroscopic diagnosis model for predicting arsenic contamination in soil[J]. Science of The Total Environment, 2019, 669(7):964-972.
[30] Drucker H, Burges C J C, Kaufman L, et al. Support vector regression machines[J]. Advances in Neural Information Processing Systems, 1997, 28(7):779-784.
[31] Ergon R. Principal component regression (PCR) and partial least squares regression (PLSR)[M]. Hoboken:Wiley, 2013:121-142.
[32] Zhang J. Rapid determination of low heavy metal mass concentrations in grassland soils around mining using vis-nir spectroscopy:a case study of Inner Mongolia, China[J]. Sensors, 2021, 21(9):3220.
Options
文章导航

/

〈 〉