联合FOD-sCARS的土壤有机质高光谱机器学习估测模型

doi:10.3964/j.issn.1000-0593(2025)01-0204-09

摘要
参考文献
相关文章 (15)

全文: PDF (24874 KB)
输出: BibTeX | EndNote (RIS)

摘要：土壤有机质（SOM）含量是表征土壤质量的关键指标，在全球碳循环系统中发挥重大作用。快速准确的SOM估算和空间制图对土壤碳库估算、作物生长监测和耕地规划管理具有重要意义。利用传统方法监测区域性SOM含量耗时费力，基于高光谱遥感影像建立SOM估测模型是现在较为合理有效的方法。为探索解决目前高光谱遥感影像建立SOM含量估测模型存在光谱数据冗余、光谱数据特征提取精度低、小样本模型泛化能力不强的问题，选择位于青海省湟中县的研究区，共采集67个土壤样本。获取资源1号02D（ZY1-02D）高光谱遥感影像并进行预处理得到样点像元光谱数据，采用分数阶微分变换（FOD）方法挖掘与SOM含量具有响应关系的敏感波段，以0.2为一个步长，利用相关性阈值法对比分析不同阶次微分处理数据挖掘能力；运用稳定性竞争性自适应重加权采样算法（sCARS）去除高光谱冗余数据获取建模特征波段，选择随机森林（RF）、极端梯度提升树、极限学习机和岭回归机器学习作为建模算法，以全波段和特征波段光谱数据分别作为模型输入变量构建SOM估测模型进行高光谱反演研究工作；最后根据最优特征变量和建模算法，基于ZY1-02D遥感影像进行了SOM空间分布制图。结果表明：采用FOD变换相比整数阶可以大大提高波段与SOM含量间的相关性，挖掘出更多细微的与SOM含量产生响应关系的光谱波段，其中0.8阶微分变换效果最优，较原始波段相比相关系数最大值提高了0.546；相较于全波段光谱数据，采用sCARS特征提取方法获取特征波段构建模型的估测精度得到较大提升，说明sCARS可以有效提升建模数据的质量，提升模型预测精度。建模算法中RF表现最优，R²_p（模型决定系数）达到0.766，RPD达到1.86，较全波段建模结果R²_p提升约7.58%；基于FOD-sCARS和RF实现了区域SOM含量估测制图。研究进一步验证利用星载高光谱遥感影像是实现区域SOM估测制图的可靠途径，研究结果可为估测区域SOM含量提供新思路，为利用星载高光谱遥感影像绘制SOM含量空间分布图提供了数据支持。

关键词：高光谱遥感影像；分数阶微分变换；稳定性竞争性自适应重加权采样算法；土壤有机质；随机森林

Abstract：Soil organic matter (SOM) content is a key index of soil quality and plays an important role in the global carbon cycle system. Rapid and accurate estimation and spatial mapping of SOM content are significant for soil carbon pool estimation, crop growth monitoring, cultivated land planning, and management. It is time-consuming and difficult to use traditional methods to monitor regional SOM content, and it is a reasonable and effective method to establish an SOM estimation model based on hyperspectral remote sensing images. However, the SOM content estimation model for hyperspectral remote sensing images has some problems, such as spectral data redundancy, low feature extraction accuracy, and weak generalization ability of a small sample model. In this paper, a total of 67 soil samples were collected in Huangzhong County, Qinghai Province. The ZY1-02D hyperspectral remote sensing image was obtained and preprocessed to obtain pixel spectral data of the sample points. The fractional-order differential transform (FOD) method explored the sensitive bands with a response relationship with SOM content. With 0.2 as a step, the correlation threshold method was used to compare and analyze different order differential processing data mining capabilities. The stable competitive adaptive reweighted sampling algorithm (sCARS) removes hyperspectral redundant data to obtain the modeling feature bands. Random forest (RF), extreme gradient lifting tree, extreme learning machine, and ridge regression machine learning are selected as modeling algorithms. The SOM estimation model is constructed using the spectral data of the full band and the characteristic band as input variables. The results show that the FOD transform can greatly improve the correlation between the band and the SOM content compared with the integer order, and more subtle spectral bands with a response relationship with SOM content can be mined. The 0.8th-order differential transform has the best effect, and the maximum correlation coefficient is increased by 0.546. Compared with full-band spectral data, the estimation accuracy of the model constructed with the sCARS feature extraction method is greatly improved, indicating that sCARS can effectively improve the quality of modeling data and the model's prediction accuracy. In the modeling algorithm, RF performance is the best, withR²_p (determination coefficient) reaching 0.766 and RPD reaching 1.86, which is about 7.58% higher than theR²_p of the full-band modeling result. Regional SOM content estimation mapping was realized based on FOD-sCARS and RF. This study further verifies that space-borne hyperspectral remote sensing images are a reliable way to achieve regional SOM estimation mapping, and the research results can provide a new idea for estimating regional SOM content and provide data support for mapping spatial distribution map of SOM content using space-borne hyperspectral remote sensing images.

Key words：Hyperspectral remote sensing images; Fractional-order differential transform (FOD);Stable competitive adaptive reweighted sampling algorithm (sCARS);Soil organic matter (SOM); Random forest (RF)

收稿日期: 2023-11-08 修订日期: 2024-04-12

中图分类号:

S153.6+21

基金资助: 吉林省教育厅科学研究项目（JJKH20230342KJ），吉林省科技发展计划项目（YDZJ202401507ZYTS），吉林省自然科学基金项目（20230101373JC）和国家自然科学基金项目（42201435）资助

作者简介: 吴梦红，女，1992年生，吉林农业大学资源与环境学院博士研究生 e-mail: wumenghong@jlju.edu.cn

引用本文:

吴梦红，窦森，林楠，姜然哲，陈思，李佳璇，付佳伟，梅显军. 联合FOD-sCARS的土壤有机质高光谱机器学习估测模型[J]. 光谱学与光谱分析, 2025, 45(01): 204-212.
WU Meng-hong, DOU Sen, LIN Nan, JIANG Ran-zhe, CHEN Si, LI Jia-xuan, FU Jia-wei, MEI Xian-jun. Hyperspectral Estimation of Soil Organic Matter Based on FOD-sCARS and Machine Learning Algorithm. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(01): 204-212.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)01-0204-09 或 https://www.gpxygpfx.com/CN/Y2025/V45/I01/204

[1] DOU Sen, LI Kai, GUAN Song(窦森, 李凯, 关松). Acta Peological Sinica(土壤学报)，2011, 48(2): 412.
[2] LIU Zhan-feng, FU Bo-jie, LIU Guo-hua, et al(刘占锋, 傅伯杰, 刘国华, 等). Acta Ecologica Sinica(生态学报), 2006，26(3): 901.
[3] XU Ming-gang, YU Rong, WANG Bo-ren(徐明岗, 于荣, 王伯仁). Soil and Fertilizer Sciences in China(中国土壤与肥料), 2000，(6): 3(doi: 10.11838/sfsc.20000601).
[4] CHEN Qing-qiang, SHEN Cheng-de, YI Wei-xi, et al(陈庆强, 沈承德, 易惟熙, 等). Advances in Earth Sciences(地球科学进展), 1998，13(6): 555.
[5] Chen Y, Wang J, Liu G, et al. Forests, 2019, 10: 217.
[6] Angelopoulou T, Chabrillat S, Pignatti S, et al. Remote Sensing, 2023, 15: 1106.
[7] Wu M, Dou S, Lin N, et al. Remote Sensing, 2023, 15(19): 4713.
[8] YAN Xiang-zhao, YAO Yan-min, ZHANG Xiao-yu, et al(颜祥照, 姚艳敏, 张霄羽, 等). China Agricultural Informatics(中国农业信息), 2019, 31(3): 13.
[9] Lin C, Zhu A-X, Wang Z, et al. International Journal of Applied Earth Observation and Geoinformation, 2020, 89: 102094.
[10] Hong Y, Guo L, Chen S, et al. Geoderma, 2020, 365: 114228.
[11] YAN Xiang-zhao, YAO Yan-min, ZHANG Xiao-yu, et al(颜祥照, 姚艳敏, 张霄羽, 等). Soil and Fertilizer Sciences in China(中国土壤与肥料), 2021，(5): 10.
[12] ZHENG Guang-hui, WANG Ming-jiang, JIAO Cai-xia, et al(郑光辉, 王明江, 焦彩霞, 等). Journal of Nanjing University of Information Science & Technology(Natural Science Edition)[南京信息工程大学学报(自然科学版)], 2013, 5(6): 481.
[13] Wu M, Lin N, Li G, et al. Environment Pollutants and Bioavailability, 2022, 34(1)：308.
[14] Hong Y, Liu Y, Chen Y, et al. Geoderma, 2019, 337: 758.
[15] Hong Y, Chen S, Liu Y, et al. Catena, 2019, 174: 104.
[16] Meng X, Bao Y, Ye Q, et al. Remote Sensing, 2021, 13: 2273.
[17] Hong Y, Chen Y, Yu L, et al. Remote Sensing, 2018, 10: 479.
[18] Lu Y, Bai Y, Yang L, et al. New Zealand Journal of Agricultural Research, 2007, 50(5): 1169.
[19] Song S, Yu H, Zhang Q, et al. Ecological Indicators, 2023, 155: 110964.
[20] Hong Y, Chen S, Zhang Y, et al. Science of the Total Environment, 2018, 644: 1232.
[21] Zhang M, Zhang M, Yang H, et al. Remote Sensing, 2021, 13: 2934.
[22] CHEN Yi-yun, QI Kun, LIU Yao-lin, et al(陈奕云, 漆锟, 刘耀林，等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2015, 35(6): 1705.
[23] Zhao M, Wang T, Lu Y, et al. PLOS ONE, 2023, 18(6): e0286825.
[24] Zheng K, Li Q, Wang J, et al. Chemometrics & Intelligent Laboratory Systems, 2012, 112: 48.
[25] Li H, Wang J, Zhang J, et al. Agronomy, 2022, 12: 638.
[26] Xu X, Chen S, Xu Z, et al. Remote Sensing, 2020, 12: 3765.
[27] Liu J, Dong Zh, Xia J, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2021, 258: 119823.
[28] Guo H, Zhang R, Dai, W, et al. Agronomy, 2022, 12: 2111.