叶片多理化参数的高光谱遥感与深度学习估算

doi:10.3964/j.issn.1000-0593(2024)10-2873-11

摘要
参考文献
相关文章 (15)

全文: PDF (23969 KB)
输出: BibTeX | EndNote (RIS)

摘要：准确的植物叶片理化参数对于监测植物生长状况至关重要。随着深度学习技术的迅速应用，结合深度学习和高光谱遥感技术的植物叶片理化参数分析应用潜力巨大；然而，现阶段结合深度学习和高光谱遥感技术在植物多叶片理化参数的联合估算研究尚少。该研究旨在挖掘结合高光谱遥感技术和深度学习技术开展高精度的多植被叶片理化参数（叶绿素含量、类胡萝卜素含量、水分含量、蛋白质含量和碳基成分含量）联合估算的潜力。首先，通过利用新型PROSPECT-PRO辐射传输模型模拟分析，确定了多个植被叶片理化参数的敏感光谱区域，并设计了LeafTraitNet模型；然后，基于Lopex93数据开展LeafTraitNet模型训练和验证，取得了高精度的叶片参数估算结果。得到以下结论：（1）基于PROSPECT-PRO辐射传输模型辅助实测开展植被光谱特征选择十分必要；叶绿素（约434和约676 nm）和类胡萝卜素（约445 nm）两类色素的光谱吸收峰均主要位于可见光区域；然而，其与叶片光谱的相关系数绝对值最大的点却不是各自的吸收峰位置，这可能是因为叶绿素和类胡萝卜素对光谱吸收的相互影响。（2）水分的吸收峰主要位于950～2 500 nm范围内，这与叶片蛋白质和碳基成分的吸收区域重叠，因此削弱了后者的高光谱遥感估算精度。基于PROSPECT-PRO辐射传输模型和Lopex93数据集的叶片参数相关性分析结果表明，叶片水分含量与950～2 500 nm范围内叶片光谱反射率相关系数绝对值接近1，而叶片蛋白质和碳基成分含量与950～2 500 nm范围内光谱反射率相关系数较低。（3）三种传统统计回归方法和LeafTraitNet模型的叶片理化参数估算精度可以基于其估算总nRMSE而排序为：LeafTraitNet（总nRMSE=0.84）< RF（总nRMSE=1.59）< MLP（总nRMSE=1.73）< MLR（总nRMSE=1.74）。研究显示基于深度学习的LeafTraitNet模型有潜力提供远高于传统统计回归模型的植物叶片理化参数估算结果，但其在冠层尺度的研究仍需要更多的实验来进一步验证。

关键词：深度学习；高光谱遥感；叶片蛋白质含量；叶片叶绿素含量；叶片类胡萝卜素含量

Abstract：Plant leaf physical and chemical parameters, such as leaf chlorophyll content, carotenoid content, water content, protein content, and Carbone-based constituents content, are crucial for accurately monitoring plant growth status. In recent years, with the rapid development of deep learning technology in vegetation remote sensing, the combined use of deep learning and hyperspectral remote sensing for plant leaf parameters estimation has been widely applied; however, currently, few leaf parameters estimation works based on the combination of deep learning and hyperspectral remote sensing technology have been conducted. This study explores the possibility of estimating leaf chlorophyll, carotenoid, water, protein, and Carbone-based constituent content by combining hyperspectral remote sensing and deep learning techniques. The main work of this paper is to propose a leaf physical and chemical parameter estimation method based on hyperspectral remote sensing and deep learning. Firstly, this study determines the sensitive spectral regions of multiple vegetation leaf physical and chemical parameters based on the PROSPECT-PRO radiative transfer model. Then, we designed a LeafTraitNet deep learning model; the LeafTraitNet model is trained and tested based on the lobex93 dataset, and a high-precision leaf parameter estimation result is obtained. The conclusions of this study are as follows: (1) It is vital to select leaf spectral absorption features based on the PROSPECT-PRO radiative transfer model. The leaf chlorophyll (434 and 676 nm) and carotenoids (445 nm) spectral absorption regions are located in the visible bands. However, the absorption regions with the most significant correlation coefficients (absolute values) are not their maximum spectral absorption bands, which the mutual influence of leaf chlorophyll and carotenoid absorptions may cause. (2) The leaf water spectral absorption regions are mainly located in the bands 950～2 500 nm, which overlaps with the spectral absorption regions of leaf protein and carbon-based component content, thus weakening the hyperspectral remote sensing estimation accuracy of the latter. The correlation coefficients between leaf protein (and carbon-based component content) and the spectral reflectance in the 950～2 500 nm range are notably lower than the leaf water. The correlation coefficients analysis results of leaf parameters and hyperspectral for the PROSPECT-PRO radiative transfer model and lobex93 dataset show similar correlation coefficients. (3) The three traditional methods and the LeafTraitNet model can be ranked as LeafTraitNet (total nRMSE=0.84) < RF (total nRMSE=1.59) < MLP (total nRMSE=1. 73) < MLR (total nRMSE=1.74), which means the leaf parameters estimation performance is notably higher than RF, MLP, and MLR. However, further experiments are needed to validate the LeafTraitNet model at the canopy scale.

Key words：Deep learning; Hyperspectral; Leaf protein content; Leaf chlorophyll content; Leaf carotenoids content

收稿日期: 2024-03-09 修订日期: 2024-07-11

中图分类号:

S127

基金资助: 国家自然科学基金项目（42101362），河南省科技攻关计划项目（232102111123）资助

通讯作者: 乔红波 E-mail: qiaohb@henau.edu.cn

作者简介: 岳继博，1992年生，河南农业大学信息与管理科学学院讲师 e-mail: yuejibo@henau.edu.cn

引用本文:

岳继博，冷梦蝶，田庆久，郭伟，刘杨，冯海宽，乔红波. 叶片多理化参数的高光谱遥感与深度学习估算[J]. 光谱学与光谱分析, 2024, 44(10): 2873-2883.
YUE Ji-bo, LENG Meng-die, TIAN Qing-jiu, GUO Wei, LIU Yang, FENG Hai-kuan, QIAO Hong-bo. Estimation of Leaf Physical and Chemical Parameters Based on Hyperspectral Remote Sensing and Deep Learning Technologies. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2024, 44(10): 2873-2883.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2024)10-2873-11 或 https://www.gpxygpfx.com/CN/Y2024/V44/I10/2873

[1] Li Dong，Chen Jingming，Yan Yan，et al. Remote Sensing of Environment，2022，283：113314.
[2] Wong C Y S, D'Odorico P, Bhathena Y，et al. Remote Sensing of Environment，2019，233：111407.
[3] Sonobe R，Hirono Y. Remote Sensing，2023，15（17）：4303.
[4] Falcioni R，Antunes W C，Demattê J A M，et al. Sensors，2023，23（8）：3843.
[5] Féret Jean-Baptiste，Berger Katja，de Boissieu Florian，et al. Remote Sensing of Environment，2021，252：112173.
[6] ZHANG Mei-ling，CHEN Yong-jie，WANG Min-juan，et al（张美玲，陈勇杰，王敏娟，等）. Spectroscopy and Spectral Analysis（光谱学与光谱分析），2024，44（3）: 865.
[7] Yue Jibo，Yang Hao，Feng Haikuan，et al. Computers and Electronics in Agriculture，2023，211：108011.
[8] Hu Jingyu，Feng Hao，Wang Qilei，et al. Remote Sensing，2024，16（5）：784.
[9] YUAN Pei-sen，DING Yi-fei，XU Huan-liang（袁培森，丁毅飞，徐焕良）. Transactions of the Chinese Society for Agricultural Machinery（农业机械学报），2024，55（2）: 258.
[10] TIE Ning，LIU Bing-yi（铁凝，刘秉义）. Acta Optica Sinica（光学学报），2023，43（24）: 75.
[11] WANG Nan，LI Zhen，LI Jia-meng，et al（王楠，李震，李佳盟，等）. Transactions of the Chinese Society for Agricultural Machinery（农业机械学报），2023，54（S2）: 260.
[12] GUO Feng-wei，WANG Peng-xin，LIU Jun-ming，et al（郭丰玮，王鹏新，刘峻明，等）. Transactions of the Chinese Society for Agricultural Machinery（农业机械学报），2024，55（1）: 164.
[13] WANG Ya-zhou，XIAO Zhi-yun（王亚洲，肖志云）. Transactions of the Chinese Society for Agricultural Machinery （农业机械学报），2024，55（1）: 196.
[14] BIAN Ming-bo，MA Yan-peng，FAN Yi-guang，et al（边明博，马彦鹏，樊意广，等）. Transactions of the Chinese Society for Agricultural Machinery（农业机械学报），2023，54（8）: 240.
[15] XIAO Tian-pu，YANG Li，ZHANG Dong-xing，et al（肖天璞，杨丽，张东兴，等）. Transactions of the Chinese Society of Agricultural Engineering（农业工程学报），2022，38（16）: 162.
[16] FU Bo-lin，DENG Liang-chao，ZHANG Li，et al（付波霖，邓良超，张丽，等）. National Remote Sensing Bulletin（遥感学报），2022，26（6）: 1182.
[17] Simonyan K，Zisserman A. Very Deep Convolutional Networks for Large-Scale Image Recognition, 2015, arXiv: 1409.1556V6[CS.CV].
[18] HUANG Lü-wen，TIAN Xu，REN Lie-hong，et al（黄铝文，田旭，任烈弘，等）. Transactions of the Chinese Society of Agricultural Engineering（农业工程学报），2023，39（2）：252.
[19] HE Zhi-fen, SHI Na, SONG Lu-kai, et al(贺致芬，石娜，宋禄楷，等). Remote Sensing Information(遥感信息)，2023，38(6)：117.
[20] CHEN Jun-qing，LI Yi，WANG Bin，et al（陈俊清，李毅，王斌，等）. Journal of Natural Disasters（自然灾害学报），2024，33（1）：74.