基于多植被指数组合的棉花叶片叶绿素含量估算

doi:10.13866/j.azr.2023.11.16

摘要/Abstract

摘要：

叶绿素含量是表征植被生长状况的重要参考指标，利用高光谱技术快速，精确地监测棉花叶片叶绿素含量，以新疆125个苗期棉花叶片样本为研究对象，通过测定其叶绿素含量与光谱数据，采用多种光谱预处理和多植被指数相结合的方法，构建了WOA-RFR棉花叶片叶绿素含量定量反演模型，并与SVR和RFR模型结果进行对比分析。结果表明：（1）光谱变换方法中对数变换、分数阶微分和连续小波变换均能有效地提高植被指数与叶绿素含量的相关性。（2）基于分数阶微分0.9阶变换的Vogelmann3、RVI、DVI、SR_[675-700]、Mndvi₇₀₅、ND、VOG1、NVI、TVI和VOG2植被指数组合的WOA-RFR模型反演效果最佳，其建模集和验证集模型R²分别为0.920和0.955，RMSE分别为0.987和0.986，MRE分别为0.013和0.014，与RFR和SVR模型相比，预测精度有所提高，WOA算法优化模型效果明显。研究结果可为棉花叶片叶绿素含量定量反演提供决策依据。

关键词: 植被指数组合, 棉花, 叶绿素含量, 鲸鱼优化算法

Abstract:

Chlorophyll content is a crucial indicator for characterizing vegetation growth. In this study, we utilized high-spectral technology to rapidly monitor the chlorophyll contents of cotton leaves. We collected 125 cotton leaf seedling samples from Xinjiang and measured their chlorophyll content and spectral data. To achieve this, we employed various spectral preprocessing techniques and used a combination of vegetation indices. Subsequently, we constructed a whale optimization algorithm/random forest regression (WOA-RFR) quantitative inversion model for cotton leaf chlorophyll content. Finally, we conducted a comparative analysis, contrasting the results of the WOA-RFR model with those obtained from the support vector regression (SVR) and RFR models. The results indicated that the spectral transformation methods (logarithm transformation, fractional order differentiation, and wavelet transformation) effectively improved the correlation between the vegetation indices and the chlorophyll content. We also found that the best inversion performance was achieved with the WOA-RFR model using a fractional order differentiation with a transformation order of 0.9 and the Vogelmann3, RVI, DVI, SR_[675-700], Mndvi₇₀₅, ND, VOG1, NVI, TVI, VOG2 combined vegetation indices. The model exhibited R² values of 0.920 and 0.955 for the training set and validation set, respectively. The corresponding RMSE values were 0.987 and 0.986, while the MRE values were 0.013 and 0.014. Compared to the RFR and SVR models, the WOA-RFR model demonstrated higher predictive accuracy, and the optimization effect of the WOA algorithm was evident. As a result, this study provides valuable decision-making support for accurately quantifying cotton leaf chlorophyll content.

Key words: combination of vegetation index, cotton, chlorophyll content, whale optimization algorithm

阿热孜古力·肉孜, 买买提·沙吾提, 何旭刚, 冶晓文. 基于多植被指数组合的棉花叶片叶绿素含量估算[J]. 干旱区研究, 2023, 40(11): 1865-1874.

Areziguli ROZI, Mamat SAWUT, HE Xugang, YE Xiaowen. Estimation of cotton leaf chlorophyll content based on combinations of multi-vegetation indices[J]. Arid Zone Research, 2023, 40(11): 1865-1874.

图/表 9

图1

表1

高光谱植被指数"

植被指数	计算公式	文献	植被指数	计算公式	文献
GNDVI	$(R 800 - R 500) / (R 800 + R 500)$	[19]	NRI	$(R 570 - R 670) / (R 570 + R 670)$	[20]
PSRI	$(R 678 - R 500) / R 750$	[20]	NPCI	$(R 680 - R 430) / (R 680 + R 430)$	[20]
VARI	$(R 555 - R 680) / (R 580 + R 680 + R 480)$	[21]	RVI	$R 800 / R 680$	[21]
NDVI	$R 800 - R 680 / R 800 + R 680$	[21]	DVI	$R 800 - R 680$	[21]
VOG1	$R 780 / R 740$	[21]	VOG2	$R 734 - R 747 / R 715 - R 726$	[21]
G	$R 780 / R 740$	[23]	Lichtenthaler1	$R 440 / R 690$	[23]
Lichtenthaler2	$(R 800 - R 680) / (R 800 + R 680)$	[23]	SIPI	$(R 800 - R 450) / (R 800 + R 450)$	[23]
PSSRa	$R 800 / R 680$	[23]	Vogelmann1	$(R 734 - R 747) / (R 715 + R 720)$	[23]
Vogelmann3	$R 740 / R 720$	[23]	ND	$(R 935 - R 705) / (R 935 + R 705)$	[23]
CIred edge	$R 750 / R 705 - 1$	[24]	CIgreen	$R 800 / R 550 - 1$	[24]
RI-half	$R 747 / R 708$	[24]	GRVI	$(R 670 / R 500) - 1$	[24]
RDVI	$(R 800 - R 670) / R 800 - R 670$	[25]	Datt1	$(R 850 - R 710) / (R 850 + R 680)$	[25]
Datt2	$R 850 / R 710$	[25]	Datt3	$R 754 / R 704$	[25]
Carte1	$R 695 / R 420$	[25]	Carte2	$R 695 / R 760$	[25]
Carte3	$R 605 / R 760$	[25]	Carte4	$R 710 / R 760$	[25]
SR_[752,690]	$R 752 / R 690$	[25]	SR_[675,700]	$R 675 / R 700$	[25]
NVI	$(R 777 - R 747) / R 673$	[25]	MSR705	$(R 750 - R 445) / (R 705 + R 445)$	[25]
SAVI	$(1 + 0.5) × (R 800 - R 670) / (R 800 + R 670 + 0.5)$				[19]
GARI	$[R 800 - [R 550 - 1.7 × (R 470 - R 670)]] [R 800 + [R 550 - 1.7 × (R 470 + R 670)]]$				[19]
OSAVI	$1.16 × [(R 800 - R 670) / (R 800 + R 670 + 0.16)$ ]				[21]
R_O	红谷(640~680 nm)内反射率最小值				[22]
mND₇₀₅	$(R 750 - R 705) / (R 750 + R 705 - 2 × R 445)$				[24]
EVI	$2.5 × [(R 800 - R 670) / (R 800 + 6 R 670 - 7.5 × R 475 + 1)$				[25]
MCARI	$(R 700 - R 670 - 0.2 × (R 700 - R 550)] × (R 700 / R 670)$				[25]
TVI	$0.5 × 120 × (R 750 - R 550 - 2.5 × R 670 - R 550]$				[25]
MTVI1	$1.2 × 1.2 × (R 800 - R 550 - 2.5 × (R 670 - R 550)$				[25]
REP	$700 + 40 × (R 670 + R 780 / 2 - R 700] / (R 740 - R 700)$				[25]
SPVI	$0.4 × 3.7 × (R 670 - R 550) - 1.2 × R 530 - R 670$				[25]
SPVI2	$0.4 × 3.7 × (R 800 - R 670) - 1.2 × R 550 - R 670$				[25]

表1

图2

图3

图4

图5

表2

表3

图6

参考文献 30

[1]	依尔夏提·阿不来提, 买买提·沙吾提, 白灯莎·买买提艾力, 等. 基于随机森林法的棉花叶片叶绿素含量估算[J]. 作物学报, 2019, 45(1): 81-90. doi: 10.3724/SP.J.1006.2019.84058
	[Ershat Ablet, Mamat Sawut, Baidengsha Maimaitiaili, et al. Estimation of leaf chlorophyll content in cotton based on the random forest approach[J]. Acta Agronomica Sinica, 2019, 45(1): 81-90.] doi: 10.3724/SP.J.1006.2019.84058
[2]	王伟东, 常庆瑞, 王玉娜. 基于UHD185成像光谱仪的冬小麦叶绿素监测[J]. 干旱区研究, 2020, 37(5): 1362-1369.
	[Wang Weidong, Chang Qingrui, Wang Yuna. SPAD monitoring of winter wheat based on a UHD185 imaging spectrometer[J]. Arid Zone Research, 2020, 37(5): 1362-1369.]
[3]	Xiao Q L, Tang W T, Zhang C, et al. Spectral preprocessing combined with deep transfer learning to evaluate chlorophyll content in cotton leaves[J]. Plant Phenomics, 2022, 2022: 9813841.
[4]	安赛日丁·艾尔肯, 买买提·沙吾提, 李金朝. 基于冠层光谱特征的长绒棉叶绿素含量估算[J]. 激光与光电子学进展, 2022, 59(5):1-11.
	[Arkin Ansardin, Sawut Mamat, Li Jinzhao. Estimation of chlorophyll content of long-staple cotton based on canopy spectrum characteristics[J]. Laser & Optoelectronics Progress, 2022, 59(5):1-11. ]
[5]	Liu N, Xing Z P, Zhao R M, et al. Analysis of chlorophyll concentration in potato crop by coupling continuous wavelet transform and spectral variable optimization[J]. Remote Sensing, 2020, 12(17): 2826. doi: 10.3390/rs12172826
[6]	茹克亚·萨吾提, 阿不都艾尼·阿不里, 尼加提·卡斯木, 等. 基于分数阶微分的春小麦叶绿素含量高光谱估算[J]. 麦类作物学报, 2019, 39(6): 738-746.
	[Rukeya Sawuti, Abuaini Abula, Nijati Kasimu, et al. Spectral estimation of chlorophyll content in spring wheat leaves based on fractional differential[J]. Journal of Triticeae Crops, 2019, 39(6): 738-746.]
[7]	张思楠, 王权, 靳佳, 等. 应用光谱指数法估算多枝柽柳同化枝叶绿素含量[J]. 干旱区研究, 2016, 33(5): 1088-1097.
	[Zhang Sinan, Wang Quan, Jin Jia, et al. Application of hyperspectral indices for estimating leaf chlorophyll content of assimilating shoots of Tamarix Ramosissima[J]. Arid Zone Research, 2016, 33(5): 1088-1097.]
[8]	李长春, 施锦锦, 马春艳, 等. 基于小波变换和分数阶微分的冬小麦叶绿素含量估算[J]. 农业机械学报, 2021, 52(8): 172-182.
	[Li Changchun, Shi Jinjin, Ma Chunyan, et al. Estimation of chlorophyll content in winter wheat based on wavelet transform and fractional differential[J]. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(8): 172-182.]
[9]	Verma B, Prasad R, Srivastava P, et al. Investigation of optimal vegetation indices for retrieval of leaf chlorophyll and leaf area index using enhanced learning algorithms[J]. Computers and Electronics in Agriculture, 2022, 192: 106581. doi: 10.1016/j.compag.2021.106581
[10]	Wu C Y, Zheng N, Tang Q, et al. Estimating chlorophyll content from hyperspectral vegetation indices: Modeling and validation[J]. Agricultural and Forest Meteorology, 2008, 148(8-9): 1230-1241. doi: 10.1016/j.agrformet.2008.03.005
[11]	张卓然. 棉花高光谱特征及其农学参数遥感反演研究[D]. 杨凌: 西北农林科技大学, 2018.
	[Zhang Zhuoran. Research on Hyperspectral Characteristic of Cotton and Remote Sensing Inversion about Cotton Agronomic Parameters[D]. Yangling: Northwest A&F University, 2018.]
[12]	Liu N F, Townsend P A, Naber M R, et al. Hyperspectral imagery to monitor crop nutrient status within and across growing seasons[J]. Remote Sensing of Environment, 2021, 255: 112303. doi: 10.1016/j.rse.2021.112303
[13]	Sun Q, Jiao Q J, Chen X D, et al. Machine learning algorithms for the retrieval of canopy chlorophyll content and leaf area index of crops using the PROSAIL-D model with the adjusted average leaf angle[J]. Remote Sensing, 2023, 15(9): 2264. doi: 10.3390/rs15092264
[14]	郭超凡, 陈雯璟, 牛明艳, 等. 基于多植被指数模型的草地地上生物量协同估算[J]. 干旱地区农业研究, 2022, 40(4): 206-213.
	[Guo Chaofan, Chen Wenjing, Niu Mingyan, et al. Collaborative estimation of aboveground biomass in grassland based on multiple vegetation index models[J]. Agricultural Research in the Arid Areas, 2022, 40(4): 206-213.]
[15]	Qi H X, Zhu B Y, Kong L X, et al. Hyperspectral inversion model of chlorophyll content in peanut leaves[J]. Applied Sciences, 2020, 10(7): 2259. doi: 10.3390/app10072259
[16]	玉苏甫·买买提, 吐尔逊·艾山, 买合皮热提·吾拉木. 新疆渭-库绿洲棉花种植面积遥感监测研究[J]. 农业现代化研究, 2014, 35(2): 240-243.
	[Yusup Mamat, Tursun Hasan, Magpirat Gulam. Remote sensing of cotton plantation areas monitoring in delta oasis of Ugan-Kucha River, Xinjiang[J]. Research of Agricultural Modernization, 2014, 35(2): 240-243.]
[17]	刘楠, 杨海波, 高飞, 等. 基于查找表法和优化光谱指数的马铃薯叶绿素反演[J]. 中国马铃薯, 2022, 36(6): 495-507.
	[Liu Nan, Yang Haibo, Gao Fei, et al. Potato chlorophyll inversion based on LUT and optimized spectral index[J]. Chinese Potato Journal, 2022, 36(6): 495-507.]
[18]	许童羽, 邢思敏, 于丰华, 等. 基于多植被指数组合的BAS-ELM粳稻冠层氮含量反演方法[J]. 沈阳农业大学学报, 2021, 52(5): 577-585.
	[Xu Tongyu, Xing Simin, Yu Fenghua, et al. A BAS-ELM inversion method of Japonica rice canopy nitrogen content based on A combination of multiple vegetation indices[J]. Journal of Shenyang Agricultural University, 2021, 52(5): 577-585.]
[19]	陈圣波, 陈彦冰, 任枫荻, 等. 基于光谱指数的玉米叶绿素含量估算[J]. 信阳师范学院学报(自然科学版), 2021, 34(2): 225-229.
	[Chen Shengbo, Chen Yanbing, Ren Fengdi, et al. Estimation of maize chlorophyll content based on spectral index[J]. Journal of Xinyang Normal University(Natural Science Edition), 2021, 34(2): 225-229.]
[20]	傅坤亚, 张文宇, 曹宏鑫, 等. 基于光谱的作物病虫害监测研究进展[J]. 中国农业科技导报, 2014, 16(5): 90-98. doi: 10.13304/j.nykjdb.2014.018
	[Fu Kunya, Zhang Wenyu, Cao Hongxin, et al. Research progress on crop diseases and insect pests monitoring based on spectrum[J]. Journal of Agricultural Science and Technology, 2014, 16(5): 90-98.] doi: 10.13304/j.nykjdb.2014.018
[21]	王芳东, 严志雁, 赵小敏, 等. 油茶叶片叶绿素含量高光谱估测的偏最小二乘模型参数选择[J]. 江西农业大学学报, 2022, 44(1): 86-96.
	[Wang Fangdong, Yan Zhiyan, Zhao Xiaomin, et al. Parameter selection of partial least squares model for hyperspectral estimation of chlorophyll content in camellia oleifera leaves[J]. Acta Agriculturae Universitatis Jiangxiensis, 2022, 44(1): 86-96.]
[22]	王鑫梅, 张劲松, 孟平, 等. 基于无人机遥感影像的核桃冠层氮素含量估算[J]. 农业机械学报, 2021, 52(2): 178-187.
	[Wang Xinmei, Zhang Jinsong, Meng Ping, et al. Estimation of nitrogen content in walnut canopy based on UAV remote sensing image[J]. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(2): 178-187.]
[23]	孙林, 程丽娟. 植被叶片生化组分的光谱响应特征分析[J]. 光谱学与光谱分析, 2010, 30(11): 3031-3035. pmid: 21284178
	[Sun Lin, Cheng Lijuan. Analysis of spectral response of vegetation leaf biochemical components[J]. Spectroscopy and Spectral Analysis, 2010, 30(11): 3031-3035.] pmid: 21284178
[24]	洪帅, 张泽, 张立福, 等. 滴灌棉花不同生育时期冠层叶片叶绿素含量的高光谱估测模型[J]. 棉花学报, 2019, 31(2): 138-146.
	[Hong Shuai, Zhang Ze, Zhang Lifu, et al. Hyperspectral estimation model of chlorophyll content in cotton canopy leaves under drip irrigation at different growth stages[J]. Cotton Science, 2019, 31(2): 138-146.]
[25]	Liang L, Di L P, Zhang L P, et al. Estimation of crop LAI using hyperspectral vegetation indices and a hybrid inversion method[J]. Remote Sensing of Environment, 2015, 165: 123-134. doi: 10.1016/j.rse.2015.04.032
[26]	Mirjalili S, Lewis A. The whale optimization algorithm[J]. Advances in Engineering Software, 2016, 95: 51-67. doi: 10.1016/j.advengsoft.2016.01.008
[27]	童新, 杨震雷, 张亦然, 等. 基于不同阶微分高光谱植被指数的牧区草场地上生物量估算[J]. 草地学报, 2022, 30(9): 2438-2448. doi: 10.11733/j.issn.1007-0435.2022.09.025
	[Tong Xin, Yang Zhenglei, Zhang Yiran, et al. Estimation of pasture aboveground biomass using different orders of differential hyperspectral vegetation indices[J]. Acta Agrestia Sinica, 2022, 30(9): 2438-2448.] doi: 10.11733/j.issn.1007-0435.2022.09.025
[28]	李莉婕, 岳延滨, 王延仓, 等. 高光谱定量反演火龙果茎枝叶绿素含量的研究[J]. 光谱学与光谱分析, 2021, 41(11): 3538-3544.
	[Li Lijie, Yue Yanbin, Wang Yancang, et al. The quantitative study on chlorophyll content of Hylocereus Polyrhizus based on hyperspectral analysis[J]. Spectroscopy and Spectral Analysis, 2021, 41(11): 3538-3544.]
[29]	竞霞, 张腾, 邹琴, 等. 基于分数阶微分光谱指数的小麦条锈病遥感监测模型构建[J]. 农业工程学报, 2021, 37(17): 142-151.
	[Jing Xia, Zhang Teng, Zou Qin, et al. Construction of remote sensing monitoring model of wheat stripe rust based on fractional-order differential spectral index[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(17): 142-151.]
[30]	Wu C Y, Niu Z, Tang Q, et al. Estimating chlorophyll content from hyperspectral vegetation indices: Modeling and validation[J]. Agricultural and Forest Meteorology, 2008, 148(8): 1230-1241. doi: 10.1016/j.agrformet.2008.03.005

光谱变换	植被指数	相关性范围
OR	SAVI，DVI，ND，RDVI，SPVI，SPVI2	0.24~0.26
LogR	VOG1，NDCI，Datt2，ND，RVI，NDVI，GNDVI，PSSRa，CIgreen，carte3	-0.27~-0.37
FD-0.3	SAVI，OSARI，DVI，ND，RDVI，Carte3，Carte4，Carte2，SPVI，SPVI2	-0.28~0.29
FD-0.6	DVI，SPVI2，SAVI，OSARI，EVI，ND，RDVI，Carte2，Carte4，VOG2	-0.33~0.34
FD-0.9	Vogelmann3，RVI，DVI，SR_[675-700]，Mndvi₇₀₅，ND，VOG1，NVI，TVI，VOG2	-0.23~0.30
CWT-3	CIred，PSSRa，MSR₇₀₅，RVI，Lichtenthaler2，VOG1，SIPI，CIgreen，GNDVI，SPVI	-0.27~0.28
CWT-7	REP，MCARI，Vogelmann1，Datt1，GRVI，MTVI1，carte1，Ro，G，NRI	-0.31~0.34

光谱变换	模型算法	建模集			验证集
光谱变换	模型算法	R²	RMSE	MRE	R²	RMSE	MRE
OR	WOA-RFR	0.889	0.014	0.015	0.918	0.050	0.015
	RFR	0.860	1.307	0.019	0.862	1.240	0.018
	SVR	0.756	1.241	0.013	0.908	0.854	0.006
LogR	WOA-RFR	0.912	1.067	0.016	0.894	1.081	0.015
	RFR	0.884	1.367	0.020	0.858	1.280	0.020
	SVR	0.658	1.460	0.015	0.943	0.651	0.006
FD-0.3	WOA-RFR	0.917	1.019	0.014	0.878	1.060	0.016
	RFR	0.872	1.357	0.020	0.850	1.269	0.019
	SVR	0.609	1.569	0.017	0.866	1.035	0.010
FD-0.6	WOA-RFR	0.890	1.085	0.015	0.925	1.008	0.014
	RFR	0.872	1.379	0.019	0.846	1.388	0.021
	SVR	0.451	1.836	0.020	0.685	1.594	0.018
FD-0.9	WOA-RFR	0.920	0.987	0.013	0.955	0.986	0.014
	RFR	0.916	1.250	0.018	0.949	1.207	0.017
	SVR	0.754	1.263	0.013	0.882	0.973	0.010
CWT-3	WOA-RFR	0.895	1.091	0.015	0.922	1.086	0.014
	RFR	0.892	1.388	0.020	0.900	1.443	0.020
	SVR	0.779	1.172	0.010	0.906	0.835	0.007
CWT-7	WOA-RFR	0.934	0.946	0.013	0.911	1.082	0.015
	RFR	0.913	1.286	0.019	0.841	1.281	0.020
	SVR	0.522	1.724	0.019	0.791	1.254	0.009