Estimation of Chlorophyll Density of Cotton Leaves Based on Spectral Index

doi:10.13304/j.nykjdb.2022.1051

Abstract

Abstract:

To construct the estimation model of chlorophyll density， cotton pot experiments with different nitrogen applications were set up to obtain reflectance data by spectral testing of cotton leaves at peak flowering stage. At the same time， leaf samples were collected to determine chlorophyll density. The spectral index was constructed by reflectance data， and the spectral index with high correlation with chlorophyll density was selected. The random forest （RF） algorithm and partial least squares regression （PLSR） were used to construct the estimation model of chlorophyll density in cotton leaves and verify the accuracy. The results showed that the R² of RF and PLSR estimation models were 0.723 5 and 0.704 7， respectively， and the RPD were 1.851 4 and 1.456 9， respectively， indicating that both models had rough prediction ability. The accuracy of RF model was higher than that of PLSR model， so it was feasible to use hyperspectral technology to estimate the chlorophyll density of cotton at peak flowering stage， which provided technical support for monitoring cotton population growth.

Key words: cotton, chlorophyll density, spectral characteristics of cotton, hyperspectral estimation modeling

摘要：

为构建叶绿素密度的估算模型，通过设置不同施氮量梯度的棉花盆栽试验，在盛花期对棉花叶片进行光谱测试获取反射率数据，同时采集叶片样本测定叶绿素密度，通过反射率数据构建光谱指数，优选与叶绿素密度相关性大的光谱指数，利用随机森林算法（radom forest，RF）和偏最小二乘法（partial least squares regression，PLSR）构建棉花叶片的叶绿素密度估算模型并进行精度验证。结果表明，RF与PLSR 2种估算模型的R²分别为0.723 5和0.704 7，RPD分别为1.851 4和1.456 9，表明2个模型均具有粗略的预测能力，其中，RF模型的精度要高于PLSR模型，故利用高光谱技术对盛花期棉花叶绿素密度进行无损快速估测是可行的，研究结果为棉花群体长势状况的监测提供技术支撑。

关键词: 棉花, 叶绿素密度, 光谱特征, 高光谱

CLC Number:

S562

Ziqin LI, Jiaqiang WANG, Zhen LI, Deqiu ZOU, Xiaogong ZHANG, Xiaoyu LUO, Weiyang LIU. Estimation of Chlorophyll Density of Cotton Leaves Based on Spectral Index[J]. Journal of Agricultural Science and Technology, 2024, 26(8): 103-111.

李紫琴, 王家强, 李贞, 邹德秋, 张小功, 罗霄玉, 柳维扬. 基于光谱指数的棉花叶片叶绿素密度估算研究[J]. 中国农业科技导报, 2024, 26(8): 103-111.

Figures/Tables 7

References 32

1	吴长山, 项月琴, 郑兰芬, 等. 利用高光谱数据对作物群体叶绿素密度估算的研究[J]. 遥感学报, 2000,4(3): 228-232.
	WU C S, XIANG Y Q, ZHENG L F, et al.. Estimating chlorophyll density of crop canopies by using hyperspectral data [J]. J. Remote Sensing, 2000,4(3): 228-232.
2	孟卓强, 胡春胜, 程一松. 高光谱数据与冬小麦叶绿素密度的相关性研究[J]. 干旱地区农业研究, 2007,25(6): 74-79.
	MENG Z Q, HU C S, CHENG Y S. Study on correlation between chlorophyll density of winter wheat and hyperspectral data [J]. Agric. Res. Arid Areas, 2007,25(6): 74-79.
3	PINAR A, CURRAN P J. Technical note grass chlorophyll and the reflectance red edge [J]. Int. J. Remote Sens., 1996, 17(2): 351-357.
4	HANSEN P M, SCHJOERRING J K. Reflectance measurement of canopy biomass and nitrogen status in wheat crops using normalized difference vegetation indices and partial least squares regression [J]. Remote Sens. Environ., 2003, 86(4): 542-553.
5	DANSON F M, PLUMMER S E. Red-edge response to forest leaf area index [J]. Remote Sens-basel., 1995, 16(1): 183-188.
6	BROGE N H, LEBLANC E. Comparing prediction power and stability of broadband and hyperspectral vegetation indices for estimation of green leaf area index and canopy chlorophyll density [J]. Remote Sens. Environ., 2001, 76(2): 156-172.
7	陈燕, 黄春燕, 王登伟, 等. 北疆棉花叶绿素密度的高光谱估算研究[J]. 新疆农业科学, 2006,43(6): 451-454.
	CHEN Y, HUANG C Y, WANG D W, et al.. Hyperspectral estimation of chlorophyll density of cotton in northern Xinjiang [J]. Xinjiang Agric. Sci., 2006,43(6): 451-454.
8	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.
	Ablet Ershat, Sawut Mamat, Maimaitiaili Baidengsha, et al.. Estimation of leaf chlorophyll content in cotton based on the random forest approach [J]. Acta Agron. Sin., 2019, 45(1):81-90.
9	赵婷婷, 张丽珍, 郑淑华, 等. 黍稷籽粒淀粉含量的近红外光谱分析[J]. 华北农学报, 2011, 26(1): 234-238.
	ZHAO T T, ZHANG L Z, ZHENG S H, et al.. Analysis of starch content of Panicum miliaceum L. seed by near infrared transmittance spectroscopy [J]. Acta Agric. Boreali-Sin., 2011, 26(1): 234-238.
10	季江, 高鹏飞, 贾南南, 等. 自适应多尺度窗口平均光谱平滑[J]. 光谱学与光谱分析, 2015, 35(5): 1445-1449.
	JI J, GAO P F, JIA N N, et al.. Spectral smoothing with adaptive multiscale window average [J]. Spectrosc. Spectral Anal., 2015, 35(5): 1445-1449.
11	刘倩. 基于非局部稀疏的图像去噪与平滑方法研究[D]. 济南: 山东大学, 2016.
	LIU Q. Research on image denoising and smoothing method based on non-local sparse [D]. Jinan: Shandong University, 2016.
12	潘文超, 李少昆, 王克如, 等. 基于棉花冠层光谱的土壤氮素监测研究[J]. 棉花学报, 2010, 22(1): 70-76.
	PAN W C, LI S K, WANG K R, et al.. Monitoring soil nitrogen and plant nitrogen based on hyperspectral of cotton canopy [J]. Cott. Sci., 2010, 22(1): 70-76.
13	王式琴.基于高光谱的棉花叶片氮素变化特征及估算建模研究[D].阿拉尔:塔里木大学,2021.
	WANG S Q. Study on variation characteristics and estimation modeling of cotton leaf nitrogen based on hyperspectral [D]. Alar: Tarim University, 2021.
14	Breiman L. Random forests [J]. Mach. Learn., 2001, 45(1): 5-32.
15	DONNELLY S, WALSH D. Quality of life assessment in advanced cancer [J]. Palliative Med., 1996, 10(4): 275-283.
16	梁智, 孙国强, 卫志农, 等. 基于变量选择与高斯过程回归的短期负荷预测[J]. 电力建设, 2017, 38(2): 122-128.
	LIANG Z, SUN G Q, WEI Z N, et al.. Short-term load forecasting based on variable selection and gaussian process regression [J]. Electric Power Construction, 2017, 38(2): 122-128.
17	GRÖMPING U. Variable importance assessment in regression: linear regression versus random forest [J]. Am. Statistician, 2009, 63(4): 308-319.
18	尼加提·卡斯木, 师庆东, 王敬哲, 等. 基于高光谱特征和偏最小二乘法的春小麦叶绿素含量估算[J]. 农业工程学报, 2017, 33(22): 208-216.
	Kasmu Nigati, SHI Q D, WANG J Z, et al.. Estimation of chlorophyll content in spring wheat based onhyperspectral characteristics and partial least square method [J]. Trans. Chin. Soc. Agric. Eng., 2017, 33(22): 208-216.
19	翁永玲, 戚浩平, 方洪宾, 等. 基于PLSR方法的青海茶卡-共和盆地土壤盐分高光谱遥感反演[J]. 土壤学报, 2010, 47(6): 1255-1263.
	WENG Y L, QI H P, FANG H B, et al.. PLSR-based hyperspectral remote sensing retrieval of soil salinity of Chaka-Gonghe basin in Qinghai province [J]. Acta Pedol. Sin., 2010, 47(6): 1255-1263.
20	李四海，任真.基于小波系数特征和偏最小二乘回归的近红外光谱定量分析［J］.长春大学学报，2018，28（4）：23-27.
	LI S H， REN Z. Quantitative analysis of near infrared spectroscopy based on wavelet coefficient characteristics and partial least-squares regression ［J］. J. Changchun Univ.， 2018， 28（4）：23-27.
21	杨灵玉.三江源区土壤特性高光谱估算方法比较研究［D］.西宁：青海师范大学，2016.
	YANG L Y. Comparison of estimation methods for obtaining soil properties from hyperspectral data in the Sanjiang Yuan Regions ［D］. Xining： Qinghai Normal University， 2016.
22	LAZRAQ A, CLÉROUX R. The PLS multivariate regression model: testing the significance of successive PLS components [J]. J. Chemom., 2001, 15(6): 523-536.
23	楚万林, 齐雁冰, 常庆瑞, 等. 棉花冠层叶片叶绿素含量与高光谱参数的相关性[J].西北农林科技大学学报(自然科学版), 2016, 44(9): 65-73.
	CHU W L, QI Y B, CHANG Q R, et al.. Relationship between chlorophyll content and hyperspectral parameters in canopy leaves of cotton [J]. J. Northwest A&F Univ. (Nat. Sci.), 2016, 44(9): 65-73.
24	齐雁冰, 楚万林, 解飞, 等. 基于高光谱的渭北旱塬区棉花冠层叶面积指数估算[J]. 干旱地区农业研究, 2017, 35(1): 114-121.
	QI Y B, CHU W L, XIE F, et al.. Estimating leaf area index of cotton canopy by hyperspectral reflectance in Weibei plateau [J]. Agric. Res. Arid Areas, 2017, 35(1): 114-121.
25	李银坤, 武雪萍, 梅旭荣, 等. 不同施氮水平下温室番茄叶片反射光谱特征分析[J].土壤通报, 2012, 43(1): 141-146.
	LI Y K, WU X P, MEI X R, et al.. Analysis of the spectral reflectance characteristics of greenhouse tomato at different nitrogen levels [J]. Chin. J. Soil Sci., 2012, 43(1): 141-146.
26	王强, 易秋香, 包安明, 等. 基于高光谱反射率的棉花冠层叶绿素密度估算[J]. 农业工程学报, 2012, 28(15): 125-132.
	WANG Q, YI Q X, BAO A M, et al.. Estimating chlorophyll density of cotton canopy by hyperspectral reflectance [J]. Trans. Chin. Soc. Agric., 2012, 28(15): 125-132.
27	张苏, 刘良云, 黄文江. 基于植被指数的叶绿素密度遥感反演建模与适用性研究[J]. 遥感信息, 2013, 28(3): 94-101.
	ZHANG S, LIU L Y, HUANG W J. Retrieval of canopy chlorophyll density based on hyperpectral vegetation indices [J]. Remote Sensing Inf., 2013, 28(3): 94-101.
28	乔星星, 冯美臣, 杨武德, 等. SG平滑处理对冬小麦地上干生物量光谱监测的影响[J]. 山西农业科学, 2016, 44(10): 1450-1454.
	QIAO X X, FENG M C, YANG W D, et al.. Effect of SG smoothing processing on predicting the above ground dry biomass of winter wheat [J]. J. Shanxi Agric. Sci., 2016, 44(10): 1450-1454.
29	陶惠林, 徐良骥, 冯海宽, 等. 基于无人机高光谱遥感数据的冬小麦产量估算[J].农业机械学报, 2020, 51(7): 146-155.
	TAO H L, XU L J, FENG H K, et al.. Winter wheat yield estimation based on UAV hyperspectral remote sensing data [J]. Trans. Chin. Soc. Agric. Mach., 2020, 51(7): 146-155.
30	吴芳, 李映雪, 张缘园, 等. 基于机器学习算法的冬小麦不同生育时期生物量高光谱估算[J]. 麦类作物学报, 2019, 39(2): 217-224.
	WU F, LI Y X, ZHANG Y Y, et al.. Hyperspectral estimation of biomass of winter wheat at different growth stages based on machine learning algorithms [J]. J. Triticeae Crops, 2019, 39(2): 217-224.
31	李世波, 林辉, 王光明, 等. 基于GF-1的森林蓄积量遥感估测[J].中南林业科技大学学报, 2019, 39(8): 70-75.
	LI S B, LIN H, WANG G M, et al.. Estimation of forest volume based on GF-1 [J]. J. Cent. South Univ. For. Technol., 2019, 39(8): 70-75.
32	朱赟, 申广荣, 项巧巧, 等. 基于不同光谱变换的土壤盐含量光谱特征分析[J]. 土壤通报, 2017, 48(3): 560-568.
	ZHU Y, SHEN G R, XIANG Q Q, et al.. Spectral characteristics of soil salinity based on different pre-processing methods [J]. Acta Pedol. Sin., 2017, 48(3): 560-568.

处理Treatment	样本数 Sample number	叶绿素密度CH.D/（g·m^-2）
处理Treatment	样本数 Sample number	最小值 Minimum	最大值 Maximum	均值 Mean	标准差 Standard deviation	变异系数 Variable coefficient/%
T0	10	0.355	0.576	0.478 b	0.073	16.19
T1	33	0.474	0.809	0.665 a	0.087	13.31
T2	16	0.521	0.704	0.616 a	0.050	8.47
T3	15	0.418	0.729	0.618 a	0.075	12.56
T4	22	0.499	0.878	0.661 a	0.080	12.44

处理Treatment	样本数 Sample number	叶绿素密度CH.D/（g·m^-2）
处理Treatment	样本数 Sample number	最小值 Minimum	最大值 Maximum	均值 Mean	标准差 Standard deviation	变异系数 Variable coefficient/%
T0	10	0.355	0.576	0.478 b	0.073	16.19
T1	33	0.474	0.809	0.665 a	0.087	13.31
T2	16	0.521	0.704	0.616 a	0.050	8.47
T3	15	0.418	0.729	0.618 a	0.075	12.56
T4	22	0.499	0.878	0.661 a	0.080	12.44

序号 Number	植被指数表达形式 Vegetation index expression form
1	R₇₄₅-R₇₄₆
2	R₇₃₉-R₇₄₇
3	R₇₃₉-R₇₄₈
4	R₇₃₂-R₈₀₁
5	R₇₃₃-R₇₉₃
6	（R₇₂₂-R₇₂₃）/（R₇₂₂+R₇₂₃）
7	（R₇₂₃-R₇₂₄）/（R₇₂₃+R₇₂₄）
8	（R₇₀₉-R₇₆₅）/（R₇₀₉+R₇₆₅）
9	（R₇₁₈-R₇₂₃）/（R₇₁₈+R₇₂₃）
10	R₇₁₈/R₇₂₄
11	R₇₁₁/R₇₆₄
12	R₇₀₈/R₇₆₃

序号 Number	植被指数表达形式 Vegetation index expression form
1	R₇₄₅-R₇₄₆
2	R₇₃₉-R₇₄₇
3	R₇₃₉-R₇₄₈
4	R₇₃₂-R₈₀₁
5	R₇₃₃-R₇₉₃
6	（R₇₂₂-R₇₂₃）/（R₇₂₂+R₇₂₃）
7	（R₇₂₃-R₇₂₄）/（R₇₂₃+R₇₂₄）
8	（R₇₀₉-R₇₆₅）/（R₇₀₉+R₇₆₅）
9	（R₇₁₈-R₇₂₃）/（R₇₁₈+R₇₂₃）
10	R₇₁₈/R₇₂₄
11	R₇₁₁/R₇₆₄
12	R₇₀₈/R₇₆₃

指标 Index		偏二小乘模型 PLSR	随机森林模型 RF
建模 Calibration （n=64）	决定系数R²	0.765 4	0.731 6
	均方根误差RMSE	0.046 6	0.045 5
	相对分析误差RPD	1.816 0	1.993 1
预测 Validation （n=32）	决定系数R²	0.730 9	0.723 5
	均方根误差RMSE	0.051 7	0.057 2
	相对分析误差RPD	1.456 9	1.851 4