Hyperspectral Estimation of Tea Leaves Water Content Under the Influence of Dust Retention

Jing JIANG; Zi-wei ZHAO; Chang CAI; Jin-song ZHANG; Zhi-qing CHENG

doi:10.3964/j.issn.1000-0593(2021)11-3532-06

Journals >Spectroscopy and Spectral Analysis >Volume 41 >Issue 11 >Page 3532 > Article

Spectroscopy and Spectral Analysis
Vol. 41, Issue 11, 3532 (2021)

Hyperspectral Estimation of Tea Leaves Water Content Under the Influence of Dust Retention

Jing JIANG^{1、1; 2;}, Zi-wei ZHAO^{1、1; 2;}, Chang CAI^{1、1; 2;}, Jin-song ZHANG^3、3;, and Zhi-qing CHENG^{1、1; 2; *;}

Author Affiliations

¹1. College of Resources and Environment, Anhui Agricultural University, Hefei 230036, China

³3. Chinese Academy of Forestry, Beijing 100091, China

show less

DOI: 10.3964/j.issn.1000-0593(2021)11-3532-06 Cite this Article

Jing JIANG, Zi-wei ZHAO, Chang CAI, Jin-song ZHANG, Zhi-qing CHENG. Hyperspectral Estimation of Tea Leaves Water Content Under the Influence of Dust Retention[J]. Spectroscopy and Spectral Analysis, 2021, 41(11): 3532 Copy Citation Text

show less

Fig. 1. Comparison of spectral reflectance between dust and clean tea leaves

Download full size | View in the Article

Fig. 2. Comparative analysis of the measured and predicted EWT in tea leaves
(a): NDVI_{(1 298, 1 340)}; (b): RVI_{(1 298, 1 325)}; (c): NDWI_{(860, 1 450)}; (d): SIWSI

Download full size | View in the Article

样本	样本数	等效水厚度范围 /(g·cm^-2)	单位滞尘率范围 /(g·cm^-2)
特征波段提取集	200	0.010~0.023	0.067~2.523
建模集	100	0.012~0.020	0.033~1.612
检验集	50	0.012~0.029	0.124~1.9403
总量	350	0.010~0.029	0.033~2.523

Table 1. Informations of samples

View in the Article

植被指数	计算公式	文献
NDVI_{(1 298, 1 340)}	(R_{1 298}-R_{1 340})/(R_{1 298}+R_{1 340})	-
RVI_{(1 298, 1 325)}	R_{1 298}/R_{1 325}	-
NDWI_{(860, 1 450)}	(R₈₆₀-R_{1 450})/(R₈₆₀+R_{1 450})	[2]
NDWI_{(860, 2 130)}	(R₈₆₀-R_{2 130})/(R₈₆₀+R_{2 130})	[2]
NDII	(R₈₀₀-R_{1 600})/(R₈₀₀+R_{1 600})	[2]
MSI	R_{1 600}/R₈₂₀	[2]
SIWSI	(R₈₆₀-R_{1 640})/(R₈₆₀+R_{1 640})	[11]
NDMSI	(R₈₂₀-R_{1 600})/(R₈₂₀+R_{1 600})	[11]

Table 2. Vegetation indexes and calculation methods

View in the Article

植被指数	相关系数		相对变率 n=100
植被指数	无尘(n=100)	有尘(n=100)	相对变率 n=100
NDVI_{(1 298, 1 340)}	0.865^**	0.848^**	0.019
RVI_{(1 298, 1 325)}	0.864^**	0.844^**	0.023
NDWI_{(860, 1 450)}	0.837^**	0.795^**	0.051
NDWI_{(860, 2 130)}	0.805^**	0.752^**	0.068
NDII	0.808^**	0.760^**	0.061
MSI	-0.805^**	-0.748^**	0.073
SIWSI	0.801^**	0.754^**	0.060
NDMSI	0.806^**	0.757^**	0.062

Table 3. Correlation coefficients of EWT with various indices in dust and clean tea leaves

View in the Article

植被指数	无尘回归方程	R²	有尘回归方程	R²	R²相对变率
NDVI_{(1 298, 1 340)}	EWT=0.291NDVI_{(1 298, 1 340)}+0.002	0.870	EWT=0.268NDVI_{(1 298, 1 340)}+0.004	0.812	0.069
RVI_{(1 298, 1 325)}	EWT=0.253RVI_{(1 298, 1 325)}-0.249	0.873	EWT=0.233RVI_{(1 298, 1 325)}-0.228	0.814	0.070
NDWI_{(860, 1 450)}	EWT=0.038NDWI_{(860, 1 450)}-0.003	0.799	EWT=0.031 NDWI_{(860, 1 450)}+0.0004	0.741	0.075
SIWSI	EWT=0.066SIWSI+0.003	0.770	EWT=0.049SIWSI+0.007	0.704	0.089

Table 4. Estimation and evaluation for EWT model using vegetation indexes of tea leaves under clean and dust conditions

View in the Article

植被指数	无尘			有尘			R²相对变率
植被指数	估算值与实测值拟合方程	R²	RMSE	估算值与实测值拟合方程	R²	RMSE	R²相对变率
NDVI_{(1 298, 1 340)}	y=0.225x+0.005	0.719	0.001	y=0.777x+0.003	0.711	0.001	0.011
RVI_{(1 298, 1 325)}	y=0.871x+0.002	0.718	0.001	y=0.733x+0.004	0.694	0.001	0.033
NDWI_{(860, 1 450)}	y=0.899x+0.002	0.667	0.001	y=0.696x+0.005	0.649	0.001	0.027
SIWSI	y=0.679x+0.006	0.565	0.001	y=0.689x+0.005	0.534	0.001	0.055

Table 5. Comparative analysis of the measured and predicted values of EWT in tea leaves

View in the Article

植被指数	(有尘+无尘)回归方程	R²
NDVI_{(1 298, 1 340)}	EWT=0.278NDVI_{(1 298, 1 340)}+0.006	0.837
RVI_{(1 298, 1 325)}	EWT=0.245RVI_{(1 298, 1 325)}-0.241	0.853
NDWI_{(860, 1 450)}	EWT=0.033NDWI_{(860, 1 450)}-0.001	0.752
SIWSI	EWT=0.053SIWSI+0.006	0.688

Table 6. Estimated models for EWT of tea leaf by using vegetation indexes under mixed conditions

Download Citation

Save the article for my favorites

Paper Information