Open Access
Add to BibSonomyAbstract
Paddy rice is one of the most important crops in China, and leaf nitrogen content (LNC) serves as a significant indictor for monitoring crop status. A reliable method is needed for precise and fast quantification of LNC. Laser-induced fluorescence (LIF) technology and reflectance spectra of crops are widely used to monitor leaf biochemical content. However, comparison between the fluorescence and reflectance spectra has been rarely investigated in the monitoring of LNC. In this study, the performance of the fluorescence and reflectance spectra for LNC estimation was discussed based on principal component analysis (PCA) and back-propagation neural network (BPNN). The combination of fluorescence and reflectance spectra was also proposed to monitor paddy rice LNC. The fluorescence and reflectance spectra exhibited a high degree of multi-collinearity. About 95.38%, and 97.76% of the total variance included in the spectra were efficiently extracted by using the first three PCs in PCA. The BPNN was implemented for LNC prediction based on new variables calculated using PCA. The experimental results demonstrated that the fluorescence spectra (R2 = 0.810, 0.804 for 2014 and 2015, respectively) are superior to the reflectance spectra (R2 = 0.721, 0.671 for 2014 and 2015, respectively) for estimating LNC based on the PCA-BPNN model. The proposed combination of fluorescence and reflectance spectra can greatly improve the accuracy of LNC estimation (R2 = 0.912, 0.890 for 2014 and 2015, respectively).
© 2016 Optical Society of America
1. Introduction
Paddy rice, one of the three major crops in China, is cultivated in approximately 30 million hectare land annually [1, 2]. A large amount of nitrogen (N) fertilizer is consumed to improve paddy rice yield. The N fertilizer consumption in China occupies about 35% of the total N fertilizer consumption worldwide per year. This condition implies that large amounts of N fertilizer are wasted and enter the environmental system, resulting in serious atmospheric and environmental pollution [3, 4], including emission of N2O and CH4, water eutrophication, and nitrate leaching in soil. Currently, precise monitoring of leaf N content (LNC) is an effective strategy to match the N fertilizer rate with crop N demand in both temporal and spatial dimensions [5–9].
In the last decades, numerous passive and active remote sensing technologies have been proposed to detect LNC in cereal crops [10–13]. Passive remote sensing is based on the crops reflectance spectra, which are closely related to the photosynthetic ability of plants and foliage chlorophyll concentration [14–16]. This sensing technique can provide effective information regarding the biophysical and biochemical composition of a crop leaf. In many cases, reflectance spectrometry can be as accurate as traditional wet-chemical procedures when crop is dried and ground [14]. Therefore, several vegetation indices (VIs) have been proposed; the correlation between VIs and LNC has also been analyzed to estimate the latter [17–19]. Passive remote sensing has been maturely applied in satellite platform and provides valuable information in agricultural production, national defense, and vegetation monitoring; however, the estimation accuracy of LNC is remains because of the asymptotic saturation problem [20].
On the other hand, laser-induced fluorescence (LIF) was proposed by Chappelle to monitor the growth status of crops; the mechanism of LIF differs from that of reflectance [21]. LIF occurs when the energy of a specific wavelength is absorbed by fluorophore, and a part of the energy is vanished by light emission at longer wavelengths within a short time [22]. In the vegetation, chlorophyll is a typical fluorophore and is closely related to the photosynthesis of green plants. Therefore chlorophyll fluorescence is widely utilized to monitor the photosynthetic activity of plants and evaluate the effect of various stress factors on it [23]. Relative investigations have been conducted to estimate crop LNC by using different fluorescence parameters based LIF technology [12, 24]. This technology displayed the advantages of rapid, non-destructive, and highly sensitivity. Therefore, LIF, as a powerful type of laser remote sensing technology, has been extensively implemented for crop monitoring [25–28].
In addition, multivariate statistical methods, such as back-propagation neural networks (BPNN) and principal component analysis (PCA), have also been employed for quantitative remote sensing of the biochemical content of vegetation, which can greatly improve the accuracy of crop monitoring [29–31]. BPNN can describe complex and intricate relationships between spectral information and various crop conditions [32]. PCA has become a useful tool for the induction of data dimensionality by analyzing major attributes. PCA uses few variables obtained by a linear combination of the original data to explain the most important variable information [33]. Thus, PCA is utilized to analyze the spectra, and BPNN is implemented to inverse the LNC based on the new variable calculated by PCA.
At present, a large amount of investigations have been conducted to monitoring crop LNC based on LIF technology or reflectance spectra [5, 8]. However, comparative investigations on the performance of fluorescence and reflectance spectra for monitoring of crop LNC remain limited. In addition, there are no relative literatures that have focused on the study of crop LNC estimation based on the combined fluorescence and reflectance spectra. Thus, the present study mainly aims to: (1) compare the performance of fluorescence and reflectance spectra in estimating paddy rice LNC from the leaf level by using the PCA-BPNN model; and (2) analyze whether the proposed combination of fluorescence and reflectance spectra can improve the LNC estimation accuracy of paddy rice.
2. Materials and methods
2.1 Study areas and site description
The experiment was performed in the Jianghan China Plain during the rice cultivating seasons of 2014 and 2015. The latitude of this area varies from29°58′ N to 31°22′ N and the longitude ranges from 113°41′ E to 115°05′ E. The rainfall and sunshine duration are above 1200 mm and over 1800 hours per year, respectively. Therefore, the area is ideal for cultivating paddy rice [34].
In 2014, the paddy rice varieties grown were Yongyou4949 and cultivated in Junchuan County, Suizhou City in the province of Hubei, China. These cultivars were seeded on April 27 and transplanted to the field on June 1. During the entire cultivation period, four different levels of urea fertilizer were applied (0, 189, 270, and 351 kg/ha) in the experimental fields. Urea fertilizer was divided into four splits: 30% at seeding, 20% at tillering, 25% at shooting, and 25% at booting. For each fertilizer treatment, the experimental field had an absolute block design with three replications under the same cultivation conditions. Other management procedures were advised by the local farm extension service in rice production. The leaf samples were collected on July 15, 2014 which corresponded to the tillering stage.
In the growing season of 2015, Yangliangyou 6 was grown in the experimental station of Huazhong Agricultural University (HAU) in Wuhan City, China. The crops were seeded on April 30, and were transplanted to the field on May 27. During the entire cultivation period, four different urea levels were implemented (0, 120, 180, and 240 kg/ha). Urea fertilizer was divided into three splits: 60% at seeding, 20% at tillering, and 20% at shooting. The experimental field had an absolute block design with three replications for each treatment. Other management procedures were advised by the local farm extension service in rice production. Leaf samples of paddy rice were collected on July 24 and 26, 2015, which corresponded to the tillering stage.
2.2 Measurement of leaf reflectance spectra
The ASD FieldSpec Pro FR (Analytical Spectral Devices, Inc., Boulder, USA), a commercial passive instrument, was implemented for leaf spectral measurements. Spectral measurement was performed process followed the investigation of Pu et al. [35]. A 100 W halogen reflectorized lamp was served as light source, and each leaf sample was measured thrice to get an average spectral curve for each sample. All spectral reflectance were measured at the nadir direction of the radiometer with a 25° field of view (FOV). Leaf radiance was measured initially within 350-1000 nm with a 1.4 nm sampling interval and 1000-2500 nm with a 2.2 nm sampling interval. The entire spectra (350-2500 nm) were resampled automatically at 1 nm spectral resolution. Raw leaf radiance was converted to spectral reflectance by referencing a standard reference whiteboard (Spectralon, Labsphere, Inc., North Sutton, NH, USA, 10 cm × 10 cm, reflectance nearly 99%) acquired several times during the procedure of leaf radiance measurements [34]. The leaf reflectance spectrum can be calculated as follows:
where, and denote the reference standard radiance and leaf radiance at wavelength , respectively.2.3 Measurements of leaf fluorescence spectra
The schematic of the LIF measurement system is illustrated in Fig. 1. The ultraviolet excitation light source is a 355 nm laser emitted by a neodymium-doped yttrium aluminum garnet (Nd:YAG) laser and a third harmonic generation. The LIF measurement system works as follows: excitation light (355 nm) was transmitted from the beam expander (10 times at 355 nm), completely reflected by M1 and M2, and transmitted perpendicularly to the targets. The back-emission LIF signal was collected by utilizing a Maksutov-Cassegrain telescope. A single-mode optical fiber with 200 µm diameter was utilized to transmit the fluorescence collected between the telescope and spectrograph (Princeton Instrument SP2500i with a spectral resolution of 0.5 nm). The excited fluorescence entered the spectrometer and was detected by implementing the intensified charge coupled device (ICCD) camera. The data were stored in a personal computer for post-processing. In this system, an additional long-pass filter (Semrock BLP01-355R-25 with an edge of 361 nm and 93% transmittance at 364.9-900 nm) was placed behind the telescope and used to eliminate the reflected light from the laser entering the optical fiber. The fluorescence spectra range from 360 nm to 800 nm, with 0.5 nm of sampling interval.
Fig. 1 Schematic of the LIF measurement system. BE: 10 times beam expander at 355 nm; M1 and M2: completely reflecting mirrors; F: filter; OFP: optical fiber probe; OF: optical fiber; ICCD: intensified charge coupled device.
2.4 Measurement of leaf nitrogen content
In each experimental field, paddy rice leaves were destructively sampled by stochastically cutting six fully expanded the second leaves from the top with three replicates. These samples were sealed in plastic bags, kept in an ice chest, and immediately transported to the laboratory for reflectance and fluorescence spectral measurements [34]. After the measurements, all samples were immediately sent to Wuhan Academy of Agricultural Science and Technology for measuring LNC. LNC was determined using the traditional Keldjahl method [36, 37].
2.5 Analytical method
PCA is widely used in a large number of fields and is one of the most efficient statistical multivariate analysis approaches. The main task of PCA is to extract the most significant characteristic variables and reduce the dimensionality of the spectra by analyzing the internal correlation of data [38]. That is to say, the number of variables can be reduced by removing the lower-level components without any notable loss of information contained in the original data set by PCA. The computed new variables called principal components (PC) are obtained as linear combinations of the original variables.
where, Xi denotes the measured spectral values at ith wavelength, Yj corresponds to the principal component, and represents the sum of kth principal components for ith wavelength, P is the spectral loadings. PCA can implement fewer new variables than the original ones to simplify the analysis [39]. In the present investigation, the spectra usually include a great deal of irrelevant, redundant information. Therefore, PCA was employed to discuss the spectra.BPNN, a supervised training algorithm, is an efficient tool for the prediction of nonlinearities and comprises individual processing units called neurons, which resemble neural activity [40]. All independent neurons can act together, and this condition allows them to analyze and solve various complex tasks simultaneously [41]. BPNN has been widely utilized in pattern recognition, classification, agriculture, and biology [41, 42], and a brief introduction to BPNN can be referenced [41, 43]. Therefore, the BPNN model was implemented to analyze the performance of the combined reflectance and fluorescence spectra in the estimation of paddy rice LNC in the present work. In the network, the biases () and network weights () were adjusted along with the gradient-decrease of mean squared error (MSE). The transformation T in Eq. (3) is a nonlinear activation function that could process iteration and provide an out-value y of the networks.
In the present investigation, a simple three layer BPNN was utilized. This network architecture consisted of three parts: one input layer, one hidden layer and one output layer. The Levenberg-Marquardt algorithm was used as the train function.
The reflectance and fluorescence characteristics were randomly divided into three equal parts, and leave-one-out cross validation was utilized. Two-thirds of the data were utilized to train the BPNN model, and the remaining one-third was utilized for testing. This procedure was repeated three times, utilizing a different third of the spectra characteristic as test sets each time. In addition, the coefficient of determination (R2), root mean square error (RMSE), and relative error (RE) in the prediction were implemented to analyze the performance of the prediction. RMSE and RE can be written as follows:
where n is the number of samples, Mi denotes the measured values, and Pi represents the predicted values. corresponds to the mean of the measured values, and RE is the relative difference between the predicted and observed values. High R2 and low RMSE and RE denote high precision and accuracy of the model to predict LNC [44].
3. Results and discussion
3.1 The spectral analysis
Spectral characteristics were analyzed for different LNC levels according to the reflectance and fluorescence spectra. The spectra were selected and divided into six different groups based on LNC levels (Table 1) [45].
Table 1. Data were divided into six groups according to LNC, and LNC changing range.
An average spectral curve for each LNC level was obtained based on the six different groups. As shown in Fig. 2, the reflectance and fluorescence spectra of the leaf displayed certain differences among the spectral characteristics with changing LNC levels.
Fig. 2 Spectra changing with different leaf nitrogen content levels. (a): Reflectance spectra; and (b): fluorescence spectra.
As shown in Fig. 2(a), the reflectance spectral differences among different LNC levels of leaf samples are significant obvious within the 500-1300 nm. The different LNC levels of paddy rice can be distinguished based on spectral characteristics obtained using the ASD. For the range of visible wavebands (500-700 nm), the rice leaf reflectance decreases with high LNC. This result is likely due to the large absorption of various leaf pigments [46]. The reflectance spectra in the region of near infrared elevated and displayed greater obvious differences among different LNC levels than those in visible wavebands. Hence, these spectral ranges are relatively sensitive to variations in paddy rice LNC. Overall, the dynamic change patterns of leaf reflectance under varied LNC provide a basis for monitoring the LNC of crops under different LNC levels [36].
As shown in Fig. 2(b), all fluorescence spectra were normalized to 1 at 460 nm. These fluorescence spectra exhibited three main fluorescence peaks at 460, 685, and 740 nm, and a peak shoulder at 525 nm. According to Chappelle et al. [47], the fluorescence peaks at 685 and 740 nm are attributed to chlorophyll a and b, respectively. Nicotinamide adenine dinucleotide (NADPH) and riboflavin are responsible for the fluorescence peak at 460 nm and the peak shoulder at 525 nm, respectively. In accordance with previous investigations [25, 28], LNC is closely related to the fluorescence peaks (685 and 740 nm). Figure 2(b) shows that the intensity of the fluorescence peaks significantly varied with increasing LNC. In contrast to that reported by McMurtrey et al. [48], the fluorescence spectra exhibited a similar changing tendency in the present study. Therefore, LIF technology can be employed to monitor alterations in LNC.
3.2 The principle component analysis (PCA) of spectra
The spectra usually include a great deal of irrelevant, redundant information (Fig. 3). The spectral data exhibited a high degree of multi-collinearity and may influence the predictability when all spectral data were utilized to estimate LNC of paddy rice. PCA was then implemented to reduce the dimensionality of the spectra and extract key characteristic variables.
Fig. 3 Pearson’s correlation analysis of all spectra combination of 2014 and 2015 for different measurement techniques. (a): Reflectance spectra; and (b): fluorescence spectra.
In this investigation, all spectra including fluorescence and reflectance spectra were analyzed using PCA. The experimental results demonstrated that when the number of PCs exceeded three, the increase of the explained variance with additional PC was reduced to less than 1%. Thus, the first three PCs were retained for further study. The explained variables and cumulative variances for the reflectance and fluorescence spectra are listed in Table 2.
Table 2. Percentage of explained variance for the first three PCs. PC1, PC2 and PC3 capture the most significant information contained in the spectral variation measured by different ways. EV: explained variance; CV: cumulative variance.
As shown in Table 2, 95.38%, and 97.76% of the total variances included in the spectra can be explained using the first three PCs. These variances correspond to reflectance and fluorescence spectra, respectively. In addition, other PCs contained less spectral information, which can be ignored without any notable loss of information contained in the original data set. The loading plots of the first three PCs are shown in Fig. 4 to comprehend the efficiency of PCA in describing the spectra.
Fig. 4 The loading weights of the first three principal components with different measured spectra. (a): Reflectance spectra; (b): fluorescence spectra.
Figure 4 shows that the first three PCs captured the most significant information in the fluorescence spectra, which can be applied for further analysis. The factor scores calculated from the first three PCs were utilized as the input variables of the PCA-BPNN models for predicting paddy rice LNC.
3.3 Estimation of LNC based on fluorescence and reflectance spectra
The PCA-BPNN model was utilized to estimate LNC based on three different spectral parameters (reflectance, fluorescence, and combination with reflectance and fluorescence) to compare the predictive capacity of the reflectance and fluorescence spectra in the estimation of LNC. In the PCA-BPNN model, the numbers of neurons in the output, hidden, and input layers were one, seven, and three, respectively. The measured spectra by using different instruments were randomly divided into three equal parts, and three-fold cross validation was used. On the basis of different input variables, the relationship between the predicted and observed LNC is established and are illustrated in Fig. 5.
Fig. 5 Relationship between the predicted LNC by using the combination of PCA and BPNN and measured LNC based on different parameters for different growing years. (a), (b), and(c): 2014 (n = 72); (d), (e), and (f): 2015 (n = 144); (a) and (d): Reflectance; (b) and (e): fluorescence; (c) and (f): reflectance combined with fluorescence. The dotted line represents the 1:1 line. The red line denotes the fitted curves for regression models.
As shown in Fig. 5, the predicted and observed LNC values were nearly in accordance with the line of 1:1 (the dotted line). Ideally, the red solid line, which is the linear regression between the predicted and observed values, should be coincided with the 1:1 line. By comparing the R2 and RMSE, Fig. 5 demonstrates that the fluorescence spectra (R2 = 0.81, 0.804 for 2014 and 2015, respectively) are superior to the reflectance spectra (R2 = 0.721, 0.671 for 2014 and 2015, respectively). The possible interpretation is that the fluorescence is related to the leaf internal fluorophore which was induced by the excitation light source and emitted light at longer wavelengths. Relative investigations demonstrated that LIF displays high measurement accuracy and is unsusceptible to ambient light [49]. Thus, LIF technology has been widely utilized to detect chlorophyll fluorescence for crop nutrient stress and has become a focus of studies on the laser remote sensing [12, 26, 50]. Compared with using fluorescence or reflectance to monitor LNC, the combined LIF technology and reflectance spectra can provide more accurate estimation of LNC. For 2014 and 2015, the R2 of regression analysis between the predicted and measured LNC can reach up to 0.912 and 0.89, respectively. The satisfactory results of the PCA-BPNN model based on the combination of fluorescence and reflectance demonstrated the promising potential of LIF technology combined with reflectance spectra for the estimation of paddy rice LNC.
In this study, the fluorescence and reflectance spectra were compared to estimate paddy rice LNC, and the combination of fluorescence and reflectance was proposed to monitor LNC. The fluorescence spectra combined with the reflectance spectra exhibited promising potential for monitoring crops LNC. However, this preliminary study only compared the spectra by using the PCA-BPNN model. Some limitations should be considered in further work. For the reflectance spectra, the VIs should be further discussed and the correlation between the VIs and LNC should be established. In addition, the fluorescence parameters should also be further analyzed in comparison studies. Although different nitrogen treatments were set in the present study, additional paddy rice cultivars, growth seasons, and other crops should also be considered in future studies to improve the generalization capability of the proposed approach and obtain a solid conclusion [39].
4. Conclusion
This investigation mainly aims to compare the performance of fluorescence and reflectance spectra for the estimation of LNC and explore the potential of the combined fluorescence and reflectance spectra for estimating LNC based on PCA-BPNN model. The spectra including fluorescence and reflectance displayed a high degree of multi-collinearity, and the PCA can efficiently extract 95.38%, and 97.76% of the total variances included in the spectra by using the first three PCs. The BPNN model was utilized to predict LNC based on the calculated new variables by using PCA. The experimental results demonstrated that the fluorescence spectra (for 2014 and 2015, the R2 was 0.810 and 0.804, respectively) are superior to reflectance spectra (for 2014 and 2015, the R2 was 0.721 and 0.671, respectively) for estimating LNC. In addition, the proposed combination of the fluorescence and reflectance spectra can significant improve the accuracy of LNC estimation (for 2014 and 2015, the R2 was 0.912 and 0.890, respectively). Thus, the research results revealed that the LNC of paddy rice can be accurately evaluated by implementing the LIF technology combined with hyperspectral reflectance. However, to obtain a more solid conclusion, more investigations are still need to be conducted with VIs and additional crops cultivars in the following works.
Funding
National Natural Science Foundation of China (Grant No. 41571370; 41127901); Natural Science Foundation of Hubei Province (Grant No. 2015CFA002); and Open Research Fund of State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing (Grant No.15R01).
Acknowledgments
The authors wish to thank College of Plant Science & Technology of Huazhong Agricultural University for providing the experimental samples and wish to thank Wuhan Academy of Agricultural Science & Technology for providing the LNCs of samples.
References and links
1. X. Liang, H. Li, S. Wang, Y. Ye, Y. Ji, G. Tian, C. van Kessel, and B. Linquist, “Nitrogen management to reduce yield-scaled global warming potential in rice,” Field Crops Res. 146, 66–74 (2013). [CrossRef]
2. M. Huang, W. Zhang, L. Jiang, and Y. Zou, “Impact of temperature changes on early-rice productivity in a subtropical environment of China,” Field Crops Res. 146, 10–15 (2013). [CrossRef]
3. J. Zou, Y. Huang, X. Zheng, and Y. Wang, “Quantifying direct N2O emissions in paddy fields during rice growing season in mainland China: dependence on water regime,” Atmos. Environ. 41(37), 8030–8042 (2007). [CrossRef]
4. P. Nugroho, M. Shimizu, H. Nakamato, A. Nagatake, S. Suwardi, U. Sudadi, and R. Hatano, “Nitrous oxide fluxes from soil under different crops and fertilizer management,” Plant Soil Environ. 61, 385–392 (2015).
5. M. Diacono, P. Rubino, and F. Montemurro, “Precision nitrogen management of wheat. A review,” Agron. Sustain. Dev. 33(1), 219–241 (2013). [CrossRef]
6. F. Li, M. L. Gnyp, L. Jia, Y. Miao, Z. Yu, W. Koppe, G. Bareth, X. Chen, and F. Zhang, “Estimating N status of winter wheat using a handheld spectrometer in the North China Plain,” Field Crops Res. 106(1), 77–85 (2008). [CrossRef]
7. J. Yang, S. Shi, W. Gong, L. Du, Y. Y. Ma, B. Zhu, and S. L. Song, “Application of fluorescence spectrum to precisely inverse paddy rice nitrogen content,” Plant Soil Environ. 61(4), 182–188 (2016). [CrossRef]
8. A. Oukarroum, F. Bussotti, V. Goltsev, and H. M. Kalaji, “Correlation between reactive oxygen species production and photochemistry of photosystems I and II in Lemna gibba L. plants under salt stress,” Environ. Exp. Bot. 109, 80–88 (2015). [CrossRef]
9. X. Jin, G. Yang, C. Tan, and C. Zhao, “Effects of nitrogen stress on the photosynthetic CO2 assimilation, chlorophyll fluorescence, and sugar-nitrogen ratio in corn,” Sci. Rep. 5, 9311 (2015). [CrossRef] [PubMed]
10. V. Aleksandrov, V. Krasteva, M. Paunov, M. Chepisheva, M. Kousmanova, H. M. Kalaji, and V. Goltsev, “Deficiency of some nutrient elements in bean and maize plants analyzed by luminescent method,” Bulg. J. Agric. Sci. 20, 24–30 (2014).
11. W. Gong, S. L. Song, B. Zhu, S. Shi, F. Li, and X. Cheng, “Multi-wavelength canopy LiDAR for remote sensing of vegetation: Design and system performance,” ISPRS J. Photogramm. 69, 1–9 (2012). [CrossRef]
12. M. P. Cendrero-Mateo, M. S. Moran, S. A. Papuga, K. R. Thorp, L. Alonso, J. Moreno, G. Ponce-Campos, U. Rascher, and G. Wang, “Plant chlorophyll fluorescence: active and passive measurements at canopy and leaf scales with different nitrogen treatments,” J. Exp. Bot. 67(1), 275–286 (2016). [CrossRef] [PubMed]
13. J. Wang, X. Xiao, Y. Qin, J. Dong, G. Zhang, W. Kou, C. Jin, Y. Zhou, and Y. Zhang, “Mapping paddy rice planting area in wheat-rice double-cropped areas through integration of Landsat-8 OLI, MODIS, and PALSAR images,” Sci. Rep. 5, 10088 (2015). [CrossRef] [PubMed]
14. B. J. Yoder and R. E. Pettigrew-Crosby, “Predicting nitrogen and chlorophyll content and concentrations from reflectance spectra (400–2500 nm) at leaf and canopy scales,” Remote Sens. Environ. 53(3), 199–211 (1995). [CrossRef]
15. M. Zivcak, H. M. Kalaji, H. B. Shao, K. Olsovska, and M. Brestic, “Photosynthetic proton and electron transport in wheat leaves under prolonged moderate drought stress,” J. Photochem. Photobiol. B 137, 107–115 (2014). [CrossRef] [PubMed]
16. H. Croft, J. Chen, Y. Zhang, and A. Simic, “Modelling leaf chlorophyll content in broadleaf and needle leaf canopies from ground, CASI, Landsat TM 5 and MERIS reflectance data,” Remote Sens. Environ. 133, 128–140 (2013). [CrossRef]
17. J.-B. Feret, C. François, G. P. Asner, A. A. Gitelson, R. E. Martin, L. P. Bidel, S. L. Ustin, G. le Maire, and S. Jacquemoud, “PROSPECT-4 and 5: Advances in the leaf optical properties model separating photosynthetic pigments,” Remote Sens. Environ. 112(6), 3030–3043 (2008). [CrossRef]
18. S. L. Ustin, A. A. Gitelson, S. Jacquemoud, M. Schaepman, G. P. Asner, J. A. Gamon, and P. Zarco-Tejada, “Retrieval of foliar information about plant pigment systems from high resolution spectroscopy,” Remote Sens. Environ. 113, S67–S77 (2009). [CrossRef]
19. J.-B. Féret, C. François, A. Gitelson, G. P. Asner, K. M. Barry, C. Panigada, A. D. Richardson, and S. Jacquemoud, “Optimizing spectral indices and chemometric analysis of leaf chemical properties using radiative transfer modeling,” Remote Sens. Environ. 115(10), 2742–2750 (2011). [CrossRef]
20. K. Zhao, D. Valle, S. Popescu, X. Zhang, and B. Mallick, “Hyperspectral remote sensing of plant biochemistry using Bayesian model averaging with variable and band selection,” Remote Sens. Environ. 132, 102–119 (2013). [CrossRef]
21. E. W. Chappelle, F. M. Wood, J. E. McMurtrey, and W. W. Newcomb, “Laser-induced fluorescence of green plants. 1: A technique for the remote detection of plant stress and species differentiation,” Appl. Opt. 23(1), 134–138 (1984). [CrossRef] [PubMed]
22. N. Tremblay, Z. Wang, and Z. G. Cerovic, “Sensing crop nitrogen status with fluorescence indicators. A review,” Agron. Sustain. Dev. 32(2), 451–464 (2012). [CrossRef]
23. H. K. Lichtenthaler and U. Rinderle, “The role of chlorophyll fluorescence in the detection of stress conditions in plants,” Crit. Rev. Anal. Chem. 19(S1), S29–S85 (1988). [CrossRef]
24. J. Yang, W. Gong, S. Shi, L. Du, J. Sun, Y. Ma, and S. Song, “Accurate identification of nitrogen fertilizer application of paddy rice using laser-induced fluorescence combined with support vector machine,” Plant Soil Environ. 61(11), 501–506 (2016). [CrossRef]
25. A. A. Gitelson, C. Buschmann, and H. K. Lichtenthaler, “The chlorophyll fluorescence ratio F735/F700 as an accurate measure of the chlorophyll content in plants,” Remote Sens. Environ. 69(3), 296–302 (1999). [CrossRef]
26. J. Yang, W. Gong, S. Shi, L. Du, J. Sun, S. Song, B. Chen, and Z. Zhang, “Analyzing the performance of fluorescence parameters in the monitoring of leaf nitrogen content of paddy rice,” Sci. Rep. 6, 28787 (2016). [CrossRef] [PubMed]
27. M. Brestic, M. Zivcak, K. Olsovska, H. B. Shao, H. M. Kalaji, and S. I. Allakhverdiev, “Reduced glutamine synthetase activity plays a role in control of photosynthetic responses to high light in barley leaves,” Plant Physiol. Biochem. 81, 74–83 (2014). [CrossRef] [PubMed]
28. J. Yang, W. Gong, S. Shi, L. Du, J. Sun, and S. L. Song, “Estimation of nitrogen content based on fluorescence spectrum and principal component analysis in paddy rice,” Plant Soil Environ. 62(4), 178–183 (2016). [CrossRef]
29. P. M. Atkinson and A. Tatnall, “Introduction neural networks in remote sensing,” Int. J. Remote Sens. 18(4), 699–709 (1997). [CrossRef]
30. N. Neeti and J. R. Eastman, “Novel approaches in Extended Principal Component Analysis to compare spatio-temporal patterns among multiple image time series,” Remote Sens. Environ. 148, 84–96 (2014). [CrossRef]
31. E. H. Czyzowska-Wisniewski, W. J. van Leeuwen, K. K. Hirschboeck, S. E. Marsh, and W. T. Wisniewski, “Fractional snow cover estimation in complex alpine-forested environments using an artificial neural network,” Remote Sens. Environ. 156, 403–417 (2015). [CrossRef]
32. M. Buscema, “Back propagation neural networks,” Subst. Use Misuse 33(2), 233–270 (1998). [CrossRef] [PubMed]
33. R. Bro and A. K. Smilde, “Principal component analysis,” Anal. Methods 6(9), 2812–2831 (2014). [CrossRef]
34. S. Song, W. Gong, B. Zhu, and X. Huang, “Wavelength selection and spectral discrimination for paddy rice, with laboratory measurements of hyperspectral leaf reflectance,” ISPRS J. Photogramm. 66(5), 672–682 (2011). [CrossRef]
35. R. Pu, S. Ge, N. Kelly, and P. Gong, “Spectral absorption features as indicators of water status in coast live oak (Quercus agrifolia) leaves,” Int. J. Remote Sens. 24(9), 1799–1810 (2003). [CrossRef]
36. X. Yao, Y. Zhu, Y. Tian, W. Feng, and W. Cao, “Exploring hyperspectral bands and estimation indices for leaf nitrogen accumulation in wheat,” Int. J. Appl. Earth Obs. Geoinf. 12(2), 89–100 (2010). [CrossRef]
37. Y. C. Tian, X. Yao, J. Yang, W. X. Cao, D. B. Hannaway, and Y. Zhu, “Assessing newly developed and published vegetation indices for estimating rice leaf nitrogen concentration with ground- and space-based hyperspectral reflectance,” Field Crops Res. 120(2), 299–310 (2011). [CrossRef]
38. L. S. Galvão, M. A. Pizarro, and J. C. N. Epiphanio, “Variations in reflectance of tropical soils: spectral-chemical composition relationships from AVIRIS data,” Remote Sens. Environ. 75(2), 245–255 (2001). [CrossRef]
39. Q. X. Yi, J. F. Huang, F. M. Wang, X. Z. Wang, and Z. Y. Liu, “Monitoring rice nitrogen status using hyperspectral reflectance and artificial neural network,” Environ. Sci. Technol. 41(19), 6770–6775 (2007). [CrossRef] [PubMed]
40. R. Zong, Y. Zhi, B. Yao, J. Gao, and A. A. Stec, “Classification and identification of soot source with principal component analysis and back-propagation neural network,” Aust. J. Forensic Sci. 46(2), 224–233 (2014). [CrossRef]
41. A. I. Samborska, V. Alexandrov, L. Sieczko, B. Kornatowska, V. Goltsev, D. C. Magdalena, and H. M. Kalaji, “Artificial neural networks and their application in biological and agricultural research,” J. Nano. Photo. Bio. Sci. 2, 14–30 (2014).
42. V. Goltsev, I. Zaharieva, P. Chernev, M. Kouzmanova, H. M. Kalaji, I. Yordanov, V. Krasteva, V. Alexandrov, D. Stefanov, S. I. Allakhverdiev, and R. J. Strasser, “Drought-induced modifications of photosynthetic electron transport in intact leaves: analysis and use of neural networks as a tool for a rapid non-invasive estimation,” Biochim. Biophys. Acta 1817(8), 1490–1498 (2012). [CrossRef] [PubMed]
43. L. E. Keiner and X. H. Yan, “A neural network model for estimating sea surface chlorophyll and sediments from thematic mapper imagery,” Remote Sens. Environ. 66(2), 153–165 (1998). [CrossRef]
44. W. Feng, X. Yao, Y. Zhu, Y. Tian, and W. Cao, “Monitoring leaf nitrogen status with hyperspectral reflectance in wheat,” Eur. J. Agron. 28(3), 394–404 (2008). [CrossRef]
45. Y. C. Tian, K. J. Gu, X. Chu, X. Yao, W. X. Cao, and Y. Zhu, “Comparison of different hyperspectral vegetation indices for canopy leaf nitrogen concentration estimation in rice,” Plant Soil 376(1-2), 193–209 (2014). [CrossRef]
46. D. Stroppiana, M. Boschetti, P. A. Brivio, and S. Bocchi, “Plant nitrogen concentration in paddy rice from field canopy hyperspectral radiometry,” Field Crops Res. 111(1-2), 119–129 (2009). [CrossRef]
47. E. W. Chappelle, J. E. McMurtrey III, and M. S. Kim, “Identification of the pigment responsible for the blue fluorescence band in the laser induced fluorescence (LIF) spectra of green plants, and the potential use of this band in remotely estimating rates of photosynthesis,” Remote Sens. Environ. 36(3), 213–218 (1991). [CrossRef]
48. J. McMurtrey III, E. Chappelle, M. Kim, J. Meisinger, and L. Corp, “Distinguishing nitrogen fertilization levels in field corn (Zea mays L.) with actively induced fluorescence and passive reflectance measurements,” Remote Sens. Environ. 47(1), 36–44 (1994). [CrossRef]
49. W. Wan, D. Hua, J. Le, T. He, Z. Yan, and C. Zhou, “Study of laser-induced chlorophyll fluorescence lifetime measurement and its correction,” Measurement 60, 64–70 (2015). [CrossRef]
50. M. Živcak, K. Olsovska, P. Slamka, J. Galambošová, V. Rataj, H. Shao, and M. Brestič, “Application of chlorophyll fluorescence performance indices to assess the wheat photosynthetic functions influenced by nitrogen deficiency,” Plant Soil Environ. 60, 210–215 (2014).
