Next Article in Journal
Thinned and Sparse Beamforming for Semicircular FDAs in the Transmit–Receive Domain
Previous Article in Journal
Construction and Application of Dynamic Threshold Model for Agricultural Drought Grades Based on Near-Infrared and Short-Wave Infrared Bands for Spring Maize
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 16
Issue 17
10.3390/rs16173261
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Leaf Nitrogen and Phosphorus Variation and Estimation of Citrus Tree under Two Labor-Saving Cultivation Modes Using Hyperspectral Data

by
Dasui Li
1,
Qingqing Hu
1,
Jinzhi Zhang
1,2,
Yuanyong Dian
1,3,
Chungen Hu
1,2 and
Jingjing Zhou
1,3,*
1
College of Horticulture & Forestry Sciences, Huazhong Agricultural University, Wuhan 430070, China
2
National Key Laboratory for Germplasm Innovation & Utilization of Horticultural Crops, Wuhan 430070, China
3
Hubei Engineering Technology Research Center for Forestry Information, Wuhan 430070, China
*
Author to whom correspondence should be addressed.
Remote Sens. 2024, 16(17), 3261; https://doi.org/10.3390/rs16173261
Submission received: 13 July 2024 / Revised: 25 August 2024 / Accepted: 28 August 2024 / Published: 3 September 2024
Browse Figures
Versions Notes

Abstract

:
Understanding canopy nitrogen (N) and phosphorus (P) differences is crucial for optimizing plant nutrient distribution and management. This study evaluated leaf N and P content in citrus trees across three cultivation modes: traditional mode (TM), wide-row and narrow-plant mode (WRNPM), and fenced mode (FM). We used hyperspectral data for non-destructive quantification and compared 1080 leaf samples from upper, middle, and lower canopy layers. Four models—Random Forest (RF), Backpropagation Neural Network (BPNN), Partial Least Squares (PLS), and Support Vector Machine (SVM)—were employed for leaf N and P estimation. Results showed that the TM had significantly lower N content compared to the WRNPM and FM, while the WRNPM exhibited higher P content. The canopy layer had minimal impact on N and P in the FM, and leaves in the upper layer had higher nutrient content in the WRNPM and TM. RF provided the best estimation accuracy, with R2 values of 0.66 for N and 0.72 for P. The cultivation mode and canopy layer significantly influenced the estimation accuracy, with the TM yielding the highest R2, followed by the WRNPM and FM obtaining the lowest accuracy. The labor-saving cultivation mode had different nutrient utilization efficiency compared to the TM. The cultivation mode and canopy layer should be considered when hyperspectral data were used for estimating the leaf N and P content. The study can offer new insights for precise fertilization strategies in fruit trees.

1. Introduction

Nitrogen (N) and phosphorus (P) are essential elements for plant growth and development. They serve as fundamental constituents of proteins, amino acids, phytohormones, nucleic acids, and other critical plant components [1,2,3,4,5,6]. To boost global food production, large quantities of N and P fertilizers have been applied to soils [7,8]. However, nearly half of these N inputs are not utilized effectively [9].
Citrus, as the world’s largest fruit category, holds significant economic importance, particularly in southern China, where it is the most important fruit tree [10]. Nitrogen plays an important role in the construction of citrus trees, material metabolism, fruit yield, and quality [11]. It is closely related to the differentiation, formation, and tree structure of citrus organs [12]. Phosphorus is mainly concentrated in flower organs, seeds, new shoots, new roots, and active parts of cell division. It plays an important role in plant carbohydrate synthesis, nitrogen metabolism, fruit enlargement, fruit quality, and storability. Excessive fertilizer application not only leads to a reduction in the yield and quality of citrus fruits but also causes environmental issues such as the eutrophication of rivers and ecosystems [8,13]. Therefore, it is urgent to develop strategies for effectively monitoring and promoting N and P use efficiency. Remote sensing, particularly through hyperspectral data, offers a more detailed and accurate method for the nutritional analysis of plants [14]. Compared to traditional chemical analysis, this approach is less time-consuming, labor-intensive, and environmentally harmful [11,15]. At present, most studies on the inversion of nitrogen and phosphorus content based on hyperspectral data focus on rice, soybean, and other grasses, primarily examining differences among various growth periods [16,17,18]. However, these studies often neglected the influence of varying canopy structures on the accuracy of nutrient estimation, particularly in woody perennials like citrus. Furthermore, although several models had been developed to predict N and P content using hyperspectral data, they often lacked adaptability to different cultivation modes and failed to account for the impact of canopy layers on nutrient estimation, which is crucial for practical application in diverse agricultural settings [19,20].
In recent years, innovative and labor-saving cultivation methods, such as the wide-row narrow-plant mode (WRNPM) and the fenced mode (FM), have gradually been adopted in citrus cultivation [21]. These modes had distinct canopy structures compared to the traditional mode (TM), particularly in vertical direction. They also had small plant spacing and large row spacing, which facilitate mechanized cultivation management and robotic harvesting [22]. Dian reported that the FM could enhance photosynthetic efficiency and nitrogen use efficiency, resulting in higher fruit quality including increased soluble solids, juice yield, solid acid ratio, and vitamin C content [22,23]. The practice of group cultivation often alters the shape of individual trees. As a result, agronomists are focusing more on improving fruit yield and quality, as well as optimizing canopy structure [24,25,26,27]. However, few studies have compared how different cultivation patterns and tree shapes affect nutrient utilization and distribution in fruit trees [28].
This study aimed to address several questions: How did different cultivation modes affect the absorption and distribution of N and P in citrus trees? What was the relationship between leaf N and P content and canopy layers of citrus trees? Could hyperspectral data effectively estimate the N and P content in citrus leaves across different canopy layers? Therefore, the main objectives of this study were (i) to evaluate the effect of cultivation modes and canopy layers on the N and P content in citrus trees and to explore the potential of labor-saving cultivation modes for nutrient utilization; (ii) to estimate the N and P content in citrus leaves using hyperspectral data and to assess different algorithms for predicting N and P content; (iii) to determine the impact of cultivation modes and canopy layers on the estimation of N and P content.

2. Materials and Methods

2.1. Study Area and Experimental Design

The study was conducted in November 2020 in orchards of Newhall navel oranges located in the Xinfeng county of Ganzhou city in Jiangxi Province, China (24°29′–27°09′N, 113°54′–116°38′E) (Figure 1). This area has a tropical monsoon climate, with an average annual temperature of 18.9 °C, average annual precipitation of 1587 mm, and an average annual sunshine duration of 1823 h [29]. The terrain is predominantly mountainous and hilly, with slopes ranging from 16° to 45° and an average altitude of 300 to 500 m [30].
A total of 120 citrus trees, representing three cultivation modes—WRNPM, FM, and TM—were selected using a stratified random sampling method, with 40 citrus trees in each cultivation mode. Tree height, base diameter, and crown width of all the trees were measured, and the results were shown in Figure 2. Each tree was accurately positioned using differential GPS, with a positional error within 8 mm. Nitrogen and phosphorus contents were measured in a total of nine leaves from the upper, middle, and lower layers of each citrus tree. The selected leaves in each layer were the third leaves from the tip of the spring shoots on annual branches. A total of 1080 leaves were analyzed in this study, and the spectral reflectance of each leaf was measured before collection.

2.2. Hyperspectral Data Acquisition

A portable ASD FieldSpecHH spectrometer (325–1075 nm range, 1 nm resolution; ASD Inc., Boulder, CO, USA) was used to measure the spectral reflectance of citrus leaves. Calibration was performed using a white reference panel under clear skies to ensure accuracy. For each leaf, 10 measurements were taken at consistent angles and distances, specifically avoiding the leaf veins to ensure the spectral data reflected the true leaf tissue characteristics. These measurements were averaged to produce a representative spectral curve. The Savitzky–Golay method was applied for smoothing, resulting in 10,800 spectral curves used for further analysis.

2.3. Determination of Nitrogen and Phosphorus Content of Citrus Leaves

Nitrogen (N) content was determined using the Kjeldahl method [31,32]. The specific steps were as follows: A 0.100 g sample, ground and passed through a 0.25 mm sieve, was weighed and placed into a digestion tube. The sample was mixed with 3 g of K2SO4, 0.2 g of CuSO4, and 10 mL of concentrated sulfuric acid and then digested at 230 °C, 350 °C, and 420 °C for 15, 15, and 60 min, respectively. After digestion and cooling, the sample was analyzed using a Kjeldahl apparatus.
Phosphorus (P) was determined using a concentrated sulfuric acid–perchloric acid digestion method [33], followed by analysis with a flow analyzer. The procedure was as follows: A 0.100 g sample was weighed, sieved, and placed into a digestion tube. Two to three drops of ultrapure water were added to moisten the sample, followed by 5 mL of concentrated sulfuric acid and 0.5 mL of perchloric acid. The mixture was gently shaken and left to stand overnight. Digestion was then carried out at 160 °C, 240 °C, and 370 °C for 40, 30, and 60 min, respectively. After digestion and cooling, the sample was stored in a polyethylene bottle, and the phosphorus content was measured using a fully automated chemical analyzer.

2.4. Spectral Information

2.4.1. Vegetation Indices

Eleven vegetation indices (VIs) related to nitrogen and phosphorus content were selected based on previous studies. These indices were extracted from hyperspectral curves and summarized in Table 1. The selected VIs included ARI (Anthocyanin Reflectance Index), TCARI (Transformed Chlorophyll Absorption Ratio), TSAVI (Transformed Soil Adjusted Vegetation Index), MNDVI (Modified Normalized Difference Vegetation Index), PRI(570,515) (Photochemical Reflectance Index), ARVI (Atmospherically Resistant Vegetation Index), DATT4, DND1, DND7, DMAX42, and MSRCHL.

2.4.2. First-Order Differential Spectral Acquisition

First-order derivative spectra were obtained based on the original spectra. Twenty first-derivative spectra from 12 visible region bands and 8 near-infrared (NIR) region bands were applied in the inversion of nitrogen content in citrus leaves. Similarly, 18 first-derivative spectra, including 10 visible region bands and 8 NIR bands, were used for the estimation of phosphorus content (Table 2).

2.5. Machine Learning Algorithms and Random Forest Importance Ranking

In this study, we utilized four machine learning algorithms—Random Forest (RF), Backpropagation Neural Network (BPNN), Support Vector Machine (SVM), and Partial Least Squares (PLS)—to estimate nitrogen (N) and phosphorus (P) content in Newhall navel orange leaves. Random Forest importance ranking evaluates the significance of each feature by assessing how much each feature contributes to improving the model’s accuracy. This involves measuring the decrease in model performance when a feature is either removed or its values were permuted.
Initially, we ranked the importance of spectral feature parameters derived from hyperspectral data to identify the most predictive features for N and P content. The most significant features were then selected as independent variables for our regression models, with leaf nutrient contents as the dependent variables. Predictive models were built using these features with each algorithm. For RF, BPNN, and SVM, hyperparameters such as the number of trees, network architecture, and kernel types were optimized based on established practices and experimental validation. PLS was employed to reduce dimensionality and model the relationship between the selected features and nutrient contents.

2.6. Data Analysis

We used Analysis of Variance (ANOVA) to evaluate the effects of cultivation modes and canopy layers on the nitrogen and phosphorus content of citrus leaves, respectively. Multiple comparisons were performed using the LSD method. ANOVA and LSD analyses were conducted using R 4.0.5 software. First, the dataset was randomly divided into a calibration dataset (2/3) and a validation dataset (1/3). Next, we compared the inversion accuracy of RF, BPNN, SVM, and PLS algorithms. We also tested the influence of cultivation modes and canopy layers on the estimation of nitrogen and phosphorus content using RF algorithms. Model building and validation were conducted using the Scikit-learn library in Python 3.7.3. All graphs were generated using R 4.0.5 and OriginPro software 2019. The predictive performance of each estimation model was evaluated using the coefficient of determination (R2), mean square error (MSE), and mean absolute error (MAE), calculated as follows:
R 2 = 1 − ∑ i = 1 n ( y i − y ^ i ) 2 ∑ i = 1 n ( y i − y ¯ ) 2
M S E = 1 n ∑ i = 1 n ( y i − y ^ i ) 2
M A E = 1 n ∑ i = 1 n y i − y ^ i
where y ^ i represents the predicted values; y i the measured values in the field; y ¯ i the average measured values; and n the sample size used in validation.

3. Results

3.1. Nitrogen and Phosphorus Content in Citrus Leaves

Figure 3 showed that there was no significant difference in leaf N content between the WRNPM and FM, regardless of whether the citrus was in the upper, middle, or lower layers. However, the leaf N content in the TM was significantly lower compared to the WRNPM and FM. Figure 4 illustrated that leaf phosphorus (P) content follows a similar trend across different layers of the canopy, with the WRNPM exhibiting notably higher P content than the FM and TM. No significant differences in P content were observed between the FM and TM.
Figure 5 demonstrated that, in the WRNPM, leaf N content in the upper layer was significantly greater than in the middle and lower layers. In contrast, the canopy layer did not significantly impact N content in the FM and TM. Figure 6 illustrated the effect of canopy layers on leaf P content across different cultivation modes. In the WRNPM, the P content in the upper leaves was higher than in the middle and lower layers, with no significant difference between the middle and lower layers. In the FM, the canopy layers did not significantly affect leaf P content, while in the TM, the P content in the upper layer was greater than in the lower layer.

3.2. RF, BPNN, SVM, and PLS Algorithms

To estimate the nitrogen (N) and phosphorus (P) content in citrus leaves, we applied four machine learning algorithms—Random Forest (RF), Backpropagation Neural Network (BPNN), Support Vector Machine (SVM), and Partial Least Squares (PLS). For nitrogen content estimation, the top-ranked spectral parameters, including 10 vegetation indices and 20 first-order differential spectra, were selected based on importance ranking (Figure 7). The resulting models showed that the RF algorithm achieved the highest accuracy, with R2, MSE, and MAE values of 0.66, 0.368, and 0.505, respectively. The SVM model yielded R2, MSE, and MAE values of 0.64, 0.382, and 0.491, respectively, while the BPNN model produced R2, MSE, and MAE values 0.59, 0.444, and 0.529, respectively. Conversely, the PLS model exhibited the lowest accuracy, with R2, MSE, and MAE values of 0.48, 0.559, and 0.608, respectively. The RF model’s superior performance was evident in its higher R2 and lower MSE and MAE, indicating high reliability for estimating nitrogen content in citrus leaves. The scatter plots comparing measured versus predicted nitrogen content for each model were shown Figure 8.
For phosphorus content estimation, we selected the top-ranked spectral parameters, including four vegetation indices and 18 first-order differential spectra (Figure 9). Again, the RF algorithm produced the most accurate model, with R2, MSE, and MAE values of 0.72, 0. 321, and 0. 457, respectively. The SVM model yielded R2, MSE, and MAE values of 0.67, 0.375, and 0.477, respectively, while the BPNN model produced R2, MSE, and MAE values 0.68, 0.362, and 0.470, respectively. Conversely, the PLS model exhibited the lowest accuracy, with R2, MSE, and MAE values of 0.58, 0.476, and 0.564, respectively. Thus, the RF-based model for phosphorus estimation also demonstrated high reliability, with overall better fitting compared to the nitrogen content estimation model. Scatter plots comparing measured versus predicted phosphorus content for each model were presented in Figure 10.

3.3. Estimation of Nitrogen and Phosphorus Content by Cultivation Mode Using RF Algorithm

Given the superior performance of the RF algorithm in estimating nitrogen and phosphorus content, we used it to assess the effects of cultivation mode and canopy layer on the accuracy of these estimations.
For the nitrogen content, the RF model showed variations in accuracy across different cultivation modes. The TM exhibited the highest accuracy, with R2, MSE, and MAE of 0.64, 0.436, and 0.528, respectively. This was followed by the WRNPM with R2, MSE, and MAE of 0.58, 0.484, and 0.567, respectively. The fenced mode (FM) had the lowest accuracy, with R2, MSE, and MAE of 0.46, 0.614, and 0.669. Measured and predicted nitrogen content for different cultivation patterns was shown in Figure 11A–C.
Similarly, for phosphorus content estimation, the accuracy of the RF model varied by cultivation mode. The TM provided the best prediction, with the R2, MSE, and MAE of the model being 0.71, 0.291, and 0.444, respectively. The WRNPM followed with R2, MSE, and MAE of 0.59, 0.439, and 0.556. The FM had the lowest accuracy, with R2, MSE, and MAE of 0.52, 0.524, and 0.567. Measured and predicted phosphorus contents for different cultivation patterns were depicted in Figure 11D–F.

3.4. Estimation of Nitrogen and Phosphorus Content by Canopy Layer Using the RF Algorithm

In addition to the cultivation modes, canopy layers significantly impacted the estimation of nitrogen (N) content in citrus leaves. Using the RF algorithm, N content was estimated separately for the upper, middle, and lower layers of citrus trees. The results indicated that the estimation of N and phosphorus (P) content was more accurate for the upper and lower layers. The R2, MSE, and MAE of the upper leaf N content estimation model were 0.70, 0.331, and 0.479, respectively, and 0.66, 0.334, and 0.481 for the lower layer. The estimation of N content for the middle layer was less effective, with R2, MSE, and MAE values of 0.49, 0.614, and 0.650, respectively. The measured versus predicted values for the nitrogen content in citrus leaves across different layers were shown in Figure 12A–C.
Canopy layers also notably influenced the estimation of P content in citrus leaves, following similar general trends as observed for nitrogen but with some variations. The estimation accuracy was highest for the lower leaves, with R2, MSE, and MAE of 0.73, 0.277, and 0.406 for the model, respectively. The upper leaf estimation was slightly less accurate, with R2, MSE, and MAE of 0.68, 0.299, and 0.456, respectively. The middle layer estimation was comparatively less accurate, with R2, MSE, and MAE of 0.59, 0.419, and 0.529, respectively. The measured versus predicted values for phosphorus content in citrus leaves across different layers were shown in Figure 12D–F.

4. Discussion

Nitrogen (N) and phosphorus (P) are crucial macronutrients for plant growth and development. Deficiencies in these nutrients can adversely affect plant health, while excessive fertilization can lead to environmental issues such as algal blooms [42]. Variations in cultivation modes, training, and pruning techniques can influence canopy structure, impacting microclimate, transpiration, and nutrient acquisition and allocation [43]. These factors cause variations in leaf nutrient content and photosynthetic capacity [16,44,45]. Understanding how different cultivation modes affect N and P content in citrus leaves can provide insights into nutrient utilization and improve fruit tree management practices. Hyperspectral remote sensing offers a non-destructive and efficient method for estimating leaf N and P content. This study compared N and P content in different canopy layers of citrus trees across various cultivation modes and developed models for estimating leaf nutrient content using hyperspectral reflectance and machine learning algorithms. The effects of cultivation modes and canopy layers on nutrient estimation were also evaluated.

4.1. Effect of Canopy Layer and Cultivation Mode on Leaf N and P Contents of Citrus Trees

Our findings indicate that the canopy layer significantly affects N and P content in citrus leaves, suggesting that variations in cultivation modes influence nutrient absorption and utilization, ultimately impacting photosynthesis [44]. This effect was particularly pronounced in trees cultivated using the TM and WRNPM, whereas it was less evident in the FM (Figure 6). Nitrogen content was notably higher in the WRNPM and FM compared to the TM across all canopy layers (Figure 3). This may be attributed to the higher yields associated with these new cultivation modes [17], which require more nutrients to support tree growth. The FM, in particular, effectively reduces light interception and shading between leaves, leading to a more uniform distribution of N and P, thus enhancing photosynthesis. Our previous study demonstrated that FM increases the photosynthetic rate across all leaf layers compared to the WRNPM or TM due to its wider row spacing and narrower crown width [23]. UAV-based LiDAR data highlighted differences in vertical canopy structure among the cultivation modes, with the FM showing higher leaf N content and, consequently, stronger photosynthetic capacity [23]. These results suggest potential strategies for optimizing N and P distribution and improving light-use efficiency in citrus canopies through the WRNPM and FM.

4.2. Effect of Canopy Layer and Cultivation Modes on Estimation of N and P Content

The spectral characteristics of plant canopies were influenced by canopy structure and the surrounding environment. Spectral preprocessing helps eliminate noise and environmental interference, enhancing spectral features. This study explored the effectiveness of different spectral parameters and first-order differential spectra for estimating N and P content in citrus canopies across various cultivation modes. Models including RF, BPNN, SVM, and PLS were developed to estimate canopy N and P content. RF proved to be the most effective model, accurately predicting both canopy N and P levels. This finding aligned with Osco et al., who noted that nonlinear methods like RF were more suitable for predicting canopy nitrogen content than linear methods [46]. Our RF model demonstrated the highest accuracy in estimating nitrogen and phosphorus across different cultivation modes, with the TM showing the highest accuracy and the WRNPM the lowest (Figure 11). This may be due to the higher variability in leaf N and P content in the TM, which benefits nutrient modeling. The RF model also effectively estimated nutrient content across different canopy layers, with R2 values of 0.70 for N in the upper canopy (Figure 12A) and 0.73 for P in the lower canopy (Figure 12F). These results emphasize the need to consider cultivation mode and canopy layer when estimating nutrient content in fruit trees.

4.3. Limitations and Future Prospects

While this study provided valuable insights into the effects of cultivation modes on N and P content in citrus leaves, it had limitations. The research was conducted in specific regions, which may not fully represent other environmental conditions or cultivation practices. Future studies should explore diverse geographical locations and incorporate additional remote sensing technologies, such as full-spectrum hyperspectral imaging, to enhance the accuracy and applicability of nutrient estimations [47,48]. It will be important to account for differences between cultivation modes and canopy heights in future research aimed at accurately modeling nutrient content in citrus trees. Further optimization of machine learning algorithms and more extensive data validation were also necessary to improve nutrient estimation accuracy. Additionally, investigating the effects of canopy management on nutrient efficiency could offer further opportunities for optimizing cultivation practices and enhancing fruit quality [49,50].

5. Conclusions

Labor-saving cultivation modes, such as FM and WRNPM, are increasingly being adopted in citrus cultivation. It is essential to evaluate their effects on nutrient utilization to select appropriate cultivation practices and improve fruit yield and quality. This study compared nutrient utilization among three cultivation modes: two labor-saving modes and one traditional mode (TM). Nitrogen content was significantly lower in the TM compared to the labor-saving modes, while phosphorus content was higher in the WRNPM than in the FM and TM. The canopy layer had minimal impact on leaf N and P content in the FM, but the upper layers of the WRNPM and TM exhibited higher nutrient content. When estimating leaf N and P content using hyperspectral reflectance and machine learning algorithms, RF performed the best, with R2 values of 0.67 for N and 0.73 for P. The accuracy of nutrient estimation was significantly influenced by canopy layer, with the middle layer proving less reliable for nutrient estimation.

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/rs16173261/s1.

Author Contributions

D.L. organized the writing, drew pictures, and polished the articles. Q.H. determined the N and P content, obtained leaf hyperspectral data, analyzed data, and drew pictures. J.Z. (Jinzhi Zhang), C.H. found appropriate citrus orchards, helped determine the experimental design, and provided necessary experimental instruments. Y.D. obtained hyperspectral data and revised the manuscript. J.Z. (Jingjing Zhou) designed the experiment, analyzed the photosynthetic factors and nitrogen content-related experimental data, and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Key Research and Development Plan (no. 2019YFD1000104) and the National Natural Fund Project (no. 31901963, 32060653 and 32060646).

Data Availability Statement

The data used in this study are available from the corresponding author upon reasonable request.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Castro-Rodríguez, V.; Cañas, R.A.; De La Torre, F.N.; Pascual, M.B.; Avila, C.; Cánovas, F.M. Molecular Fundamentals of Nitrogen Uptake and Transport in Trees. J. Exp. Bot. 2017, 68, 2489–2500. [Google Scholar] [CrossRef] [PubMed]
  2. Luo, J.; Zhou, J.-J. Growth Performance, Photosynthesis, and Root Characteristics Are Associated with Nitrogen Use Efficiency in Six Poplar Species. Environ. Exp. Bot. 2019, 164, 40–51. [Google Scholar] [CrossRef]
  3. Luo, J.; Zhou, J.-J.; Masclaux-Daubresse, C.; Wang, N.; Wang, H.; Zheng, B. Morphological and Physiological Responses to Contrasting Nitrogen Regimes in Populus Cathayana Is Linked to Resources Allocation and Carbon/Nitrogen Partition. Environ. Exp. Bot. 2019, 162, 247–255. [Google Scholar] [CrossRef]
  4. Liu, D. Root Developmental Responses to Phosphorus Nutrition. J. Integr. Plant Biol. 2021, 63, 1065–1090. [Google Scholar] [CrossRef] [PubMed]
  5. Li, Z.; Deng, S.; Zhu, D.; Wu, J.; Zhou, J.; Shi, W.; Fayyaz, P.; Luo, Z.-B.; Luo, J. Proteomic Reconfigurations Underlying Physiological Alterations in Poplar Roots in Acclimation to Changing Nitrogen Availability. Environ. Exp. Bot. 2023, 211, 105367. [Google Scholar] [CrossRef]
  6. Lu, Y.; Zheng, B.; Zhang, C.; Yu, C.; Luo, J. Wood Formation in Trees Responding to Nitrogen Availability. Ind. Crops Prod. 2024, 218, 118978. [Google Scholar] [CrossRef]
  7. Robertson, G.P.; Vitousek, P.M. Nitrogen in Agriculture: Balancing the Cost of an Essential Resource. Annu. Rev. Environ. Resour. 2009, 34, 97–125. [Google Scholar] [CrossRef]
  8. Powers, S.M.; Bruulsema, T.W.; Burt, T.P.; Chan, N.I.; Elser, J.J.; Haygarth, P.M.; Howden, N.J.K.; Jarvie, H.P.; Lyu, Y.; Peterson, H.M.; et al. Long-Term Accumulation and Transport of Anthropogenic Phosphorus in Three River Basins. Nat. Geosci. 2016, 9, 353–356. [Google Scholar] [CrossRef]
  9. Schroeder, J.I.; Delhaize, E.; Frommer, W.B.; Guerinot, M.L.; Harrison, M.J.; Herrera-Estrella, L.; Horie, T.; Kochian, L.V.; Munns, R.; Nishizawa, N.K.; et al. Using Membrane Transporters to Improve Crops for Sustainable Food Production. Nature 2013, 497, 60–66. [Google Scholar] [CrossRef]
  10. Guo, W.; Ye, J.; Deng, X. Fruit scientific research in New China in the past 70 years: Citrus. J. Fruit Sci. 2019, 36, 1264–1272. [Google Scholar] [CrossRef]
  11. Osco, L.P.; Ramos, A.P.M.; Pereira, D.R.; Moriya, É.A.S.; Imai, N.N.; Matsubara, E.T.; Estrabis, N.; de Souza, M.; Junior, J.M.; Gonçalves, W.N.; et al. Predicting Canopy Nitrogen Content in Citrus-Trees Using Random Forest Algorithm Associated to Spectral Vegetation Indices from UAV-Imagery. Remote Sens. 2019, 11, 2925. [Google Scholar] [CrossRef]
  12. Li, W.; Zhang, M.; Shu, H. Physiological Effects of Nitrogen on Fruit Trees. J. Shandong Agric. Univ. 2002, 33, 96–100. [Google Scholar]
  13. Cameron, K.C.; Di, H.J.; Moir, J.L. Nitrogen Losses from the Soil/Plant System: A Review. Ann. Appl. Biol. 2013, 162, 145–173. [Google Scholar] [CrossRef]
  14. Osco, L.P.; Ramos, A.P.M.; Faita Pinheiro, M.M.; Moriya, É.A.S.; Imai, N.N.; Estrabis, N.; Ianczyk, F.; Araújo, F.F.; Liesenberg, V.; Jorge, L.A.d.C.; et al. A Machine Learning Framework to Predict Nutrient Content in Valencia-Orange Leaf Hyperspectral Measurements. Remote Sens. 2020, 12, 906. [Google Scholar] [CrossRef]
  15. Román, J.R.; Rodríguez-Caballero, E.; Rodríguez-Lozano, B.; Roncero-Ramos, B.; Chamizo, S.; Águila-Carricondo, P.; Cantón, Y. Spectral Response Analysis: An Indirect and Non-Destructive Methodology for the Chlorophyll Quantification of Biocrusts. Remote Sens. 2019, 11, 1350. [Google Scholar] [CrossRef]
  16. Mahajan, G.R.; Sahoo, R.N.; Pandey, R.N.; Gupta, V.K.; Kumar, D. Using Hyperspectral Remote Sensing Techniques to Monitor Nitrogen, Phosphorus, Sulphur and Potassium in Wheat (Triticum aestivum L.). Precis. Agric. 2014, 15, 499–522. [Google Scholar] [CrossRef]
  17. Zhu, Y.; Abdalla, A.; Tang, Z.; Cen, H. Improving Rice Nitrogen Stress Diagnosis by Denoising Strips in Hyperspectral Images via Deep Learning. Biosyst. Eng. 2022, 219, 165–176. [Google Scholar] [CrossRef]
  18. Zhang, Y.; Guan, M.; Wang, L.; Cui, X.; Li, T.; Zhang, F. In Situ Nondestructive Detection of Nitrogen Content in Soybean Leaves Based on Hyperspectral Imaging Technology. Agronomy 2024, 14, 806. [Google Scholar] [CrossRef]
  19. Liu, Y.; Lyu, Q.; He, S.; Yi, S.; Liu, X.; Xie, R.; Zheng, Y.; Deng, L. Prediction of Nitrogen and Phosphorus Contents in Citrus Leaves Based on Hyperspectral Imaging. Int. J. Agric. Biol. Eng. 2015, 8, 80–88. [Google Scholar]
  20. Siedliska, A.; Baranowski, P.; Pastuszka-Woźniak, J.; Zubik, M.; Krzyszczak, J. Identification of Plant Leaf Phosphorus Content at Different Growth Stages Based on Hyperspectral Reflectance. BMC Plant Biol. 2021, 21, 28. [Google Scholar] [CrossRef]
  21. Li, D.; Ruan, S.; Hu, Q.; Zhang, J.; Zhang, Y.; Dian, Y.; Hu, C.; Liu, Y.; Lei, H.; Zhou, J. Nitrogen estimation and spatial analysis of orchard canopy based on UAV remote sensing. J. Huazhong Agric. Univ. 2023, 42, 158–166. [Google Scholar] [CrossRef]
  22. Hu, Q.; Dian, Y.; Gong, Z.; Zhang, J.; Hu, C.; Liu, Y.; Lei, H.; Yuan, K.; Zhou, J. Analyzing fruit quality of Newhall navel oranges with different cultivation patterns. J. Huazhong Agric. Univ. 2022, 41, 108–115. [Google Scholar] [CrossRef]
  23. Dian, Y.; Liu, X.; Hu, L.; Zhang, J.; Hu, C.; Liu, Y.; Zhang, J.; Zhang, W.; Hu, Q.; Zhang, Y.; et al. Characteristics of Photosynthesis and Vertical Canopy Architecture of Citrus Trees under Two Labor-Saving Cultivation Modes Using Unmanned Aerial Vehicle (UAV)-Based LiDAR Data in Citrus Orchards. Hortic. Res. 2023, 10, uhad018. [Google Scholar] [CrossRef] [PubMed]
  24. Zhao, C.; Wang, Q.; Han, M.; Wang, A.; Liu, H. Effects of tree shape on the quality of leaf and fruit and the yield in peach. J. Northwest Agric. For. Univ. 2010, 38, 160–164+170. [Google Scholar] [CrossRef]
  25. Sharma, Y.; Singh, H.; Thakur, A. Effect of Training System and in Row Spacing on Yield and Fruit Quality of Peach in the Sub-Tropical Regions. Ind. J. Hort. 2017, 74, 440. [Google Scholar] [CrossRef]
  26. Seki, T.; Hirose, K.; Shibata, K. Yield and Fruit Quality of Japanese Pear in “Joint V-Shaped Trellis”. Acta Hortic. 2021, 1303, 171–176. [Google Scholar] [CrossRef]
  27. Liu, L.; Li, Q.L.; Gao, D.T.; Wei, Z.F.; Shi, C.Y.; Wang, Z.Q.; Liu, J.W. Effects of tree shapes on growth, yield and quality of peach. J. Fruit Sci. 2022, 39, 36–46. [Google Scholar] [CrossRef]
  28. Liu, X.; Hu, D.; Ma, X.; Xiang, P.; Yuan, X. Spatial distribution of spring shoot, leaf nutrition and fruit in citrus canopy with different tree shapes. J. Gansu Agric. Univ. 2023, 58, 126–135. [Google Scholar] [CrossRef]
  29. Wang, X.; Zhang, Y.; Huang, G.; Ma, W.; Chen, X.; Xia, C. Forestland Site Quality and Productivity Potential in Ganzhou City. J. Jiangxi Agric. Univ. 2014, 36, 1159–1166. [Google Scholar] [CrossRef]
  30. Zhou, X.; Zhu, W.; Ma, G.; Zhang, D.; Zheng, Z. Assessing the Loss Value of Soil and Water Conservation Resulted from the Mining of Rare Earth Ore in Ganzhou, Jiangxi Province. J. Nat. Resour. 2016, 31, 982–993. [Google Scholar]
  31. Li, H.; Li, M.; Luo, J.; Cao, X.; Qu, L.; Gai, Y.; Jiang, X.; Liu, T.; Bai, H.; Janz, D.; et al. N-Fertilization Has Different Effects on the Growth, Carbon and Nitrogen Physiology, and Wood Properties of Slow- and Fast-Growing Populus Species. J. Exp. Bot. 2012, 63, 6173–6185. [Google Scholar] [CrossRef] [PubMed]
  32. Li, Z.; Guan, L.; Zhang, C.; Zhang, S.; Liu, Y.; Lu, Y.; Luo, J. Nitrogen Assimilation Genes in Poplar: Potential Targets for Improving Tree Nitrogen Use Efficiency. Ind. Crops Prod. 2024, 216, 118705. [Google Scholar] [CrossRef]
  33. Tang, J.-W.; Cao, M.; Zhang, J.-H.; Li, M.-H. Litterfall Production, Decomposition and Nutrient Use Efficiency Varies with Tropical Forest Types in Xishuangbanna, SW China: A 10-Year Study. Plant Soil 2010, 335, 271–288. [Google Scholar] [CrossRef]
  34. Gitelson, A.A.; Merzlyak, M.N.; Chivkunova, O.B. Optical Properties and Nondestructive Estimation of Anthocyanin Content in Plant Leaves. Photochem. Photobiol. 2007, 74, 38–45. [Google Scholar] [CrossRef]
  35. Haboudane, D.; Miller, J.R.; Pattey, E.; Zarco-Tejada, P.J.; Strachan, I.B. Hyperspectral Vegetation Indices and Novel Algorithms for Predicting Green LAI of Crop Canopies: Modeling and Validation in the Context of Precision Agriculture. Remote Sens. Environ. 2004, 90, 337–352. [Google Scholar] [CrossRef]
  36. Sims, D.A.; Gamon, J.A. Relationships between Leaf Pigment Content and Spectral Reflectance across a Wide Range of Species, Leaf Structures and Developmental Stages. Remote Sens. Environ. 2002, 81, 337–354. [Google Scholar] [CrossRef]
  37. Hernández-Clemente, R.; Navarro-Cerrillo, R.M.; Suárez, L.; Morales, F.; Zarco-Tejada, P.J. Assessing Structural Effects on PRI for Stress Detection in Conifer Forests. Remote Sens. Environ. 2011, 115, 2360–2375. [Google Scholar] [CrossRef]
  38. Kaufman, Y.J.; Tanre, D. Atmospherically Resistant Vegetation Index (ARVI) for EOS-MODIS. IEEE Trans. Geosci. Remote Sens. 1992, 30, 261–270. [Google Scholar] [CrossRef]
  39. Datt, B. Remote Sensing of Chlorophyll a, Chlorophyll b, Chlorophyll A + b, and Total Carotenoid Content in Eucalyptus Leaves. Remote Sens. Environ. 1998, 66, 111–121. [Google Scholar] [CrossRef]
  40. Shah, S.H.; Angel, Y.; Houborg, R.; Ali, S.; McCabe, M.F. A Random Forest Machine Learning Approach for the Retrieval of Leaf Chlorophyll Content in Wheat. Remote Sens. 2019, 11, 920. [Google Scholar] [CrossRef]
  41. Kooistra, L.; Leuven, R.S.E.W.; Wehrens, R.; Nienhuis, P.H.; Buydens, L.M.C. A Comparison of Methods to Relate Grass Reflectance to Soil Metal Contamination. Int. J. Remote Sens. 2003, 24, 4995–5010. [Google Scholar] [CrossRef]
  42. Li, J.; Guo, Q.; Zhang, J.; Korpelainen, H.; Li, C. Effects of Nitrogen and Phosphorus Supply on Growth and Physiological Traits of Two Larix Species. Environ. Exp. Bot. 2016, 130, 206–215. [Google Scholar] [CrossRef]
  43. Gao, Y.; Gao, S.; Jia, L.; Dai, T.; Wei, X.; Duan, J.; Liu, S.; Weng, X. Canopy Characteristics and Light Distribution in Sapindus Mukorossi Gaertn. Are Influenced by Crown Architecture Manipulation in the Hilly Terrain of Southeast China. Sci. Hortic. 2018, 240, 11–22. [Google Scholar] [CrossRef]
  44. Génard, M.; Baret, F.; Simon, D. A 3D Peach Canopy Model Used to Evaluate the Effect of Tree Architecture and Density on Photosynthesis at a Range of Scales. Ecol. Model. 2000, 128, 197–209. [Google Scholar] [CrossRef]
  45. Peng, L.L.; Wei, L.; Wang, H.; Niu, Z.; Xie, P. Effect of Canopy Structure on Foliar Photosynthetic Characteristics and Fruit Quality of Pears. Acta Bot. Boreal.-Occident. Sin. 2020, 40, 1180–1191. [Google Scholar] [CrossRef]
  46. Reda, R.; Saffaj, T.; Ilham, B.; Saidi, O.; Issam, K.; Brahim, L.; El Hadrami, E.M. A Comparative Study between a New Method and Other Machine Learning Algorithms for Soil Organic Carbon and Total Nitrogen Prediction Using near Infrared Spectroscopy. Chemom. Intell. Lab. Syst. 2019, 195, 103873. [Google Scholar] [CrossRef]
  47. Tian, D.; Kattge, J.; Chen, Y.; Han, W.; Luo, Y.; He, J.; Hu, H.; Tang, Z.; Ma, S.; Yan, Z.; et al. A Global Database of Paired Leaf Nitrogen and Phosphorus Concentrations of Terrestrial Plants. Ecology 2019, 100, e02812. [Google Scholar] [CrossRef]
  48. Peng, X.; Chen, D.; Zhou, Z.; Zhang, Z.; Xu, C.; Zha, Q.; Wang, F.; Hu, X. Prediction of the Nitrogen, Phosphorus and Potassium Contents in Grape Leaves at Different Growth Stages Based on UAV Multispectral Remote Sensing. Remote Sens. 2022, 14, 2659. [Google Scholar] [CrossRef]
  49. Kviklys, D.; Viškelis, J.; Liaudanskas, M.; Janulis, V.; Laužikė, K.; Samuolienė, G.; Uselis, N.; Lanauskas, J. Apple Fruit Growth and Quality Depend on the Position in Tree Canopy. Plants 2022, 11, 196. [Google Scholar] [CrossRef]
  50. Kaučić, M.; Vuković, M.; Gašpar, L.; Fruk, G.; Vidrih, R.; Nečemer, M.; Fruk, M.; Jatoi, M.A.; Fu, D.; Kobav, M.B.; et al. The Effect of Canopy Position on the Fruit Quality Parameters and Contents of Bioactive Compounds and Minerals in ‘Braeburn’ Apples. Agronomy 2023, 13, 2523. [Google Scholar] [CrossRef]
Remotesensing 16 03261 g001
Figure 1. Overview map of the study area.
Figure 1. Overview map of the study area.
Remotesensing 16 03261 g001
Remotesensing 16 03261 g002
Figure 2. Summarized data on inter-row plant spacing, within-row tree spacing, crown width, tree height, and diameter under three cultivation patterns. Crown width, tree height, and diameter were presented as mean ± standard deviation. The three cultivation modes were represented as follows: (A) (WRNPM), (B) (FM), and (C) (TM) [23].
Figure 2. Summarized data on inter-row plant spacing, within-row tree spacing, crown width, tree height, and diameter under three cultivation patterns. Crown width, tree height, and diameter were presented as mean ± standard deviation. The three cultivation modes were represented as follows: (A) (WRNPM), (B) (FM), and (C) (TM) [23].
Remotesensing 16 03261 g002
Remotesensing 16 03261 g003
Figure 3. One-way ANOVA results for leaf nitrogen content across three cultivation modes. (A–C) denote upper, middle, and lower layers, respectively. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Figure 3. One-way ANOVA results for leaf nitrogen content across three cultivation modes. (A–C) denote upper, middle, and lower layers, respectively. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Remotesensing 16 03261 g003
Remotesensing 16 03261 g004
Figure 4. One-way ANOVA results for leaf phosphorus content across three cultivation modes. (A–C) denote upper, middle, and lower layers, respectively. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Figure 4. One-way ANOVA results for leaf phosphorus content across three cultivation modes. (A–C) denote upper, middle, and lower layers, respectively. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Remotesensing 16 03261 g004
Remotesensing 16 03261 g005
Figure 5. One-way ANOVA results for leaf nitrogen content across different canopy layers and cultivation modes (WRNPM, FM, and TM). Panel (A) shows the results for the upper canopy layer, (B) for the middle canopy layer, and (C) for the lower canopy layer. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Figure 5. One-way ANOVA results for leaf nitrogen content across different canopy layers and cultivation modes (WRNPM, FM, and TM). Panel (A) shows the results for the upper canopy layer, (B) for the middle canopy layer, and (C) for the lower canopy layer. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Remotesensing 16 03261 g005
Remotesensing 16 03261 g006
Figure 6. One-way ANOVA results for leaf phosphorus content across different canopy layers and cultivation modes (WRNPM, FM, and TM). Panel (A) shows the results for the upper canopy layer, (B) for the middle canopy layer, and (C) for the lower canopy layer. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Figure 6. One-way ANOVA results for leaf phosphorus content across different canopy layers and cultivation modes (WRNPM, FM, and TM). Panel (A) shows the results for the upper canopy layer, (B) for the middle canopy layer, and (C) for the lower canopy layer. Identical lowercase letters indicate no significant differences, while different letters denote significant differences between groups (p < 0.05).
Remotesensing 16 03261 g006
Remotesensing 16 03261 g007
Figure 7. Importance ranking of hyperspectral parameters selected for leaf nitrogen content.
Figure 7. Importance ranking of hyperspectral parameters selected for leaf nitrogen content.
Remotesensing 16 03261 g007
Remotesensing 16 03261 g008
Figure 8. Measured vs. predicted leaf nitrogen content along a 1:1 line for RF, SVM, BPNN, and PLS models. Panel (A) shows results for the RF model, (B) for the SVM model, (C) for the BPNN model, and (D) for the PLS model. Black lines represent the fitted lines for each model. (Test set results are shown, and train set results are in Supplementary Table S1).
Figure 8. Measured vs. predicted leaf nitrogen content along a 1:1 line for RF, SVM, BPNN, and PLS models. Panel (A) shows results for the RF model, (B) for the SVM model, (C) for the BPNN model, and (D) for the PLS model. Black lines represent the fitted lines for each model. (Test set results are shown, and train set results are in Supplementary Table S1).
Remotesensing 16 03261 g008
Remotesensing 16 03261 g009
Figure 9. Importance ranking of hyperspectral parameters selected for leaf phosphorus content.
Figure 9. Importance ranking of hyperspectral parameters selected for leaf phosphorus content.
Remotesensing 16 03261 g009
Remotesensing 16 03261 g010
Figure 10. Measured vs. predicted leaf phosphorus content along a 1:1 line for RF, SVM, BPNN, and PLS models. Panel (A) shows results for the RF model, (B) for the SVM model, (C) for the BPNN model, and (D) for the PLS model. Black lines represent the fitted lines for each model. (Test set results are shown, and train set results are in Supplementary Table S2).
Figure 10. Measured vs. predicted leaf phosphorus content along a 1:1 line for RF, SVM, BPNN, and PLS models. Panel (A) shows results for the RF model, (B) for the SVM model, (C) for the BPNN model, and (D) for the PLS model. Black lines represent the fitted lines for each model. (Test set results are shown, and train set results are in Supplementary Table S2).
Remotesensing 16 03261 g010
Remotesensing 16 03261 g011
Figure 11. Measured vs. RF model-predicted leaf nitrogen (A–C) and phosphorus (D–F) content along a 1:1 line. (A,D) represent WRNPM, (B,E) represent FM, and (C,F) represent TM. Black lines represent the fitted lines for each model. (Test set results are shown, and train set results are in Supplementary Table S3).
Figure 11. Measured vs. RF model-predicted leaf nitrogen (A–C) and phosphorus (D–F) content along a 1:1 line. (A,D) represent WRNPM, (B,E) represent FM, and (C,F) represent TM. Black lines represent the fitted lines for each model. (Test set results are shown, and train set results are in Supplementary Table S3).
Remotesensing 16 03261 g011
Remotesensing 16 03261 g012
Figure 12. Measured vs. RF model-predicted leaf nitrogen (A–C) and phosphorus (D–F) content along a 1:1 line. (A,D) represent the upper layer, (B,E) represent the middle layer, and (C,F) represent the lower layer. (Test set results are shown, and train set results are in Supplementary Table S4).
Figure 12. Measured vs. RF model-predicted leaf nitrogen (A–C) and phosphorus (D–F) content along a 1:1 line. (A,D) represent the upper layer, (B,E) represent the middle layer, and (C,F) represent the lower layer. (Test set results are shown, and train set results are in Supplementary Table S4).
Remotesensing 16 03261 g012
Table 1. Spectral parameters evaluated in this study.
Table 1. Spectral parameters evaluated in this study.
NO.NameFormulationReference
1Anthocyanin Reflectance IndexARI2 = R803(1/R549 − 1/R702)[34]
2Transformed Chlorophyll Absorbtion RatioTCARI = (R700/R670)/((1 + 0.16) (R800 − R670))[35]
3Transformed Soil Adjusted Vegetation IndexTSAVI = 1.09 ( R 800 − 1.09 R 600 − 4.04 ) ( R 600 + 1.09 R 800 − 4.04 + 0.18 ) [36]
4Modified Normalized Difference Vegetation IndexmNDVI705 = (R750 − R705)/(R705 + R705 − 2R445)[36]
5Photochemical Reflectance Index 570/515PRI(570,515) = (R570 − R515)/(R570 + R515)[37]
6Atmospherically Resistant Vegetation IndexARVI = (R872 − R488)/(R872 − R488 + 2R661)[38]
7DATT4DATT4 = R672/(R550 × R708)[39]
8Derivative of Normalized Difference 1DND1 = (D742 − D529)/(D742 + D529)[40]
9Derivative of Normalized Difference 7DND7 = (D742 − D702)/(D742 + D702)[40]
10Derivative Maximum DMAX42 = D712/D742[40]
11Modified Simple RatioMSRCHL = (R800 − R445)/(R680 − R445)[41]
Note: R represents spectral reflectance, and D represents the first derivative.
Table 2. First-order differential spectra in nitrogen and phosphorus inversion model.
Table 2. First-order differential spectra in nitrogen and phosphorus inversion model.
Band AreaBand Name
Nitrogen inversion modelVisible light regionBand 458, Band 462, Band 466, Band 467, Band 481, Band 551, Band 555, Band 623, Band 653, Band 654, Band 661, Band 743
Near infrared regionBand 911, Band 926, Band 927, Band 949, Band 1017, Band 1023, Band 1034, Band 1035
Phosphorus inversion modelVisible light regionBand 421, Band 430, Band 437, Band 446, Band 448, Band 453, Band 454, Band 460, Band 658, Band 660
Near infrared regionBand 985, Band 1018, Band 1019, Band 1027, Band 1038, Band 1040, Band 1042, Band 1055
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Li, D.; Hu, Q.; Zhang, J.; Dian, Y.; Hu, C.; Zhou, J. Leaf Nitrogen and Phosphorus Variation and Estimation of Citrus Tree under Two Labor-Saving Cultivation Modes Using Hyperspectral Data. Remote Sens. 2024, 16, 3261. https://doi.org/10.3390/rs16173261

AMA Style

Li D, Hu Q, Zhang J, Dian Y, Hu C, Zhou J. Leaf Nitrogen and Phosphorus Variation and Estimation of Citrus Tree under Two Labor-Saving Cultivation Modes Using Hyperspectral Data. Remote Sensing. 2024; 16(17):3261. https://doi.org/10.3390/rs16173261

Chicago/Turabian Style

Li, Dasui, Qingqing Hu, Jinzhi Zhang, Yuanyong Dian, Chungen Hu, and Jingjing Zhou. 2024. "Leaf Nitrogen and Phosphorus Variation and Estimation of Citrus Tree under Two Labor-Saving Cultivation Modes Using Hyperspectral Data" Remote Sensing 16, no. 17: 3261. https://doi.org/10.3390/rs16173261

APA Style

Li, D., Hu, Q., Zhang, J., Dian, Y., Hu, C., & Zhou, J. (2024). Leaf Nitrogen and Phosphorus Variation and Estimation of Citrus Tree under Two Labor-Saving Cultivation Modes Using Hyperspectral Data. Remote Sensing, 16(17), 3261. https://doi.org/10.3390/rs16173261

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop