Hyperspectral imaging for estimating leaf, flower, and fruit macronutrient concentrations and predicting strawberry yields

Dung, Cao Dinh; Trueman, Stephen J.; Wallace, Helen M.; Farrar, Michael B.; Gama, Tsvakai; Tahmasbian, Iman; Bai, Shahla Hosseini

doi:10.1007/s11356-023-30344-8

Hyperspectral imaging for estimating leaf, flower, and fruit macronutrient concentrations and predicting strawberry yields

Research Article
Open access
Published: 19 October 2023

Volume 30, pages 114166–114182, (2023)
Cite this article

Download PDF

You have full access to this open access article

Environmental Science and Pollution Research Aims and scope Submit manuscript

Hyperspectral imaging for estimating leaf, flower, and fruit macronutrient concentrations and predicting strawberry yields

Download PDF

Cao Dinh Dung^1,2,3,
Stephen J. Trueman⁴,
Helen M. Wallace^1,2,4,
Michael B. Farrar⁴,
Tsvakai Gama^1,2,
Iman Tahmasbian⁵ &
…
Shahla Hosseini Bai ORCID: orcid.org/0000-0001-8646-6423⁴

1498 Accesses
2 Citations
Explore all metrics

Abstract

Managing the nutritional status of strawberry plants is critical for optimizing yield. This study evaluated the potential of hyperspectral imaging (400–1,000 nm) to estimate nitrogen (N), phosphorus (P), potassium (K), and calcium (Ca) concentrations in strawberry leaves, flowers, unripe fruit, and ripe fruit and to predict plant yield. Partial least squares regression (PLSR) models were developed to estimate nutrient concentrations. The determination coefficient of prediction (R²_P) and ratio of performance to deviation (RPD) were used to evaluate prediction accuracy, which often proved to be greater for leaves, flowers, and unripe fruit than for ripe fruit. The prediction accuracies for N concentration were R²_P = 0.64, 0.60, 0.81, and 0.30, and RPD = 1.64, 1.59, 2.64, and 1.31, for leaves, flowers, unripe fruit, and ripe fruit, respectively. Prediction accuracies for Ca concentrations were R²_P = 0.70, 0.62, 0.61, and 0.03, and RPD = 1.77, 1.63, 1.60, and 1.15, for the same respective plant parts. Yield and fruit mass only had significant linear relationships with the Difference Vegetation Index (R² = 0.256 and 0.266, respectively) among the eleven vegetation indices tested. Hyperspectral imaging showed potential for estimating nutrient status in strawberry crops. This technology will assist growers to make rapid nutrient-management decisions, allowing for optimal yield and quality.

Estimation and map** of nitrogen content in apple trees at leaf and canopy levels using hyperspectral imaging

Article 30 April 2019

Identification of plant leaf phosphorus content at different growth stages based on hyperspectral reflectance

Article Open access 07 January 2021

Application of Vegetative Indices for Leaf Nitrogen Estimation in Sugarcane Using Hyperspectral Data

Article 17 October 2023

Use our pre-submission checklist

Avoid common mistakes on your manuscript.

Introduction

Global demand for food is increasing due to the continued rise in the world population (Godfray et al. 2010). Fertilizers play an important role in maximizing food production but fruit, nut, and seed yields are often sustained by excessive fertilizer use (Islam et al. 2022; Trejo-Téllez and Gómez-Merino 2014). Non-optimal fertilizer applications cause negative impacts on plant growth and development (Kulkarni and Goswami 2019). Over-fertilization, for example, can reduce yield and have negative impacts on the environment, while insufficient fertilization can also reduce yield and decrease product quality (González et al. 2015; Hapuarachchi et al. 2022; Pereira et al. 2015; Trejo-Téllez and Gómez-Merino 2014). Farmers often make decisions on nutrient amendments without full knowledge of the nutrient status of their crop (Bindraban et al. 2015; Islam et al 2022; Sheriff 2005) because current methods to examine plant nutrient status are laborious, costly, and time consuming (Yanli et al. 2015). Knowledge of plant nutrient status in real time would allow timely decisions on how much fertilizer needs to be added to a crop.

Hyperspectral imaging has been applied widely in agriculture, food, medicine, and other fields to estimate the internal qualities of scanned objects (Bai et al. 2018; Davur et al. 2023; ElMasry et al. 2012; Farrar et al. 2023; Gowen et al. 2007; Han et al. 2021; Huang et al. 2014; Malmir et al. 2020; Moscetti et al. 2015). Hyperspectral imaging combines spectroscopy with imaging techniques to acquire both spectral and spatial information simultaneously (ElMasry et al. 2012; Huang et al. 2014). Hyperspectral imaging is potentially non-destructive, low-cost, and reliable and has been applied to fruit, nuts, grains, and vegetables to estimate internal qualities such as total soluble solid concentration and moisture content, as well as firmness, ripeness, and shelf life (Bai et al. 2018; Davur et al. 2023; Gómez et al. 2006; Han et al. 2021; Han et al. 2023; Peng and Lu 2008; Pérez-Marín et al. 2009; Rajkumar et al. 2012; Ravikanth et al. 2017). Hyperspectral imaging allows the estimation of mineral nutrient concentrations such as nitrogen (N), phosphorus (P), potassium (K), and calcium (Ca) in the soil and leaves of many crops (Ferwerda et al. 2005; Mahajan et al. 2017; Pacumbaba and Beyl 2011; Pandey et al. 2017; Rodriguez et al. 2006; Tahmasbian et al. 2018; Yu et al. 2014). Hyperspectral imaging can also be used to estimate the concentrations of mineral nutrients including N, P, K, and Ca in avocado fruit (Kämper et al. 2020). Hyperspectral images obtained from the canopy or leaves have also been used to predict crop yield (Aparicio et al. 2000; Babar et al. 2006; Cao et al. 2015; Prasad et al. 2007; ** models to estimate nitrogen, phosphorus, potassium, and calcium concentrations. Scale bars = 1 cm

Full size image

Hyperspectral imaging

We used a laboratory-based hyperspectral imaging system (Benchtop System, Resonon, Bozeman, MT) with a 12-bit line scanner camera (Pika XC2, Bozeman, MT) to capture images (Farrar et al. 2021). The system comprised a hyperspectral camera with a 23-mm focal-length lens, a linear transition stage operated by a stepper motor, four current-controlled wide-spectrum quartz-halogen lights, and a main-control computer (Fig. A1).

We placed the samples on a black background on the transition stage of the camera. Images were captured using a visible/near-infrared hyperspectral imaging system in the spectral range of 400–1,000 nm. The spectral resolution was 1.3 nm, producing a total of 462 grayscale images that can be generated to form 462 bands of each hyperspectral image. The exposure time was adjusted to 27.3 ms and the stage moved at 1.23 mm s⁻¹.

SpectrononPro software package (Version 2.94, Resonon, Bozeman, MT) was used to extract reflectance data (spectral information) of the acquired leaf, flower, unripe, and ripe fruit images. The mean corrected relative reflectance (R) was calculated from the raw spectral reflectance, R₀, within SpectrononPro as shown in Eq. 1 (ElMasry et al. 2007):

$$R=\frac{{R}_{0}-D}{W-D}$$

(1)

where R₀ was the raw spectral reflectance, D was the reflectance of a dark image (camera lens covered), and W was the reflectance of a white Teflon board that reflected approximately 99% of incident light. This corrected for the spectral curve of the leaf, flower, or fruit surface. The 100% reflectivity was scaled to 10,000 (integers) by default. The mean corrected relative reflectance was used for model development.

Nutrient analysis

Nitrogen concentrations of leaf, flower, unripe-fruit, and ripe-fruit samples were determined by combustion analysis using a LECO 928 analyzer (LECO, Saint Joseph, MI) (McGeehan and Naylor 1988; Muñoz-Huerta et al. 2013; Rayment and Higginson 1992). Calcium, phosphorus, and potassium concentrations were analyzed by using inductively coupled plasma–atomic emission spectroscopy on samples that were open-vessel digested with a 5:1 mixture of nitric and perchloric acids (Munter and Grande 1980).

Data analysis and model development

Image acquisition and data extraction were conducted using SpectrononPro software (Version 2.94, Resonon, Bozeman, MT). The mean raw reflectance was extracted by marking a region of interest (ROI) for each image. The ROI for leaf, flower, unripe-fruit, and ripe-fruit images contained the surface of one side of the leaf, flower, unripe fruit, or ripe fruit, respectively (Fig. A2). We mixed three fruit together for nutrient analysis in the case of composite fruit samples and so hyperspectral data from these three fruit were averaged prior to data analysis and model development. Spectral outliers in the samples, if any, were detected using a Hotelling’s T² test (with 95% level of confidence) and removed from the data set (Farrar et al. 2021). The remaining data were divided randomly into two data sets, one used for calibration (80%) and the other (20%) used as a test data set to examine the precision of prediction using developed models (Table A1). We used a leave-one-out (full) cross-validation to evaluate the performance of the model (Dai et al. 2014; Tahmasbian et al. 2017; Zhang et al. 2013). This method uses the calibration data set but leaves one sample out of the calibration set each time and assesses the model using the remaining data. In the next iteration, another sample is left out randomly for the validation and this process continues until every sample is left out of the model once (Dai et al. 2014; Tahmasbian et al. 2017; Zhang et al. 2013).

We developed partial least square regression (PLSR) models. PLSR is one of the most frequently used modeling methods applied in hyperspectral imaging studies (De Silva et al. 2023; Han et al. 2023; Kämper et al. 2020; Mayr et al. 2021). The PLSR is commonly recommended when the dataset is small (Wold et al. 2001a). The PLSR is a linear multivariate model and relates data matrices of X and Y, the predicted and observable variables, respectively, by identifying smaller sets of predictors to perform a series of regressions (Wold et al. 2001a). These predictors have linear combinations, and the model is able to analyze datasets with noise and incomplete variables (Wold et al. 2001b). We also applied spectral pre-processing methods to decrease noise and improve model performance (Qin et al. 2013). The applied pre-processing methods were Smoothing Savitzky-Golay (Smoothing S-Golay, 1st derivative), Normalize, Derivative Savitzky-Golay (Derivative S-Golay), and Standard Normal Variate (SNV) (Qin et al. 2013). Developed models were assessed using the following indices: determination coefficients of calibration (R²_C), validation (R²_V), and prediction (R²_P); root mean squares error (RMS_E) of calibration (RMSE_C), validation (RMSE_V), and prediction (RMSE_P); and ratio of performance to deviation (RPD). Only the model with highest coefficient of determination (R²), ratio of performance to deviation (RPD), and lowest root mean squares error (RMSE) was selected to determine the accuracy of estimation for each targeted nutrient (Table 1). The R² and RMSE were calculated using Eqs. 2 and 3 (Yanli et al. 2015):

$${R}^{2}=1- \frac{{\sum }_{i=1}^{n}{\left({y}_{\mathrm{i}}-{\widehat{y}}_{\mathrm{i}}\right)}^{2}}{{\sum }_{i=1}^{n}{\left({y}_{\mathrm{i}}- \overline{y }\right)}^{2}}$$

(2)

$$\mathrm{RMSE}=\sqrt{\frac{\sum_{n=1 }^{n}{\left({\widehat{y}}_{\mathrm{i}} -{y}_{\mathrm{i}}\right)}^{2}}{n}}$$

(3)

where n indicated the number of samples, y_i and ŷ_i represented the reference and predicted values of the ith sample, respectively, and ȳ represented the mean of each reference value.

We then calculated the RPD using the test set (Farrar et al. 2021; Morellos et al. 2016). The RPD indicates the appropriateness of prediction. RPD classifications include RPD of 1.5–2.0 discriminates between high and low values, RPD of 2.0–2.5 provides a coarse quantitative prediction, and RPD > 2.5 and RPD > 3.0 provide good and excellent predictions, respectively (Nicolaï et al. 2007). The higher the RPD, the more robust is the model (Farrar et al. 2021; Kamruzzaman et al. 2012). RPD was defined using Eq. 4:

$$\mathrm{RPD}={\mathrm{SD}}_{\mathrm{TEST}}/{\mathrm{RMSE}}_{\mathrm{TEST}}$$

(4)

where SD_TEST was the standard deviation of the observed values and RMSE_TEST was the root mean square error of the prediction from the test set.

We identified specific wavelengths that were important for predicting N, P, K, and Ca concentrations using β-coefficient values that carry predictive information. The wavelengths with the highest β-coefficients contribute most to the predictive ability of the models (Iqbal et al. 2013; Malmir et al. 2019; Tahmasbian et al. 2021; Xu et al. 2018). We assessed the model accuracies in predicting nutrient concentrations of fruit, leaves, and flowers by comparing the best-fit model for each plant part, with this model being the one that had the highest RPD and R² of the dataset. Unscrambler® X software version 11 (CAMO Software Inc., Trondheim, Norway) was used for all computation, spectral data transformations, PLSR computations, outlier detection, and model development (Farrar et al. 2021).

Predicting yield and fruit mass

We attempted to predict both fruit yield and fruit mass using hyperspectral images. We also attempted to predict fruit yield and fruit mass from the macronutrient concentrations in leaf samples (n = 150) (Fig. 1). Yield was calculated as the total mass of fruit harvested from each plant during the study period of July to October 2018. Hyperspectral data from leaves, nutrient concentrations of leaves, and fruit mass were averaged for each plant, while yield was calculated as total fruit mass per plant. Stepwise regression was performed to evaluate linear regressions between the concentrations of N, P, K, and Ca in leaves or vegetation indices as the independent variable and fruit mass or yield as the dependent variable. Vegetation indices used to predict yield were Difference Vegetation Index (DVI), Modified Chlorophyll Absorption Ratio Index (MCARI), Modified Triangle Vegetation Index (MTVI), Normalized Difference Vegetation Index (NDVI), Photochemical Reflectance Index (PRI), Enhanced Vegetation Index (EVI), Ratio Vegetation Index (RVI), Infrared Percentage Vegetation Index (IPVI), Structure Independent Pigments Index (SIPI), and Red Edge Vegetation Stress Index (RVSI) (Rathod et al. 2013; Wang et al. 2018; Yu et al. 2018). The equations for all indices are presented in Table A2. Linear regressions were regarded as significant at p < 0.05.

Results

Descriptive statistics

Reflectance of the Vis/NIR spectrum (400–1,000 nm) from strawberry leaves, unripe fruit, and ripe fruit had low standard deviations, while the spectra of randomly harvested flowers (with sepals) had large standard deviations (Fig. 3). The calibration and test data sets used for develo** each of the models had comparable means and ranges (Table A1).

Estimating nitrogen concentrations

The PLSR model estimated N concentration in leaves with R² of 0.71, 0.57, and 0.64 and RMSE values of 2.44, 2.98, and 2.69 mg/kg for the calibration, validation, and test sets, respectively (Fig. 4 a, b). The RPD of foliar N prediction was 1.64 (Table A3; Fig. 4b). The R² values in estimating flower N concentrations were 0.56–0.71, RMSE values were 1.86–2.32 mg/kg, and the RPD was 1.59 (Table A3; Fig. 4 c, d). Estimation of N concentration in unripe fruit had high accuracy, with R² from 0.80–0.81, RMSE values from 47.61–55.17 mg/100 g, and RPD of 2.46 (Table A3; Fig. 4 e, f). In contrast, the PLSR model for estimating N concentration in ripe fruit had poor accuracy, with R² below 0.30, RMSE values from 34.27–37.50 mg/100 g, and RPD of 1.31 (Table A3; Fig. 4 g, h). High β-coefficients were observed at 450 nm, 550 nm, 620 nm, 690 nm, and 960 nm for estimating N concentration in leaves and ripe fruit; at 420 nm, 440 nm, 540 nm, 690 nm, 720 nm, and 960 nm for flowers; and at 400 nm, 520 nm, 620 nm, 720 nm, and 960 nm for unripe fruit (Fig. 5).

Estimating phosphorus concentrations

The R² values in estimating foliar P concentration were 0.49–0.66, RMSE values were 0.42–0.54 mg/kg, and the RPD value was 1.36 (Table A3; Fig. 6 a, b). Similarly, the R² values in estimating flower P concentration were 0.34–0.66, RMSE values were 0.30–0.48 mg/kg, and the RPD was 1.24 (Table A3; Fig. 6 c, d). High estimation accuracy was obtained in estimating the P concentration of unripe fruit, with R² of 0.81 for the calibration, validation, and test sets; RMSE values from 6.61–8.05 mg/100 g; and RPD of 2.30 (Table A3; Fig. 6 e, f). In contrast, estimation of P concentration for ripe fruit had lower accuracy, with R² values being 0.40–0.49, RMSE values being 3.83–4.17 mg/100 g, and the RPD being 1.54 (Table A3; Fig. 6 g, h). Prominent peaks were observed in the 520 nm, 590 nm, 660 nm, 790 nm, 930 nm, and 960 nm regions in the PLSR models for estimating P concentration in leaves and ripe fruit, and in the 400 nm, 550 nm, 690 nm, 760 nm, and 960 nm regions for flowers and unripe fruit (Fig. 7).

Estimating potassium concentrations

The R² values in estimating leaf K concentration were 0.66–0.76, RMSE values were from 2.07–2.48 mg/kg, and the RPD was 1.74 (Table A3; Fig. 8 a, b). The PLSR models to estimate K concentration in flowers had R² values from 0.36–0.49, RMSE values from 0.89–1.38 mg/kg, and RPD of 1.39 (Table A3; Fig. 8 c, d). The R² values in estimating K concentration of unripe fruit were 0.46–0.66, RMSE values were 38.78–42.25 mg/100 g, and RPD was 1.67 (Table A3; Fig. 8 e, f). However, low estimation accuracy was obtained in predicting the K concentration of ripe fruit, with R² values being 0.05–0.10 (Fig. 8 g, h). Weighted β-coefficients with high peaks were observed in the 400 nm, 750 nm, 960 nm, and 990 nm regions in the developed models for estimating K concentration in leaves, flowers, unripe fruit, and ripe fruit (Fig. 9).

Estimating calcium concentrations

The PLSR model estimated leaf Ca concentration with R² values of 0.75, 0.62, and 0.70 and RMSE values of 1.30, 1.62, and 1.42 mg/kg for the calibration, validation, and test set, respectively, and RPD of 1.77 (Table A3; Fig. 10 a, b). The model for estimating Ca concentration of flowers had R² values from 0.54–0.87, RMSE values from 0.41–0.76 mg/kg, and RPD of 1.63 (Table A3; Fig. 10 c, d). The model for estimating Ca concentration of unripe fruit provided R² values from 0.30–0.61, RMSE values from 16.72–17.80 mg/100 g, and RPD of 1.60 (Table A3; Fig. 10 e, f). The model for estimating Ca concentration of ripe fruit had R² values from 0.03–0.07, RMSE values from 26.72–27.54 mg/100 g, and RPD of 1.15 (Table A3; Fig. 10 g, h). High peaks were observed in the 410 nm, 420 nm, 620 nm, 650 nm, 690 nm, 920 nm, 960 nm, and 990 nm regions in the models for estimating leaf Ca concentration (Fig. 11). Prominent peaks were observed in the 400 nm, 420 nm, 550 nm, 660 nm, and 990 nm regions for fresh flowers, in the 400 nm, 550 nm, 590 nm, 690 nm, 920 nm, and 990 nm regions for unripe fruit, and in the 400 nm, 490 nm, 590 nm, 790 nm, 920 nm, and 990 nm regions for ripe fruit (Fig. 11).

Predicting yield and fruit mass

Fruit yield per plant (R² = 0.256) and fruit mass (R² = 0.226) had significant linear relationships with the Difference Vegetation Index, DVI _{[800, 670]}, based on the leaf reflectance spectrum (Table 1). Fruit yield per plant (R² = 0.294) and fruit mass (R² = 0.307) were also related significantly to leaf N concentration (Table 1). Fruit yield per plant and fruit mass were not related significantly to other vegetation indices or leaf nutrient concentrations (Table 1).

Table 1 Stepwise regressions to identify factors explaining fruit yield and fruit mass of strawberry using vegetation indices or leaf nutrient concentrations as independent parameters

Full size table

Discussion

Our results showed that visible and near infrared wavelengths (400–1,000 nm) had high accuracy in estimating nitrogen, phosphorus, potassium, and calcium concentrations in strawberry unripe fruit, and in estimating nitrogen, potassium and calcium concentrations in strawberry leaves and flowers. However, macronutrient concentrations in ripe fruit could not be estimated accurately. We also found that fruit yield per plant and fruit mass could be predicted using a hyperspectral-based vegetation index or leaf nitrogen concentrations. Hyperspectral imaging, therefore, has great potential for diagnosing the nutrient status of strawberry plants and providing decision support to amend fertilizer scheduling.

Nitrogen, phosphorus, potassium, and calcium concentrations in strawberry leaves, flowers, and unripe fruit were estimated with higher accuracies than in ripe fruit. The models for estimating nitrogen and phosphorus concentrations in unripe fruit provided coarse quantitative predictions, while the models for potassium and calcium concentrations in unripe fruit only provided discrimination between low and high values. Furthermore, the models for nitrogen and calcium concentrations in leaves and flowers and potassium concentrations in leaves also provided simple discrimination between low and high values. The ratio of performance to deviation (RPD) is one of the most commonly used criteria in determining model robustness for accurately predicting a variable. PLSR models can discriminate between low and high values, but not provide good quantitative predictions, when the RPD value is between 1.5 and 2.0 (Nicolaï et al. 2007). High-accuracy estimations of leaf macronutrient concentrations have been achieved in previous studies of strawberry and other crops such as cacao, citrus, corn, maize, pepper, and soybean using hyperspectral imaging, mainly under field conditions (Ercoli et al. 1993; España-Boquera et al. 2006; Pandey et al. 2017; Yanli et al. 2015; Yu et al. 2014; Zhu et al. 2006). We found that hyperspectral imaging could also be used to estimate macronutrient concentrations of strawberry flowers and unripe fruit using laboratory-based hyperspectral imaging, suggesting another potential approach for fertilizer planning.

Some mineral nutrients do not absorb light in the visible and near infrared regions and so visible and near infrared spectroscopy cannot detect these minerals directly (Manley 2014). Hyperspectral imaging can possibly detect these minerals and their concentrations indirectly, for example, when minerals bind to organic complexes (Manley 2014). Nitrogen exists in plant parts in several forms, including in amino acids, proteins, and chlorophyll molecules. Hyperspectral imaging systems that use visible and near infrared wavelengths can possibly detect nitrogen indirectly through chlorophyll-related compounds, because chlorophylls are strongly absorptive in the blue and red parts of the visible region (Ercoli et al. 1993; Pacumbaba and Beyl 2011; Pandey et al. 2017). Phosphorus is a component part of proteins and nucleic acids and so hyperspectral imaging can possibly detect phosphorus indirectly by detecting these organic macromolecules (Raven 2013). Potassium has a vital role in plant water absorption and osmotic potential regulation and accumulates as a mineral element in the vacuole, and so hyperspectral imaging may detect potassium concentrations indirectly through changes in water potential and solute concentrations that are related to potassium concentrations (Egilla et al. 2005; Malmir et al. 2020; Vago et al. 2009). Calcium may not be spectrally active, but estimation is possible when it binds with molecules that have covalent bonds such as N–H, S–H, O–H, C–H, C–O, or C = C (Bellon-Maurel et al. 2010; Manley 2014).

We were able to identify specific wavelengths important for estimating nitrogen, phosphorus, potassium, and calcium concentrations. Wavelengths with the highest β-coefficients contribute most to the predictive ability of models (Bai et al. 2018; Malmir et al. 2019). High β-coefficients at 400–450 nm, 530–550 nm, 590–650 nm, 700–780 nm, and 960–1000 nm have also been reported when estimating nitrogen, phosphorus, potassium, and calcium concentrations in cacao, pepper, and wheat leaves and avocado fruit (Hosseini-Bai et al. 2019; Kämper et al. 2020; Malmir et al. 2020; Yao et al. 2010; Yu et al. 2014). We found that prominent wavelengths for estimating nitrogen concentrations were in the regions of 400–550 nm, 620–720 nm, and 960–990 nm, which includes the blue and red regions of the visible spectrum at which chlorophyll is highly absorptive (Pandey et al. 2017). Chlorophyll concentrations are often significantly correlated with nitrogen concentrations (Bojović and Marković 2009). The wavelengths with high β-coefficients for estimating phosphorus, potassium, and calcium concentrations were in the regions of 570–600 nm, 710–730 nm, 770–820 nm, and 860–990 nm. Reflectance in the region of 460 nm or 670 nm is the result of electron transitions in chlorophyll a and b, and reflectance in the region of 950–1000 nm is associated with prominent molecular bonds such as O–H, C–H, and N–H in water, starch, and proteins (Curran 1989; Zur et al. 2000).

Strawberry fruit yield and fruit mass could be predicted using hyperspectral imaging of leaves using the Difference Vegetation Index (DVI). This index was the only vegetation parameter, among the 11 vegetation indices we examined, that had a significant linear relationship with yield or mass. Grain yield can also be predicted using DVI based on hyperspectral canopy reflectance (350–2,500 nm), but with higher r² values of 0.77–0.81 (Cao et al. 2015) compared with < 0.30 in our study. PLSR usually outperforms other deep learning models when small datasets are used, although deep learning techniques can be applied to increase the prediction accuracy when a larger dataset exists (Ludwig et al. 2019).

Strawberry yield was also predicted from foliar nitrogen concentrations, but not phosphorus, potassium, or calcium concentrations. Plant yield often has strong positive relationships with foliar nitrogen and chlorophyll concentrations (Reis et al. 2009). However, leaf nitrogen concentrations had a significant negative relationship (r = − 0.54) with yield in our study and in studies of apple (r = − 0.82) and green pepper (r = − 0.69) (Drake et al. 2002; Hassan et al. 1993). Nitrogen excess can reduce strawberry yield and fruit size (Trejo-Téllez and Gómez-Merino 2014). Highest yields of Tudla strawberry correspond to foliar nitrogen concentrations of 2.1–3.0% (Drake et al. 2002). Foliar nitrogen concentrations in Redlands Joy leaves were often between 3.0 and 4.5% (Dung et al. 2022), which was higher than in Tudla strawberry, and these high concentrations might explain the negative correlation between yield and nitrogen concentration in our study. The recommended concentrations for phosphorus, potassium, and calcium in Tudla strawberry leaves range from 0.20–0.38%, 1.84–2.21%, and 0.77–1.48%, respectively (Almaliotis et al. 2002). The concentrations of phosphorus, potassium, and calcium in our study were within these ranges (Dung et al. 2022), although there are no published recommendations for nutrient levels in Redlands Joy strawberry plants.

Conclusion

Laboratory-based hyperspectral imaging showed great potential for estimating nitrogen, phosphorus, potassium, and calcium concentrations in strawberry plants, with often-high estimation accuracies for leaves, flowers, and unripe fruit. The technology also showed potential for predicting yield and fruit mass using the Difference Vegetation Index. Hyperspectral imaging may, therefore, be used by strawberry growers to monitor plant nutrient status and manage fertilizer inputs in real time during flowering and fruit growth to ensure the best possible fruit yield and quality. Prediction of nitrogen, phosphorus, potassium, and calcium concentrations using hyperspectral imaging of ripe fruit will require further investigation using deep learning techniques.

Data availability

Data will be made available upon the request.

References

Almaliotis D, Velemis D, Bladenopoulou S, Karapetsas N (2002) Leaf nutrient levels of strawberries (cv. Tudla) in relation to crop yield. Acta Hortic 567:447–450
CAS Google Scholar
Aparicio N, Villegas D, Casadesus J, Araus JL, Royo C (2000) Spectral vegetation indices as nondestructive tools for determining durum wheat yield. Agron J 92:83–91
Google Scholar
Babar MA, Reynolds MP, van Ginkel M, Klatt AR, Raun WR, Stone ML (2006) Spectral reflectance indices as a potential indirect selection criteria for wheat yield under irrigation. Crop Sci 46:578–588
Google Scholar
Bai SH, Tahmasbian I, Zhou J, Nevenimo T, Hannet G, Walton DA, Randall B, Gama T, Wallace HM (2018) A non-destructive determination of peroxide values, total nitrogen and mineral nutrients in an edible tree nut using hyperspectral imaging. Comput Electron Agric 151:492–500
Google Scholar
Bellon-Maurel V, Fernandez-Ahumada E, Palagos B, Roger J-M, McBratney A (2010) Critical review of chemometric indicators commonly used for assessing the quality of the prediction of soil attributes by NIR spectroscopy. TrAC Trends Anal Chem 29:1073–1081
CAS Google Scholar
Bindraban PS, Dimkpa C, Nagarajan L, Roy A, Rabbinge R (2015) Revisiting fertilisers and fertilisation strategies for improved nutrient uptake by plants. Biol Ferttil Soils 51:897–911
CAS Google Scholar
Bojović B, Marković A (2009) Correlation between nitrogen and chlorophyll content in wheat (Triticum aestivum L.). Kragujevac J Sci 31:69–74
Google Scholar
Cao X, Luo Y, Zhou Y, Fan J, Xu X, West JS, Duan X, Cheng D (2015) Detection of powdery mildew in two winter wheat plant densities and prediction of grain yield using canopy hyperspectral reflectance. PLoS One 10:e0121462
Google Scholar
Chen F, Liu H, Yang H, Lai S, Cheng X, **n Y, Yang B, Hou H, Yao Y, Zhang S, Bu G, Deng Y (2011) Quality attributes and cell wall properties of strawberries (Fragaria annanassa Duch.) under calcium chloride treatment. Food Chem 126:450–459
CAS Google Scholar
Curran PJ (1989) Remote sensing of foliar chemistry. Remote Sens Environ 30:271–278
Google Scholar
Dai Q, Sun DW, **ong Z, Cheng JH, Zeng XA (2014) Recent advances in data mining techniques and their applications in hyperspectral image processing for the food industry. Compr Rev Food Scie Food Saf 13:891–905
Google Scholar
Davur YJ, Kämper W, Khoshelham K, Trueman SJ, Bai SH (2023) Estimating the ripeness of Hass avocado fruit using deep learning with hyperspectral imaging. Horticulturae 9(5):599
Google Scholar
De Silva AL, Trueman SJ, Kämper W, Wallace HM, Nichols J, Hosseini Bai S (2023) Hyperspectral imaging of adaxial and abaxial leaf surfaces as a predictor of macadamia crop nutrition. Plants 12:558
Google Scholar
Drake SR, Raese JT, Smith TJ (2002) Time of nitrogen application and its influence on ‘Golden Delicious’ apple yield and fruit quality. J Plant Nutr 25:143–157
CAS Google Scholar
Dung CD, Wallace HM, Bai SH, Ogbourne SM, Trueman SJ (2021) Cross-pollination affects fruit colour, acidity, firmness and shelf life of self-compatible strawberry. PLoS ONE 16:0256964
Google Scholar
Dung CD, Wallace HM, Bai SH, Ogbourne SM, Trueman SJ (2022) Biomass and mineral nutrient partitioning among self-pollinated and cross-pollinated fruit on the same strawberry plant. PLoS One 17:e0269485
Google Scholar
Dung CD, Wallace HM, Bai SH, Ogbourne SM, Trueman SJ (2023) Fruit size and quality attributes differ between competing self-pollinated and cross-pollinated strawberry fruit. Int J Fruit Sci 2023:1–12
Google Scholar
Egilla JN, Davies FT, Boutton TW (2005) Drought stress influences leaf water content, photosynthesis, and water-use efficiency of Hibiscus rosa-sinensis at three potassium concentrations. Photosynthetica 43:135–140
CAS Google Scholar
ElMasry G, Wang N, ElSayed A, Ngadi M (2007) Hyperspectral imaging for nondestructive determination of some quality attributes for strawberry. Food Eng 81:98–107
CAS Google Scholar
ElMasry G, Kamruzzaman M, Sun DW, Allen P (2012) Principles and applications of hyperspectral imaging in quality evaluation of agro-food products: a review. Crit Rev Food Sci Nutr 52:999–1023
Google Scholar
Ercoli L, Mariotti M, Masoni A, Massantini F (1993) Relationship between nitrogen and chlorophyll content and spectral properties in corn leaves. Eur J Agon 2:113–117
Google Scholar
España-Boquera ML, Cárdenas-Navarro R, López-Pérez L, Castellanos-Morales V, Lobit P (2006) Estimating the nitrogen concentration of strawberry plants from its spectral response. Commun Soil Sci Plant Anal 37:2447–2459
Google Scholar
FAOSTAT (2021) Retrieved from http://www.fao.org/faostat/en/#rankings/countries_by_commodity. Accessed 12 March 2021
Farrar MB, Wallace HM, Brooks P, Yule CM, Tahmasbian I, Dunn PK, Hosseini Bai S (2021) A performance evaluation of Vis/NIR hyperspectral imaging to predict curcumin concentration in fresh turmeric rhizomes. Remote Sens 13:1807
Google Scholar
Farrar MB, Wallace HM, Tahmasbian I, Yule CM, Dunn PK, Bai SH (2023) Rapid assessment of soil carbon and nutrients following application of organic amendments. Catena 223:106928
CAS Google Scholar
Ferwerda JG, Skidmore AK, Mutanga O (2005) Nitrogen detection with hyperspectral normalized ratio indices across multiple plant species. Int J Remote Sens 26408:3–4095
Google Scholar
Giampieri F, Tulipani S, Alvarez-Suarez JM, Quiles JL, Mezzetti B, Battino M (2012) The strawberry: composition, nutritional quality, and impact on human health. Nutr 28:9–19
CAS Google Scholar
Godfray HCJ, Beddington JR, Crute IR, Haddad L, Lawrence D, Muir JF, Pretty J, Robinson S, Thomas SM, Toulmin C (2010) Food security: the challenge of feeding 9 billion people. Sci 327:812–818
CAS Google Scholar
Gómez AH, He Y, Pereira AG (2006) Non-destructive measurement of acidity, soluble solids and firmness of Satsuma mandarin using Vis/NIR-spectroscopy techniques. J Food Eng 77:313–319
Google Scholar
González ME, Cea M, Medina J, González A, Diez MC, Cartes P, Monreal C, Navia R (2015) Evaluation of biodegradable polymers as encapsulating agents for the development of a urea controlled-release fertilizer using biochar as support material. Sci Total Environ 505:446–453
Google Scholar
Gowen AA, O’Donnell CP, Cullen PJ, Downey G, Frias JM (2007) Hyperspectral imaging – an emerging process analytical tool for food quality and safety control. Trends Food Sci Technol 18:590–598
CAS Google Scholar
Han Y, Liu Z, Khoshelham K, Bai SH (2021) Quality estimation of nuts using deep learning classification of hyperspectral imagery. Comput Electron Agric 180:105868
Google Scholar
Han Y, Bai SH, Trueman SJ, Khoshelham K, Kämper W (2023) Predicting the ripening time of ‘Hass’ and ‘Shepard’ avocado fruit by hyperspectral imaging. Precis Agric 1–17
Hannum SM (2004) Potential impact of strawberries on human health: a review of the science. Crit Rev Food Sci Nutr 44:1–17
CAS Google Scholar
Hapuarachchi NS, Kämper W, Wallace HM, Hosseini Bai S, Ogbourne SM, Nichols J, Trueman SJ (2022) Boron effects on fruit set, yield, quality and paternity of Hass avocado. Agron 12:1479
CAS Google Scholar
Hassan SA, Gerrer JM, Splittstoesser WE (1993) Growth and yield potential of green pepper as affected by nitrogen at transplanting. Pertanika J Trop Agric Sci 16:101–105
Google Scholar
Hosseini-Bai S, Trueman SJ, Nevenimo T, Hannet G, Randall B, Wallace HM (2019) The effects of tree spacing regime and tree species composition on mineral nutrient composition of cocoa beans and canarium nuts in 8-year-old cocoa plantations. Environ Sci Pollut Res 26:22021–22029
CAS Google Scholar
Huang H, Liu L, Ngadi MO (2014) Recent developments in hyperspectral imaging for assessment of food quality and safety. Sens 14:7248–7276
CAS Google Scholar
Iqbal A, Sun DW, Allen P (2013) Prediction of moisture, color and pH in cooked, pre-sliced turkey hams by NIR hyperspectral imaging system. J Food Eng 117:42–51
CAS Google Scholar
Islam MS, Bell RW, Miah MAM, Alam MJ (2022) Farmers’ fertilizer use gaps relative to government recommendations in the saline coastal zone of the Ganges Delta. Agron Sustain Dev 42:59
CAS Google Scholar
Kämper W, Trueman SJ, Tahmasbian I, Bai SH (2020) Rapid determination of nutrient concentrations in Hass avocado fruit by Vis/NIR hyperspectral imaging of flesh or skin. Remote Sens 12:3409
Google Scholar
Kamruzzaman M, ElMasry G, Sun DW, Allen P (2012) Non-destructive prediction and visualization of chemical composition in lamb meat using NIR hyperspectral imaging and multivariate regression. Innov Food Sci Emerg Technol 16:218–226
CAS Google Scholar
Kulkarni S, Goswami A (2019) Effect of excess fertilizers and nutrients: a review on impact on plants and human population. In: Proceedings of International Conference on Sustainable Computing in Science, Technology and Management (SUSCOM), Amity University Rajasthan, Jaipur-India, pp 2094–2099
Ludwig B, Murugan R, Parama VR, Vohland M (2019) Accuracy of estimating soil properties with mid-infrared spectroscopy: implications of different chemometric approaches and software packages related to calibration sample size. Soil Sci Soc Am J 83:1542–1552
CAS Google Scholar
Mahajan GR, Pandey RN, Sahoo RN, Gupta VK, Datta SC, Kumar D (2017) Monitoring nitrogen, phosphorus and sulphur in hybrid rice (Oryza sativa L.) using hyperspectral remote sensing. Precis Agric 18:736–761
Google Scholar
Mahmood T, Anwar F, Iqbal T, Ahmad I, Bhatti IA, Ashraf M (2012) Mineral composition of strawberry, mulberry and cherry fruits at different ripening stages as analyzed by inductively coupled plasma-optical emission spectroscopy. J Plant Nutr 35:111–122
CAS Google Scholar
Malmir M, Tahmasbian I, Xu Z, Farrar MB, Bai SH (2019) Prediction of soil macro-and micro-elements in sieved and ground air-dried soils using laboratory-based hyperspectral imaging technique. Geoderma 340:70–80
CAS Google Scholar
Malmir M, Tahmasbian I, Xu Z, Farrar MB, Bai SH (2020) Prediction of macronutrients in plant leaves using chemometric analysis and wavelength selection. J Soils Sed 20:249–259
CAS Google Scholar
Manley M (2014) Near-infrared spectroscopy and hyperspectral imaging: non-destructive analysis of biological materials. Chem Soc Rev 43:8200–8214
CAS Google Scholar
Mayr S, Beć KB, Grabska J, Wiedemair V, Pürgy V, Popp MA, Bonn GK, Huck CW (2021) Challenging handheld NIR spectrometers with moisture analysis in plant matrices: performance of PLSR vs. GPR vs. ANN modelling. Spectrochim Acta A Mol Biomol Spectrosc 249:119342
CAS Google Scholar
McGeehan SL, Naylor DV (1988) Automated instrumental analysis of carbon and nitrogen in plant and soil samples. Commun Soil Sci Plant Anal 19:493–505
CAS Google Scholar
Morellos A, Pantazi XE, Moshou D, Alexandridis T, Whetton R, Tziotzios G, Wiebensohn J, Bill R, Mouazen AM (2016) Machine learning based prediction of soil total nitrogen, organic carbon and moisture content by using VIS-NIR spectroscopy. Biosyst Eng 1521:104–116
Google Scholar
Moscetti R, Saeys W, Keresztes JC, Goodarzi M, Cecchini M, Danilo M, Massantini R (2015) Hazelnut quality sorting using high dynamic range short-wave infrared hyperspectral imaging. Food Bioproc Tech 8:1593–1604
Google Scholar
Muñoz-Huerta RF, Guevara-Gonzalez RG, Contreras-Medina LM, Torres-Pacheco I, Prado-Olivarez J, Ocampo-Velazquez RV (2013) A review of methods for sensing the nitrogen status in plants: advantages, disadvantages and recent advances. Sens 13:10823–10843
Google Scholar
Munter RC, Grande RA (1980) Plant tissue and soil extract analyses by ICP-atomic emission spectrometry. In: Developments in atomic plasma spectrochemical analysis: proceedings of international winter conference, San Juan, Puerto Rico, vol 7, no 11, pp 653–672
Nagata M, Tallada JG, Kobayashi T, Cui Y, Yoshinori G (2004) Predicting maturity quality parameters of strawberries using hyperspectral imaging. Am Soc Agric Biol Engin 1829:043033
Google Scholar
Nestby R, Lieten F, Pivot D, Lacroix CR, Tagliavini M (2005) Influence of mineral nutrients on strawberry fruit quality and their accumulation in plant organs: a review. Int J Fruit Sci 5:139–156
Google Scholar
Nicolaï BM, Beullens K, Bobelyn E, Peirs A, Saeys W, Theron KI, Lammertyn J (2007) Nondestructive measurement of fruit and vegetable quality by means of NIR spectroscopy: a review. Postharvest Biol Technol 46:99–118
Google Scholar
Pacumbaba RO, Beyl CA (2011) Changes in hyperspectral reflectance signatures of lettuce leaves in response to macronutrient deficiencies. Adv Space Res 48:32–42
CAS Google Scholar
Pandey P, Ge Y, Stoerger V, Schnable JC (2017) High throughput in vivo analysis of plant leaf chemical properties using hyperspectral imaging. Front Plant Sci 8:1348
Google Scholar
Peng Y, Lu R (2008) Analysis of spatially resolved hyperspectral scattering images for assessing apple fruit firmness and soluble solids content. Postharvest Biol Technol 48:52–62
Google Scholar
Pereira EI, Da Cruz CC, Solomon A, Le A, Cavigelli MA, Ribeiro C (2015) Novel slow-release nanocomposite nitrogen fertilizers: the impact of polymers on nanocomposite properties and function. Ind Eng Chem Res 54:3717–3725
CAS Google Scholar
Pérez-Marín D, Sánchez MT, Paz P, Soriano MA, Guerrero JE, Garrido-Varo A (2009) Non-destructive determination of quality parameters in nectarines during on-tree ripening and postharvest storage. Postharvest Biol Technol 52:180–188
Google Scholar
Prasad B, Carve BF, Stone ML, Babar MA, Raun WR, Klatt AR (2007) Potential use of spectral reflectance indices as a selection tool for grain yield in winter wheat under Great Plains conditions. Crop Sci 47:1426–1440
Google Scholar
Qin J, Chao K, Kim MS, Lu R, Burks TF (2013) Hyperspectral and multispectral imaging for evaluating food safety and quality. J Food Eng 118:157–171
CAS Google Scholar
Rajkumar P, Wang N, Elmasry G, Raghavan GSV, Gariepy Y (2012) Studies on banana fruit quality and maturity stages using hyperspectral imaging. J Food Eng 108:194–200
Google Scholar
Rathod PH, Rossiter DG, Noomen MF, van der Meer FD (2013) Proximal spectral sensing to monitor phytoremediation of metal-contaminated soils. Int J Phytoremediation 15:405–426
CAS Google Scholar
Raven JA (2013) RNA function and phosphorus use by photosynthetic organisms. Front Plant Sci 4:536
Google Scholar
Ravikanth L, Jayas DS, White NDG, Fields PG, Sun DW (2017) Extraction of spectral information from hyperspectral data and application of hyperspectral imaging for food and agricultural products. Food Bioproc Tech 10:1–33
CAS Google Scholar
Rayment GE, Higginson FR (1992) Australian laboratory handbook of soil and water chemical methods. Inkata, Melbourne, pp 330
Reis AR, Favarin JL, Malavolta E, Júnior JL, Moraes MF (2009) Photosynthesis, chlorophylls, and SPAD readings in coffee leaves in relation to nitrogen supply. Commun Soil Sci Plant Anal 40:1512–1528
CAS Google Scholar
Rodriguez D, Fitzgerald GJ, Belford R, Christensen LK (2006) Detection of nitrogen deficiency in wheat from spectral reflectance indices and basic crop eco-physiological concepts. Aust J Agric Res 57:781–789
CAS Google Scholar
Shao Y, He Y (2007) Nondestructive measurement of the acidity of strawberry based on wavelet transform and partial least squares. In: 2007 IEEE Instrumentation & Measurement Technology Conference IMTC 2007. IEEE, pp 1–5
Sheriff G (2005) Efficient waste? Why farmers over-apply nutrients and the implications for policy design. Rev Agric Econ 27:542–557
Google Scholar
Tahmasbian I, Xu Z, Abdullah K, Zhou J, Esmaeilani R, Nguyen TTN, Bai SH (2017) The potential of hyperspectral images and partial least square regression for predicting total carbon, total nitrogen and their isotope composition in forest litterfall samples. J Soils Sed 17:2091–2103
CAS Google Scholar
Tahmasbian I, Xu Z, Boyd S, Zhou J, Esmaeilani R, Che R, Bai SH (2018) Laboratory-based hyperspectral image analysis for predicting soil carbon, nitrogen and their isotopic compositions. Geoderma 330:254–263
CAS Google Scholar
Tahmasbian I, Wallace HM, Gama T, Bai SH (2021) An automated non-destructive prediction of peroxide value and free fatty acid level in mixed nut samples. LWT - Food Sci Technol 143:110893
CAS Google Scholar
Tian H, Wang T, Liu Y, Qiao X, Li Y (2020) Computer vision technology in agricultural automation – a review. Inf Process Agric 7:1–19
CAS Google Scholar
Trejo-Téllez LI, Gómez-Merino FC (2014) Nutrient management in strawberry: effects on yield, quality and plant health. In: Malone N (ed) Strawberries. Nova Science Publishers, New York, pp 239–267
Google Scholar
Vago I, Tolner L, Loch J (2009) Effect of chloride anionic stress on the yield amount and some quality parameters of strawberry (Fragaria ananassa). Cereal Res Commun 37:81–84
CAS Google Scholar
Wang F, Gao J, Zha Y (2018) Hyperspectral sensing of heavy metals in soil and vegetation: feasibility and challenges. J Photogramm Remote Sens 136:73–84
Google Scholar
Wold S, Sjöström M, Eriksson L (2001a) PLS-regression: a basic tool of chemometrics. Chemometr Intell Lab Syst 58:109–130
CAS Google Scholar
Wold S, Trygg J, Berglund A, Antti H (2001b) Some recent developments in PLS modeling. Chemometr Intell Lab Syst 58:131–150
CAS Google Scholar
**e Y, Wang C, Yang W, Feng M, Qiao X, Song J (2020) Canopy hyperspectral characteristics and yield estimation of winter wheat (Triticum aestivum) under low temperature injury. Sci Rep 10:244
CAS Google Scholar
Xu S, Zhao Y, Wang M, Shi X (2018) Comparison of multivariate methods for estimating selected soil properties from intact soil cores of paddy fields by VIS–NIR spectroscopy. Geoderma 310:29–43
CAS Google Scholar
Yanli L, Qiang L, Shaolan H, Shilai Y, Xuefeng L, Rang** X, Yongqiang Z, Lie D (2015) Prediction of nitrogen and phosphorus contents in citrus leaves based on hyperspectral imaging. Int J Agric Biol Eng 8:80–88
Google Scholar
Yao X, Zhu Y, Tian Y, Feng W, Cao W (2010) Exploring hyperspectral bands and estimation indices for leaf nitrogen accumulation in wheat. Int J Appl Earth Obs Geoinf 12:89–100
Google Scholar
Yu KQ, Zhao YR, Li XL, Shao YN, Liu F, He Y (2014) Hyperspectral imaging for map** of total nitrogen spatial distribution in pepper plant. PLoS One 9:e116205
Google Scholar
Yu H, Kong B, Wang G, Du R, Qie G (2018) Prediction of soil properties using a hyperspectral remote sensing method. Arch Agron Soil Sci 64:546–559
Google Scholar
Zhang X, Liu F, He Y, Gong X (2013) Detecting macronutrients content and distribution in oilseed rape leaves based on hyperspectral imaging. Biosyst Eng 115:56–65
Google Scholar
Zhu Y, Li Y, Feng W, Tian Y, Yao X, Cao W (2006) Monitoring leaf nitrogen in wheat using canopy reflectance spectra. Can J Plant Sci 86:1037–1046
Google Scholar
Zur Y, Gitelson AA, Chivkunova OB, Merzlyak MN (2000) The spectral contribution of carotenoids to light absorption and reflectance in green leaves. Proceedings of the 2nd International Conference of Geospatial Information in Agriculture and Forestry. Lake Buena Vista, pp 7

Download references

Acknowledgements

Thanks are due to Steven Ogbourne for manuscript editing and Tracey McMahon, Bruce Randall, Darren Morrow, and Michael Nielsen for assistance.

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions Open Access funding enabled and organized by CAUL and its Member Institutions Cao Dinh Dung was supported by a scholarship from the Vietnam Ministry of Education and Training (MOET) and the University of the Sunshine Coast (USC). The study was supported by Project PH16001 of the Hort Frontiers Pollination Fund, part of the Hort Frontiers strategic partnership initiative developed by Hort Innovation, with co-investment from USC, Griffith University, Plant & Food Research Ltd and contributions from the Australian Government.

Author information

Authors and Affiliations

Centre for Bioinnovation, University of the Sunshine Coast, 90 Sippy Downs Drive, Sippy Downs, QLD, 4556, Australia
Cao Dinh Dung, Helen M. Wallace & Tsvakai Gama
School of Science, Technology and Engineering, University of the Sunshine Coast, 90 Sippy Downs Drive, Sippy Downs, QLD, 4556, Australia
Cao Dinh Dung, Helen M. Wallace & Tsvakai Gama
Potato, Vegetable and Flower Research Center – Institute of Agricultural Science for Southern Vietnam, Thai Phien Village, Ward 12, Da Lat, Lam Dong, Vietnam
Cao Dinh Dung
Centre for Planetary Health and Food Security, School of Environment and Science, Griffith University, Nathan, Brisbane, QLD, 4111, Australia
Stephen J. Trueman, Helen M. Wallace, Michael B. Farrar & Shahla Hosseini Bai
Department of Agriculture and Fisheries, Queensland Government, Toowoomba, QLD, 4350, Australia
Iman Tahmasbian

Authors

Cao Dinh Dung
View author publications
You can also search for this author in PubMed Google Scholar
Stephen J. Trueman
View author publications
You can also search for this author in PubMed Google Scholar
Helen M. Wallace
View author publications
You can also search for this author in PubMed Google Scholar
Michael B. Farrar
View author publications
You can also search for this author in PubMed Google Scholar
Tsvakai Gama
View author publications
You can also search for this author in PubMed Google Scholar
Iman Tahmasbian
View author publications
You can also search for this author in PubMed Google Scholar
Shahla Hosseini Bai
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization, S.J.T., H.M.W., S.H.B., and C.D.D.; methodology, S.J.T., H.M.W., S.H.B., and C.D.D.; formal analysis, C.D.D., S.J.T., and H.M.W.; writing—original draft preparation, C.D.D.; writing—review and editing, S.J.T., H.M.W., S.H.B., M.B.F., and T.G, I.B.; supervision, S.J.T., H.M.W., and S.H.B.; funding acquisition, S.J.T., H.M.W., S.H.B., and C.D.D. All authors have read and agreed to the published version of the manuscript.

Corresponding author

Correspondence to Shahla Hosseini Bai.

Ethics declarations

Ethical approval and consent to participate

Not applicable.

Consent for publication

All co-authors have agreed to publish this paper.

Conflict of interest

The authors declare no competing interests.

Additional information

Responsible Editor: Philippe Garrigues

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 1325 KB)

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Dung, C.D., Trueman, S.J., Wallace, H.M. et al. Hyperspectral imaging for estimating leaf, flower, and fruit macronutrient concentrations and predicting strawberry yields. Environ Sci Pollut Res 30, 114166–114182 (2023). https://doi.org/10.1007/s11356-023-30344-8

Download citation

Received: 13 January 2023
Accepted: 04 October 2023
Published: 19 October 2023
Issue Date: November 2023
DOI: https://doi.org/10.1007/s11356-023-30344-8

Keywords

Use our pre-submission checklist

Avoid common mistakes on your manuscript.

Hyperspectral imaging for estimating leaf, flower, and fruit macronutrient concentrations and predicting strawberry yields

Abstract

Similar content being viewed by others

Estimation and map** of nitrogen content in apple trees at leaf and canopy levels using hyperspectral imaging

Identification of plant leaf phosphorus content at different growth stages based on hyperspectral reflectance

Application of Vegetative Indices for Leaf Nitrogen Estimation in Sugarcane Using Hyperspectral Data

Introduction

Hyperspectral imaging

Nutrient analysis

Data analysis and model development

Predicting yield and fruit mass

Results

Descriptive statistics

Estimating nitrogen concentrations

Estimating phosphorus concentrations

Estimating potassium concentrations

Estimating calcium concentrations

Predicting yield and fruit mass

Discussion

Conclusion

Data availability

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Ethical approval and consent to participate

Consent for publication

Conflict of interest

Additional information

Publisher's Note

Supplementary Information

Supplementary file1 (DOCX 1325 KB)

Rights and permissions

About this article

Cite this article

Share this article

Keywords

Search

Navigation