PLSR Models Research Articles

Geographical origin differentiation and quality determination of saffron using a portable Hyperspectral imaging system

A portable Hyperspectral Imaging (HSI) system (400–1000 nm) coupled with chemometrics techniques was developed to identify the geographical origin and to determine the Crocins content of saffron rapidly and non-destructively. Reflectance spectra of 160 saffron samples from two geographical origins were captured and preprocessed using the Savitzky–Golay (SG) algorithm. PCA as an unsupervised clustering and PLS-DA, RBF, and MLP as supervised techniques were applied for the classification and geographical origin differentiation of the samples. Afterward, the sample’s Crocins content was measured and their associations with their spectra were modeled using PLSR, RBF, and MLP models. The PCA results indicated that saffron samples were originally distributed in two different clusters. PLS-DA model with a spectral range from 400 to 1000 nm achieved the best clustering performance (Sensitivity and Specificity: 100 %). The RBF model showed good predictive capabilities in the saffron Crocins content prediction and quality evaluation (R2cal > 0.94 and R2pred > 0.83).

Hyperspectral Imaging Combined with Chemometrics Analysis for Monitoring the Textural Properties of Modified Casing Sausages with Differentiated Additions of Orange Extracts

The textural properties (hardness, springiness, gumminess, and adhesion) of 16-day stored sausages with different additions of orange extracts to the modified casing solution were estimated by response surface methodology (RSM) and a hyperspectral imaging system in the spectral range of 390-1100 nm. To improve the model performance, normalization, 1st derivative, 2nd derivative, standard normal variate (SNV), and multiplicative scatter correction (MSC) were applied for spectral pre-treatments. The raw, pretreated spectral data and textural attributes were fit to the partial least squares regression model. The RSM results show that the highest R2 value achieved at adhesion (77.57%) derived from a second-order polynomial model, and the interactive effects of soy lecithin and orange extracts on adhesion were significant (p < 0.05). The adhesion of the PLSR model developed from reflectance after SNV pretreatment possessed a higher calibration coefficient of determination (0.8744) than raw data (0.8591). The selected ten important wavelengths for gumminess and adhesion can simplify the model and can be used for convenient industrial applications.

Rapid detection of waste cooking oil using low-field nuclear magnetic resonance

The illegal circulation of waste cooking oil (WCO) in the market makes the development of an efficient detection method an urgent task. Therefore, 20 batches of WCOs were collected and compared with 15 kinds of edible oils using low-field nuclear magnetic resonance (LF-NMR) combined with the chemometrics method. LDA, PLS-DA and SVM-DA models were proven to totally discriminate WCO from 15 kinds of edible oils, and T 22 -related parameters were crucial factors in the discrimination processes. WCO was blended at a ratio from 5% to 50% into corn oil (CO), olive oil (OO), peanut oil (PO), rapeseed oil (RO) and soybean oil (SO) in the following adulteration assay. The SVM-DA model showed a better classification effect than the LDA and PLS-DA models in differentiating edible oils from corresponding WCO-adulterated oils, and WCO-adulterated CO was relatively more difficult (80%<validation accuracy<88.6%) to classify than the other groups (82.9%<validation accuracy<100%). Finally, four quantitative models of WCO-adulterated OO, PO, RO and SO were well established using the PLSR method, with R 2 > 0.956, RMSEC<0.035 and RMSEP<0.038. These results illustrate that LF-NMR combined with chemometrics analysis can provide promising rapid screening of WCO. • LF-NMR detection distinguished 20 batches of WCOs from 15 kinds of edible oils. • The parameters of T 22 fraction were crucial factors to discriminate WCOs from edible oils. • SVM-DA model showed better classification effect than the LDA and PLS-DA models. • PLSR model was established to calculate the adulteration level of WCOs.

Screening for pesticide residues in cocoa (Theobroma cacao L.) by portable infrared spectroscopy

Rapid assessment of pesticide residues ensures cocoa bean quality and marketability. In this study, a portable FTIR instrument equipped with a triple reflection attenuated total reflectance (ATR) accessory was used to screen cocoa beans for pesticide residues. Cocoa beans (n = 75) were obtained from major cocoa growing regions of Peru and were quantified for pesticides by gas chromatography (GC) or liquid chromatography (LC) coupled with mass spectrometry (MS). The FTIR spectra were used to detect the presence of pesticides in cocoa beans or lipid fraction (butter) by using a pattern recognition (Soft Independent Modeling by Class Analogy, SIMCA) algorithm, which produced a significant discrimination for cocoa nibs (free or with pesticides). The variables related to the class grouping were assigned to the aliphatic (3200–2800 cm−1) region with an interclass distance (ICD) of 3.3. Subsequently, the concentration of pesticides in cocoa beans was predicted using a partial least squares regression analysis (PLSR), using an internal validation of the PLRS model, the cross-validation correlation coefficient (Rval = 0.954) and the cross-validation standard error (SECV = 14.9 mg/kg) were obtained. Additionally, an external validation was performed, obtaining the prediction correlation coefficient (Rpre = 0.940) and the standard error of prediction (SEP = 16.0 μg/kg) with high statistical performances, which demonstrates the excellent predictability of the PLSR model in a similar real application. The developed FTIR method presented limits of detection and quantification (LOD = 9.8 μg/kg; LOQ = 23.1 μg/kg) with four optimum factors (PC). Mid-infrared spectroscopy (MIR) offered a viable alternative for field screening of cocoa.

Detection of mango soluble solid content using hyperspectral imaging technology

Soluble solid content (SSC) is an important indicator for evaluating mango quality. The main task of this study is to develop a partial least squares (PLS) regression model for SSC by combinating the visible and near infrared (400–1000 nm) hyperspectral imaging. The PLSR model can be used to assess the quality grading of mangoes. By comparing the performance of five preprocessing full-band models, the standard normal variable transformation (SNV) and multiplicative scatter correction algorithm (MSC) are selected for this study. Otherwise, three variable selection methods, including successive projection algorithm (SPA), competitive adaptive reweighted sampling (CARS) method, and the genetic algorithm (GA) are used for the identification of the characteristic wavelengths. The screened feature bands are used to build PLS regression models.The SNV-CARS-PLS model is found to show the best prediction performance. The correlation coefficient for the predicted value for the mango SSC and its root mean square error are determined to be 0.9001 and 0.6162, respectively. These results suggest that the SNV-CARS-PLS model is an effective method for predicting mango SSC.

Accurate Detection of Cd2+ and Pb2+ Concentrations in Soils by Stripping Voltammetry Peak Areas under the Mutual Interference of Multiple Heavy Metals

The accurate detection of Cd2+ and Pb2+ in soils by square-wave anodic stripping voltammetry (SWASV) faces great challenges because the interaction between multiple heavy metal ions (HMIs) interferes seriously with their SWASV signals. To detect Cd2+ and Pb2+ by SWASV with high accuracy, an overlooked but informative signal, i.e., stripping current peak area, was employed and combined with chemometric methods to suppress the above mutual interference. An easy-to-prepare electrode, i.e., in-site electroplating bismuth film modified glassy carbon electrode, was used to sense the multiple HMIs. Two machine learning algorithms, including SVR and PLSR, were used to establish the detection models of Cd2+ and Pb2+. In addition, this study developed a homemade algorithm to automatically acquire the stripping peak heights and stripping peak areas of Zn2+, Cd2+, Pb2+, Bi3+, and Cu2+, which acted as the inputs of machine learning models. Then, the detection performance of various SVR and PLSR models were compared based on the R2 and RMSE values of the validation dataset. Results showed that the SVR detection models established by the algorithmically acquired peak areas presented the best stability and accuracy for detecting both Cd2+ and Pb2+ concentrations under the existence of Zn2+ and Cu2+. The R2 and RMSE values of the SVR models built using the peak heights of HMIs acquired by electrochemical workstation control software (Imanu-SVR) were 0.7650 and 5.3916 μg/L for Cd2+, and 0.8791 and 20.0015 μg/L for Pb2+, respectively; the R2 and RMSE values of the SVR models built using the peak area automatically acquired by the developed algorithm (Aalgo-SVR) were 0.9204 and 2.9906 μg/L for Cd2+, and 0.9756 and 13.1574 μg/L for Pb2+, respectively. More importantly, the detection results of the proposed method in real soil extracts for Cd2+ and Pb2+ concentrations were close to those of ICP-MS, verifying its practicability. This study provides a new solution for the accurate detection of targeted heavy metals under the co-existence of multiple HMIs by the SWASV method.

Application of portable near‐infrared spectroscopy for rapid detection and quantification of adulterants in baobab fruit pulp

SummaryBaobab fruit pulp powder (BFPP) is susceptible to economically driven adulteration owing to its incredible nutrient density and rapidly expanding demand worldwide. In this study, a portable NIR spectroscopy (wavelength 900–1700 nm) coupled with chemometrics was used for the detection of BFPP adulteration. BFPP samples separately adulterated with rice flour (RF), wheat flour (WF), and maize flour (MF) at 0%, 1%, 3%, 5%, 10%, 15%, 20%, 25%, 30%, 40%, 50% and 60% concentrations were subjected to NIR spectroscopy. Two‐class models proved to be reliable with sensitivity and specificity of above 0.98 and an error of below 0.01. Four‐class models attained sensitivity and specificity of above 0.68 and an error of below 0.276. The correlation coefficient (R2) and root mean square error (RMSE) of the prediction set were above 0.88 and below 6.20% respectively for PLSR models. The LODs were also below 13.79%. Therefore, NIR spectroscopy has a promising potential for rapid screening of BFPP adulterations.

Rapid determination of rebaudioside A content in the macroporous resin elution process using a portable near-infrared spectrometer

Portable NIR spectroscopy is used to monitor the macroporous resin elution process. The PLSR and SVMR models displayed good performance for determining the RA content. The PLSR model displayed higher predictive accuracy than the SVMR model.

Visualization and prediction of TVB-N content in chilled pork by hyperspectral imaging

TVB-N is one of the important indexes of the freshness of chilled pork. To present the freshness more accurately and completely, a visualization technology of chemical composition was applied and optimized for pork freshness detection. Two hyperspectral cameras covering visible and near-infrared wavebands were used to detect the TVB-N content. GA, SPA, iPLSR, and siPLSR were employed to select characteristic wavebands from the spectra data extracted from the hyperspectral images, and PLSR models were established on full wavebands and characteristic wavebands. According to the models established on the spectral data, TVB-N distribution maps were obtained, and URV and ICV were employed to evaluate the performance of distribution maps. The results showed that, both PLSR models established on the characteristic wavebands and full wavebands achieved good accuracy, and that the characteristic wavebands cluster was in the spectral range of 600-750 nm, 1020-1120 nm, and 1450-1570 nm. To optimize the performance of distribution maps, prediction models were reconstructed under the guidance of URV and ICV. This study provided a method of predicting TVB-N content of pork in both spectral and imaging aspect for online testing the freshness of chilled pork, which improves the efficiency and quality of testing.

Quantitative analysis of Cd based on the stress effect of minerals in rice by laser-induced breakdown spectroscopy.

Cadmium (Cd) is a highly toxic heavy metal that can accumulate in the food chain, posing a significant threat to human health. One of the key food sources through which Cd is often observed is rice. Therefore, determining heavy metals in rice is essential to assess the risk status of rice. Laser-induced breakdown spectroscopy (LIBS) has the advantages of simple sample preparation and fast analysis, which is expected to achieve real-time and rapid detection of rice. In this work, 40 naturally matured rice samples growing from the area that is possibly contaminated with Cd were collected to determine the Cd reference content in rice by graphite furnace atomic absorption spectroscopy as recommended by the Chinese National Standard. LIBS spectral acquisition and analysis are adopted as well. The Cd characteristic spectral lines were selected to predict the Cd content directly using PCA, PLSR, and ELM models, and the coefficient of determination (R2) of the models' training and prediction sets was 0.9278, 0.8920; 0.9036, 0.9771; 0.7940, and 0.8409, respectively. Further, based on the Cd stress effect in rice, the spectra of elements Mn, Mg, K, and Na with highly significant and significant correlation with Cd were selected and coupled with the Cd characteristic spectra to form a new matrix of the same size for quantitative analysis. Based on the stress effect, R2 of models' training and prediction sets was improved to 0.9786, 0.9753; 0.9395, 0.9900; 0.9798, and 0.9927, respectively. It is demonstrated that combining the stress effect when using LIBS for quantitative analysis of Cd in rice reduces the overfitting and further improves the model's prediction accuracy. This work indicates that using LIBS combined with suitable mathematical models to predict the Cd content of naturally matured rice based on stress effects in rice is feasible. It is promising to evaluate the safety of rice by analyzing LIBS spectra.

A method based on a one-dimensional convolutional neural network for UV-vis spectrometric quantification of nitrate and COD in water under random turbidity disturbance scenario.

This paper proposed a novel spectrometric quantification method for nitrate and COD concentration in water using a double-channel 1-D convolution neural network for relatively long UV-vis absorption spectra data (2600 points). To improve the model's ability to resist turbidity disturbance, a new dataset augmentation method was applied and the absorption spectra of nitrate and COD under different turbidity disturbances were successfully simulated. Compared to the PLSR model, the value of RRMSEP for the CNN model was reduced from 6.1% to 1.4% in nitrate solution and 4.5% to 1.3% in COD solution. Compared to the PLSR model, the regression accuracy of the CNN model was increased from 56% to 93% in nitrate solution and 68% to 91% in COD solution. The test on the actual solution under different turbidity disturbances shows that the 1D-CNN model had a bias rate of less than 2% in both nitrate and COD solutions, while the worst bias rate in the PLSR method was 15%.

Rapid Determination of Geniposide and Baicalin in Lanqin Oral Solution by Near-Infrared Spectroscopy with Chemometric Algorithms during Alcohol Precipitation

The selection of key variables is an important step that improves the prediction performance of a near-infrared (NIR) real-time monitoring system. Combined with chemometrics, NIR spectroscopy was employed to construct high predictive accuracy, interpretable models for the rapid detection of the alcohol precipitation process of Lanqin oral solution (LOS). The variable combination population analysis-iteratively retaining informative variables (VCPA-IRIV) was innovatively introduced into the variable screening process of the model of geniposide and baicalin. Compared with the commonly used synergy interval partial least squares regression, competitive adaptive reweighted sampling, and random frog, VCPA-IRIV achieved the maximum compression of variable space. VCPA-IRIV-partial least squares regression (PLSR) only needs to use about 1% of the number of variables of the original data set to construct models with Rp values greater than 0.95 and RMSEP values less than 10%. With the advantages of simplicity and strong interpretability, the prediction ability of the PLSR models had been significantly improved simultaneously. The VCPA-IRIV-PLSR models met the requirements of rapid quality detection. The real-time detection system can help researchers to understand the quality rules of geniposide and baicalin in the alcohol precipitation process of LOS and provide a reference for the optimization of a LOS quality control system.

VIS-NIR-SWIR Hyperspectroscopy Combined with Data Mining and Machine Learning for Classification of Predicted Chemometrics of Green Lettuce

VIS-NIR-SWIR hyperspectroscopy is a significant technique used in remote sensing for classification of prediction-based chemometrics and machine learning. Chemometrics, together with biophysical and biochemical parameters, is a laborious technique; however, researchers are very interested in this field because of the benefits in terms of optimizing crop yields. In this study, we investigated the hypothesis that VIS-NIR-SWIR could be efficiently applied for classification and prediction of leaf thickness and pigment profiling of green lettuce in terms of reflectance, transmittance, and absorbance data according to the variety. For this purpose, we used a spectroradiometer in the visible, near-infrared, and shortwave ranges (VIS-NIR-SWIR). The results showed many chemometric parameters and fingerprints in the 400–2500 nm spectral curve range. Therefore, this technique, combined with rapid data mining, machine learning algorithms, and other multivariate statistical analyses such as PCA, MCR, LDA, SVM, KNN, and PLSR, can be used as a tool to classify plants with the highest accuracy and precision. The fingerprints of the hyperspectral data indicated the presence of functional groups associated with biophysical and biochemical components in green lettuce, allowing the plants to be correctly classified with higher accuracy (99 to 100%). Biophysical parameters such as thickness could be predicted using PLSR models, which showed R2P and RMSEP values greater than &gt;0.991 and 6.21, respectively, according to the relationship between absorbance and reflectance or transmittance spectroscopy curves. Thus, we report the methodology and confirm the ability of VIS-NIR-SWIR hyperspectroscopy to simultaneously classify and predict data with high accuracy and precision, at low cost and with rapid acquisition, based on a remote sensing tool, which can enable the successful management of crops such as green lettuce and other plants using precision agriculture systems.

Classification and Prediction by Pigment Content in Lettuce (Lactuca sativa L.) Varieties Using Machine Learning and ATR-FTIR Spectroscopy.

Green or purple lettuce varieties produce many secondary metabolites, such as chlorophylls, carotenoids, anthocyanins, flavonoids, and phenolic compounds, which is an emergent search in the field of biomolecule research. The main objective of this study was to use multivariate and machine learning algorithms on Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy (ATR-FTIR)-based spectra to classify, predict, and categorize chemometric attributes. The cluster heatmap showed the highest efficiency in grouping similar lettuce varieties based on pigment profiles. The relationship among pigments was more significant than the absolute contents. Other results allow classification based on ATR-FTIR fingerprints of inflections associated with structural and chemical components present in lettuce, obtaining high accuracy and precision (>97%) by using principal component analysis and discriminant analysis (PCA-LDA)-associated linear LDA and SVM machine learning algorithms. In addition, PLSR models were capable of predicting Chla, Chlb, Chla+b, Car, AnC, Flv, and Phe contents, with R2P and RPDP values considered very good (0.81−0.88) for Car, Anc, and Flv and excellent (0.91−0.93) for Phe. According to the RPDP metric, the models were considered excellent (>2.10) for all variables estimated. Thus, this research shows the potential of machine learning solutions for ATR-FTIR spectroscopy analysis to classify, estimate, and characterize the biomolecules associated with secondary metabolites in lettuce.

The Effect of Suspended Particulate Matter on the Supraglacial Lake Depth Retrieval from Optical Data

Supraglacial lakes (SGL) are a specific phenomenon of glaciers. They are important for ice dynamics, surface mass balance, and surface hydrology, especially during ongoing climate changes. The important characteristics of lakes are their water storage and drainage. Satellite-based remote sensing is commonly used not only to monitor the area but also to estimate the depth and volume of lakes, which is the basis for long-term spatiotemporal analysis of these phenomena. Lake depth retrieval from optical data using a physical model requires several basic assumptions such as, for instance, the water has little or no dissolved or suspended matter. Several authors using these assumptions state that they are also potential weaknesses, which remain unquantified in the literature. The objective of this study is to quantify the effect of maximum detectable lake depth for water with non-zero suspended particulate matter (SPM). We collected in-situ concurrent measurements of hyperspectral and lake depth observations to a depth of 8 m. Additionally, we collected water samples to measure the concentration of SPM. The results of empirical and physically based models proved that a good relationship still exists between the water spectra of SGL and the lake depth in the presence of 48 mg/L of SPM. The root mean squared error for the models ranged from 0.163 m (Partial Least Squares Regression—PLSR model) to 0.243 m (physically based model), which is consistent with the published literature. However, the SPM limited the maximum detectable depth to approximately 3 m. This maximum detectable depth was also confirmed by the theoretical concept of Philpot (1989). The maximum detectable depth decreases exponentially with an increase in the water attenuation coefficient g, which directly depends on the water properties.

NIR-based models for estimating selected physical and chemical wood properties from fast-growing plantations

As a faster, reliable, and low cost technique, applicable to large samplings, near infrared (NIR) spectroscopy technology has been widely applied for high-throughput phenotyping in forest breeding programmes. The aim of this study was to develop multivariate models for estimating the chemical and physical properties of juvenile wood based on NIR signatures of milled wood. Moreover, two approaches, namely, external validation by clone and by age, were tested to validate the model for estimating extractive content. NIR spectra of wood specimens taken from three clones of Eucalyptus urophylla (one to six years old) grown in southern Brazil were used to calibrate and validate models for predicting the wood basic density, total extractives, ash content, holocellulose content, syringyl to guaiacyl ratio (S/G) and elementary components of the wood. PLS-R models were validated by an independent set of wood specimens and presented promising statistics for the estimating wood density (R2p = 0.768), extractives (R2p = 0.912), ash (R2p = 0.936) and carbon (R2p = 0.697) contents from NIR signatures measured in the milled wood of young trees. Furthermore, NIR models for estimating the extractive content of wood were validated using the clones or ages left out of the training sets. Most models presented satisfactory statistics (R2 > 90%) and could be applied to routine laboratory analyses or to select potential trees in Eucalyptus breeding programmes.

Detection of small yellow croaker freshness by hyperspectral imaging

Rapid discrimination of fresh and repeatedly thawed small yellow croaker is of great significance for monitoring quality and ensuring consumer safety. In this study, we selected 160 fish samples of similar size and divided them equally into four groups (fresh, freeze-thaw once, freeze-thaw twice and freeze-thaw three times). Hyperspectral images were acquired from fresh and repeatedly frozen-thawed fish samples. The grey scale co-occurrence matrix (GLCM) was then applied to extract texture information from the first three principal component (PC) images, and a library for support vector machines (LIBSVM) were employed to discriminate fresh and repeatedly frozen-thawed fish samples using spectral characteristics, texture features and their fusion, respectively. The results indicated that LIBSVM model using the fused data showed the highest classification accuracy of 96.88%, and freshness degradation of fish samples after three freeze-thaw cycles was observed in the second PC image. In the freshness level for validation model, PLSR model achieved good performance withRV2= 0.90 and RPD= 3.17. The current findings demonstrate that hyperspectral imaging(HSI) is feasible for non-destructive determination of small yellow croaker freshness, providing technical guidance for the storage and marketing of aquatic products.

Classification and determination of sulfur content in crude oil samples by infrared spectrometry

Determining and classifying the sulfur content of crude oil has long been of great importance because of its adverse economic and environmental effects. In this study, the total sulfur concentration in crude oil samples was determined and classified using attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy and chemometric methods. The methods designed for the analysis of crude oils are rapid, economical and non-destructive in the production process of the petroleum industry. Two sets of 70 and 31 samples in regression models were considered for the calibration and prediction sets, respectively. The calibration models were developed using the partial least squares regression model (PLS-R) and support vector machine regression model (SVM-R). Different pre-processing methods were also evaluated for the development of models. The preprocessing methods based on baseline correction, standard normal variate (SNV) and the auto scale were selected for regression and classification models. The use of SVM-R as a non-linear regression provided a model with significantly better root mean square error of prediction (RMSEP) values than the PLS-R model as a linear model. The ATR-FTIR spectral data were also applied by supervised classification method using the partial least squares-discriminant analysis (PLS-DA) and support vector machine-discriminant analysis (SVM-DA) for classifying crude oils based on sulfur content. The samples were classified into two classes according to the sulfur content into sweet and sour crude oil. The result of the classification found an accuracy of 96% and a classification error of 0.0384 for the prediction set in the PLS-DA algorithm. The results indicated that ATR-FTIR spectroscopy associated with multivariate calibration and classification models is a rapid and reliable approach for parallel quantification and qualification of the sulfur content present in crude oils.

Estimating technological parameters and stem productivity of sugarcane treated with rock powder using a proximal spectroradiometer Vis-NIR-SWIR

This study aimed to evaluate the use of proximal Vis-NIR-SWIR spectroscopy to estimate technological parameters and tons of sugarcane per hectare (TSH) with the application of rock powder to the soil. It was carried out in an Arenosol area in Paranavaí , Brazil. The treatments were arranged within a split-plot system, designed in randomized blocks with four repetitions. For the experimental plots, inputs supplying calcium, magnesium, and sulfur were applied, whereas inputs supplying potassium were used for the subplots. The first and second cycles (sugarcane) were completed at 14 and 26 months, respectively, after the inputs application. During both growth cycles, technological parameters of the crop and of the TSH were determined. In addition, the stem spectrum of the crop was collected with a Vis-NIR-SWIR proximal spectroradiometer for later prediction of the parameters and TSH through the Partial Least Square Regression technique. It was possible to adjust models in the prediction phase with R 2 p > 0.50 and RPD p > 1.50 for all evaluated attributes, with emphasis on three variables, namely purity, reducing sugars, and brix, which had R 2 p and RPD p values above 0.86 and 2.75, respectively. The results show that proximal Vis-NIR-SWIR spectroscopy can be used to predict technological parameters and TSH of sugarcane with the application of rock powder. It offers advantages in relation to usual methods, since it is less time-consuming and low-cost, moreover it does not involve the use of toxic reagents. • The application of rock powders enabled a wide diversity of spectral curves. • Technological parameters and sugarcane yield can be quantified by Vis-NIR-SWIR spectroscopy. • The SWIR region was important for the construction of the PLSR model. • Vis-NIR-SWIR spectroscopy is presented as an alternative to the analyses of sugarcane.

Evaluation of groundwater quality for agricultural under different conditions using water quality indices, partial least squares regression models, and GIS approaches

Evaluating grouLindwater quality and associated hydrochemical properties is critical to manage groundwater resources in arid and semiarid environments. The current study examined groundwater quality and appropriateness for agriculture in the alluvial aquifer of Makkah Al-Mukarramah Province, Saudi Arabia, utilizing several irrigation water quality indices (IWQIs) such as irrigation water quality index (IWQI), total dissolved solids (TDS), sodium adsorption ratio (SAR), potential salinity (PS), magnesium hazard (MH), and residual sodium carbonate (RSC) assisted by multivariate modeling and GIS tools. One hundred fourteen groundwater wells were evaluated utilizing several physicochemical parameters, which indicating that the primary cation and anion concentrations were as follows: Na+ > Ca2+ > Mg2+ > K+, and Cl− > SO42˗ > HCO3˗ > NO3˗ > CO32˗, respectively, reflecting Ca–HCO3, Na–Cl, and mixed Ca–Mg–Cl–SO4 water facies under the stress of evaporation, saltwater intrusion, and reverse ion exchange processes. The IWQI, TDS, SAR, PS, MH, and RSC across two studied regions had mean values of 64.86, 2028.53, 4.98, 26.18, 38.70, and − 14.77, respectively. For example, the computed IWQI model indicated that approximately 31% of samples fell into the no restriction range, implying that salinity tolerance crops should be avoided, while approximately 33% of samples fell into the low to moderate restriction range, and approximately 36% of samples fell into the high to severe restriction range for irrigation, implying that moderate to high salt sensitivity crops should be irrigated in loose soil with no compacted layers. The partial least squares regression model (PLSR) produced a more accurate assessment of six IWQIs based on values of R2 and slope. In Val. datasets, the PLSR model generated strong estimates for six IWQIs with R2 varied from 0.72 to 1.00. There was a good slope value of the linear relationship between measured and predicted for each parameter and the highest slope value (1.00) was shown with RSC. In the PLSR models of six IWQIs, there were no overfitting or underfitting between the measuring, calibrating, and validating datasets. In conclusion, the combination of physicochemical characteristics, WQIs, PLSR, and GIS tools to assess groundwater suitability for irrigation and their regulating variables is beneficial and provides a clear picture of water quality.