NIR Spectra Research Articles

Electrochromic device with hierarchical metal mesh electrodes: Transmittance switching in the full spectral range of solar radiation

Hierarchical metal mesh structures are fabricated for use as transparent conductive electrodes in electrochromic devices with large-band transmittance switching. Electrochromic materials (WO3 and Ta:NiO) are deposited by plasma sputtering on the metal mesh and characterized in a custom-made electrochemical setup allowing in-situ photo-spectrometer measurement. Both films deposited on metal mesh exhibit high transmittance, not only in the visible but also in the NIR spectrum. Electrochromic devices based on the WO3 cathode and Ta:NiO anode combined with the hierarchical metal mesh are fabricated. The resulting device achieves remarkable optical modulation in the full spectral range of solar radiation compared to conventional ITO-based devices. This study paves the way for development of hierarchical metal mesh-based electrochromic devices which offer high potential for energy savings through wide band modulation of solar heat gain in buildings in countries with warm and cold seasons.

Near room temperature tunability of thermochromic VO2 thin films prepared via thermal oxidation method, influence of CO2:N2 gas flux ratios

In this study, we have investigated the thermochromic characterizations of VO2 thin films synthesized by the thermal oxidation method. The oxidation process of a DC sputtered metallic vanadium layer on glass substrates, at 450 °C for 1 h, was carried out in the presence of CO2:N2 gases with different flux ratios of 30:70, 40:60, 50:50, 60:40, and 70:30, respectively. Using CO2, as the oxidizing gas, provides an easy control route for obtaining the VO2 phase among various vanadium oxide phases. The layers were characterized by FESEM, XRD & Raman spectra, sheet resistance vs. temperature (20–120 °C), and UV–Vis–NIR spectra at 25 and 90 °C. The XRD and Raman spectra confirmed all prepared layers have a VO2 polycrystalline structure in monoclinic phase. We found among the studied samples CN30-70 and CN50-50 having desirable optical characteristics of peak visible transmittance of about 55%, with low transition temperatures (Tcr) of ∼42 and 31 °C, and also relatively high amounts of ΔT1700nm, ΔTsol and Tlum,av are good candidates for thermochromic smart windows, both from optical properties and economical fabrication method points of view.

Augmenting NIR Spectra in deep regression to improve calibration

Deep learning, particularly with convolutional neural networks, shows promise in modelling near-infrared spectroscopy (NIRS), but the lack of robust generalisation across instruments often affects performance in practice. Here, we investigate a method to increase the robustness of this approach. The proposed method involves using a simple data augmentation technique during the training process. The performance of convolutional neural network regression is compared to partial least squares regression (PLSR) using kiwifruit data collected from multiple handheld devices over three seasons and mango data collected from a single device over four seasons. The results suggest that data augmentation for NIR spectra can prevent overfitting. In particular, augmenting the training data to mimic spectra collected over multiple devices results in a neural network model with improved performance over PLSR.

Mid-Level Data Fusion Combined with the Fingerprint Region for Classification DON Levels Defect of Fusarium Head Blight Wheat.

In this study, a method of mid-level data fusion with the fingerprint region was proposed, which was combined with the characteristic wavelengths that contain fingerprint information in NIR and FT-MIR spectra to detect the DON level in FHB wheat during wheat processing. NIR and FT-MIR raw spectroscopy data on normal wheat and FHB wheat were obtained in the experiment. MSC was used for pretreatment, and characteristic wavelengths were extracted by CARS, MGS and XLW. The variables that can effectively reflect fingerprint information were retained to build the mid-level data fusion matrix. LS-SVM and PLS-DA were applied to investigate the performance of the single spectroscopic model, mid-level data fusion model and mid-level data fusion with fingerprint information model, respectively. The experimental results show that mid-level data fusion with a fingerprint information strategy based on fused NIR and FT-MIR spectra represents an effective method for the classification of DON levels in FHB wheat samples.

Soluble Biobased Substances in soil or Salicylic Acid on leaves affect the foliar pH and soil biovariability of Grapes-as explained by the NIR Spectroscopy of Litterbags and Teabags

The purpose of the trial was to check the effects of two grapevine treatments on the plant activity and on the bioactivity and biovariability of the soil. An alkaline complex of Soluble Biobased Substances (SBS) was used in soil at 30 g per plant in a single solution. Salicylic Acid (SA) was used on leaves at 150 mg l-1 every two weeks at 50 ml plant. The plants were examined for their foliar pH and NIR spectra. The soil bioactivity was monitored by means of hay-Litterbag-NIRS (LBN) in combination with the Teabag Index (TBI), using rooibos and green tea that had been buried for 60 days. The evolution of the TBI presented here concerns the TBI-NIRS spectroscopic method used for discriminant analysis. A new algorithm was used to estimate the soil microbiome from the green Teabag spectra. The obtained results showed that the plants and the soil responded to the treatments. In fact, SBS, but not SA, lowered the leaf pH by 5%, an unexpected and original result. Both treatments increased the variability of the leaf composition, with a lower discrimination, based on the NIR spectra, from the Control (75%) to 44% (SA) and 38% (SBS). The TBI method, which is based on weights, was less efficient (67%) than the TBI-NIRS of rooibos (96%) or the LBN of hay (80%), but it was like the TBI-NIRS of green tea (74%). The LBN analyses indicated that the mycorrhizal index had increased by 8% in SA but had reduced by 7% in SBS, while both treatments reduced the activity of the microbes, which did not affect the soil respiration rate. The mineral N in the soil was substantially raised by about 11÷69%. The Taxa profiles showed marked deviations from the Control. Moreover, the SBS treatment reduced the Glomeromycota by 35%, which matched to the reduction in the mycorrhizal index. The most favored Bacteria from the treatments were Proteobateriaand Actinobacteria, with Mortierellomycota being the most penalized. Neither treatment affected the production, but both delayed the technological maturity by 9-11%, while the SBS retarded the phenolic maturity by about 18%. It has been concluded that a simple treatment of vines can affect the bioactivity in the leaves and berries as well as the biovariability of the soil.

Optimised use of data fusion and memory‐based learning with an Austrian soil library for predictions with infrared data

AbstractInfrared spectroscopy in the visible to near‐infrared (vis–NIR) and mid‐infrared (MIR) regions is a well‐established approach for the prediction of soil properties. Different data fusion and training approaches exist, and the optimal procedures are yet undefined and may depend on the heterogeneity present in the set and on the considered scale. The objectives were to test the usefulness of partial least squares regressions (PLSRs) for soil organic carbon (SOC), total carbon (Ct), total nitrogen (Nt) and pH using vis–NIR and MIR spectroscopy for an independent validation after standard calibration (use of a general PLSR model) or using memory‐based learning (MBL) with and without spiking for a national spectral database. Data fusion approaches were simple concatenation of spectra, outer product analysis (OPA) and model averaging. In total, 481 soils from an Austrian forest soil archive were measured in the vis–NIR and MIR regions, and regressions were calculated. Fivefold calibration‐validation approaches were carried out with a region‐related split of spectra to implement independent validations with n ranging from 47 to 99 soils in different folds. MIR predictions were generally superior over vis–NIR predictions. For all properties, optimal predictions were obtained with data fusion, with OPA and spectra concatenation outperforming model averaging. The greatest robustness of performance was found for OPA and MBL with spiking with R2 ≥ 0.77 (N), 0.85 (SOC), 0.86 (pH) and 0.88 (Ct) in the validations of all folds. Overall, the results indicate that the combination of OPA for vis–NIR and MIR spectra with MBL and spiking has a high potential to accurately estimate properties when using large‐scale soil spectral libraries as reference data. However, the reduction of cost‐effectiveness using two spectrometers needs to be weighed against the potential increase in accuracy compared to a single MIR spectroscopy approach.

Evaluation of near-infrared spectroscopy as a contactless method for health monitoring of resin-based coating materials applied to concrete surfaces.

The surfaces of concrete structures are often coated with protective materials to minimize corrosion and weathering-based deterioration. Therefore, it is important to monitor the aging of the coating materials and their overall condition to extend the service lifetime of the structure effectively. Near-infrared spectroscopy (NIRS) is a contactless, nondestructive, rapid, and convenient method for material characterization; therefore, it is useful for onsite inspection of coating materials. Hence, in this study, we attempt to determine whether NIRS can be used for simple inspection for health monitoring of organic resin-based coating materials. In addition to identifying different severities of peeling damage, we characterize the ultraviolet-induced deterioration of coating materials with different thicknesses using diffuse reflection spectra acquired in the near-infrared wavelength region. For independent comparison with the NIR spectra, the state of the coating materials on the mortar specimens was analyzed using a combination of Fourier-transform infrared spectroscopy and scanning electron microscopy, while the state of the underlying mortar specimens was analyzed using permeability and salt-water immersion tests. The results confirm that the NIRS could detect the degradation of coating materials at early stages of deterioration before their permeability had been affected. NIRS offers the possibility of intermittent monitoring of coating deterioration. In addition, because the NIR spectrometer is portable, it can help in inspecting high-rise areas and areas that are difficult to reach. Therefore, we believe that NIRS is a simple, safe, and inexpensive method for inspection of surface coating materials.

Prediction of fat content in salmon fillets based on hyperspectral imaging and residual attention convolution neural network

The fat content of salmon is an essential indicator to evaluate its quality, and is related to its commercial value. An attention residual convolutional neural network (CNN), namely RACNN, which added channel attention and residual modules to the traditional CNN structure, was proposed to combine near-infrared hyperspectral imaging (NIR-HSI) (900-1700 nm) to predict fat contents in salmon fillets. Partial least squares regression (PLSR), least squares support vector machine (LSSVM) and CNN models were compared with the proposed RACNN models. Various spectrum preprocessing methods were used at the time. And Three feature extraction algorithms (successive projections algorithm (SPA), uninformative variable elimination (UVE), and competitive adaptive reweighted sampling (CARS)) were used to extract the effective wavelengths from the preprocessed NIR spectra. The experimental results showed that, the optimal PLSR and LSSVM models obtained coefficient of determination of prediction (Rp2) values of 0.8781 and 0.8952, respectively. The RACNN model based on full spectrum gained the highest prediction accuracies with Rp2 = 0.9033, root mean square error of prediction (RMSEP) = 1.5143, and residual predictive deviation (RPD) = 3.2707. Therefore, the proposed RACNN model enables the rapid, non-destructive and accurate prediction of the fat content in salmon fillets without additional processing on raw spectra.

Fabrication, structural, thermo-mechanical and opto-electrical behavior of anthranilamide compound

Anthranilamide (AAM) compound were fabricated via an alcoholic solvent assisted slow evaporation at room temperature for investigating the multi functional device applications. The solubility test was performed with different concentrations of AAM salt dispersed in 100 mL ethanol solvent to understand the crystallization process of fabricated material. The cell parameters, and diffraction peaks of the title compound were collected through nondestructive tool like single crystal XRD and powder XRD studies, and thus results showed the monoclinic structure. The fabricated material’s functional group, thermal stability, and hardness behavior were tested by various tool viz FT-IR, TG-DTA, and Vickers hardness analysis. With the aid of UV–vis–NIR spectrum, and frequency conversion test technique, optical transmittance and double harmonic generation efficiency values are detected to be 92% and 4.1 mV. Using electrical studies, the dielectric parameters and specific conductance of the title compound values are varied by applying a low frequency to a high frequency region at different temperatures (313, 323, and 333 K).

Third-order nonlinearity and functional data analysis of ADP-doped PVA/PVP polymer composites

The structural dependence of nonlinear optical properties associated with ADP-doped PVA/PVP polymer composites is discussed in this work. Polymer composites are fabricated using the solution casting method. X-ray diffraction (XRD) was used to perform structural analysis on these samples whose average crystallite size increases with an increase in dopant concentration. Results from UV–Vis–NIR spectra show that the samples are transparent in the IR region. The Z-scan technique was established by following the method described by Sheik-Bahae et al. The shape of the open aperture plot suggests that the samples exhibit two-photon absorption. The NLO properties such as nonlinear absorption coefficient ([Formula: see text], two-photon absorption cross-section ([Formula: see text] and the imaginary part of third-order nonlinear susceptibility [Formula: see text] were investigated at 1064 nm and these values were found to be in the order of [Formula: see text][Formula: see text]W/cm, [Formula: see text][Formula: see text]cm4/GW and [Formula: see text][Formula: see text]esu, respectively. The open aperture Z-scan method yielded optical limiting data, from which the optical limiting threshold was calculated to be in the order of [Formula: see text][Formula: see text]W/m2. Functional data analysis uses a suitable basis representation to assess the functional correlation between global and local variables.

Defect-induced Sr1−xPrxTiO3 crystallites by burial sintering and its optoelectronic applications

The band structure of SrTiO3 was altered by introducing Pr3+ in the Sr2+ site and subsequent graphite burial sintering. The asymmetric nature of the (310) peak in XRD and the emergence of Eg and B1g modes in the Raman spectrum suggest the transformation from cubic to tetragonal phase for the Sr1-xPrxTiO3 system with x ≥ 0.1. The development of the forbidden first-order Raman peaks and novel spectral properties are attributed to the local vibrational modes linked to oxygen vacancies. Structural variations in terms of lattice strain were investigated using a W–H plot, and morphological changes were analyzed by high-resolution FE-SEM. High infrared absorption in UV–Vis–NIR spectra and a significant reduction in band gap from 2.8 eV to 1.8 eV were accomplished through increased disorder in the lattice and the reduction of Ti4+. The blue shift in absorption spectra with Pr concentration can be used to tune the absorption range for sensing purposes. The formation of new electronic states observed here offers a novel approach for band gap modification in perovskites for photocatalytic and photovoltaic applications.

Profitable approach for the synthesis of Mn3O4 phase materials and its interesting pigmentation applications

A simple precipitation synthesis of Mn3O4 phase with average crystalline size of 27.8 nm in the form of as-precipitated product crystallized in tetragonal symmetry, space group, I41/amd (141) with the lattice parameters of a = 0.5761(3) nm and c = 0.9451(6) nm. The rod-shaped (diameter of ∼ 400 nm) morphology with a spherical particle size of ∼ 50 nm is observed for the as-precipitated product. A weight loss of ∼ 4% is noted for the as-precipitated product powder up to 1000 °C with neither exothermic nor endothermic peaks in the thermal profiles. The UV–Vis absorbance spectrum reveals an absorption edge at 550 nm, which is due to optical band gap absorption with a calculated band gap (Eg) of 1.95 eV from the Tauc plot. The NIR (750 – 2500 nm) spectrum shows the NIR reflectance of 27–66 % and 50–69 % in the solar and color pigment regions, respectively. As a result, the present obtained Mn3O4 powder can be used in the color pigmentation industry. The Raman reveals a dominant peak at ∼ 652 cm−1 and two minor peaks at 315 and 369 cm−1 are due to A1g, Eg, and T2g Raman active modes of tetragonal Mn3O4 phase materials.

A Vis/NIR spectra-based approach for identifying bananas infected with Colletotrichum musae.

Anthracnose of banana caused by Colletotrichum species is one of the most serious post-harvest diseases, which can cause significant yield losses. Clarifying the infection mechanism of the fungi using non-destructive methods is crucial for timely discriminating infected bananas and taking preventive and control measures. This study presented an approach for tracking growth and identifying different infection stages of the C. musae in bananas using Vis/NIR spectroscopy. A total of 330 banana reflectance spectra were collected over ten consecutive days after inoculation, with a sampling rate of 24 h. The four-class and five-class discriminant patterns were designed to examine the capability of NIR spectra in discriminating bananas infected at different levels (control, acceptable, moldy, and highly moldy), and different time at early stage (control and days 1-4). Three traditional feature extraction methods, i.e. PC loading coefficient (PCA), competitive adaptive reweighted sampling (CARS) and successive projections algorithm (SPA), combining with two machine learning methods, i.e. partial least squares discriminant analysis (PLSDA) and support vector machine (SVM), were employed to build discriminant models. One-dimensional convolutional neural network (1D-CNN) without manually extracted feature parameters was also introduced for comparison. The PCA-SVM and·SPA-SVM models had good performance with identification accuracies of 93.98% and 91.57%, 94.47% and 89.47% in validation sets for the four- and five-class patterns, respectively. While the 1D-CNN models performed the best, achieving an accuracy of 95.18% and 97.37% for identifying infected bananas at different levels and time, respectively. These results indicate the feasibility of identifying banana fruit infected with C. musae using Vis/NIR spectra, and the resolution can be accurate to one day.

Effect of Sm3+ ions doping on the structural and optical properties, Judd-Ofelt and radiative parameters of ZnS phosphor materials

Undoped ZnS and Sm3+ doped with varying concentration of ZnS phosphor materials were synthesized via low temperature solid state reaction method. The samples were investigated for the influence of different concentrations of Sm3+ ions on their structural and spectroscopic properties. The presence of various absorption bands in UV–Vis–NIR spectra revealed the transition from ground state to different excited states of Sm3+ ions. The Judd-Ofelt (JO) analyses were evaluated to identify the structure and the bonding around the Sm3+ ions. The high value of Ω2 intensity parameter reveals that the site symmetry is lower. The parameters such as radiative transition probabilities, radiative lifetime, branching ratio, stimulated emission cross-sections and the optical gain parameters are also evaluated. The PL spectra shows five high intensity emission bands centered at 567 nm, 590 nm, 619 nm, 641 nm and 667 nm corresponding to 4G5/2→6Hj (j = 5/2, 7/2,9/2, 11/2, 13/2) transitions and showed optimum for concentration at x = 0.03 in Zn(1-x)S. The quantum yield of the prepared optimal phosphor using integrating sphere method was found to be 10.08%. This indicates that the materials synthesized may be favorable for the applications of laser and other optical devices.

Rapid screening of benzyl dodecyl dimethyl ammonium bromide co-metabolic degrading bacteria based on NIR hyperspectral imaging technology

As a typical disinfectant, the use of benzyl dodecyl dimethyl ammonium bromide (BDAB) has dramatically increased since the emergence of SARS-CoV-2, posing a threat to environmental balance and human health. Screening BDAB co-metabolic degrading bacteria is required for efficient microbial degradation. Conventional methods for screening co-metabolic degrading bacteria are laborious and time-consuming, especially when the number of strains is large. This study aimed to develop a novel method for the rapid screening of BDAB co-metabolic degrading bacteria from the cultured solid medium using near-infrared hyperspectral imaging (NIR-HSI) technology. Based on NIR spectra, the concentration of BDAB in the solid medium can be well predicted by partial least squares regression (PLSR) models, non-destructively and rapidly, with Rc2 > 0.872 and Rcv2 > 0.870. The results show that the predicted BDAB concentrations decrease after degrading bacteria utilization, comparing with the regions where no degrading bacteria grew. The proposed method was applied to directly identify the BDAB co-metabolic degrading bacteria cultured on the solid medium, and two kinds of co-metabolic degrading bacteria RQR-1 and BDAB-1 were correctly identified. This method provides a high-efficiency method for screening BDAB co-metabolic degrading bacteria from a large number of bacteria.

Synthesis, crystal growth, electronic structures and optical properties of wide bandgap semiconductor (CH3NH3)2HgI4

The synthesis, crystal growth and electronic structures, as well as optical properties of Hg-based hybrid organic-inorganic compound (CH3NH3)2HgI4 are reported. Single crystal was grown through the anti-solvent method, yielding the faint yellow (CH3NH3)2HgI4 flake crystals. The structural analysis revealed that (CH3NH3)2HgI4 crystallized in the orthorhombic space group Pbca with a 0D crystal structure featuring anionic [HgI4]2- units that were surrounded by CH3NH3+ cations. Furthermore, (CH3NH3)2HgI4 does not decompose when being exposed to air, but the crystallinity slowly decays over time in air at room temperature. TGA and DSC show that it is stable for (CH3NH3)2HgI4 from room temperature to 223 °C. It shows that (CH3NH3)2HgI4 is a slightly indirect nature with 3.276 eV of band gap based on density functional theory calculations, which is close to 3.18 eV derived from UV–vis–NIR spectra of (CH3NH3)2HgI4 crystal. In addition, room temperature photoluminescence (PL) spectra of (CH3NH3)2HgI4 yields broad peaks at 540 nm with full width at half maximum (FWHM) values of 201 nm. These PL peaks are tentatively assigned to self-trapped excitons following the crystal and electronic structure considerations. The bands around the Fermi level have small dispersions, which is indicative of high charge localization with significant exciton binding energies in (CH3NH3)2HgI4. PL analysis indicates that (CH3NH3)2HgI4 may have potential research value as white-light-emitting phosphors and scintillator.

Detection of TiO and VO in the Atmosphere of WASP-121b and Evidence For its Temporal Variation

We report the transit observations of the ultra-hot Jupiter WASP-121b using the Goodman High Throughput Spectrograph at the 4 m ground-based Southern Astrophysical Research Telescope, covering the wavelength range 502–900 nm. By dividing the target and reference star into 19 spectroscopic passbands and applying differential spectrophotometry, we derive spectroscopic transit light curves and fit them using a Gaussian process framework to determine transit depths for every passband. The obtained optical transmission spectrum shows a steep increased slope toward the blue wavelength, which seems to be too steep to be accounted for by Rayleigh scattering alone. We note that the transmission spectrum from this work and other works differ obviously from each other, which was pointed out previously by Wilson et al. as evidence for temporal atmospheric variation. We perform a free chemistry retrieval analysis on the optical transmission spectra from this work and the literature HST/WFC3 NIR spectrum. We determine TiO, VO and H2O with abundances of dex, dex and dex, respectively. We compare the abundances of all three of these molecules derived from this work and previous works, and find that they are not consistent with each other, indicating the chemical compositions of the terminator region may change over long timescales. Future multi-epoch and high-precision transit observations are required to further confirm this phenomenon. We note that when combining the transmission spectra in the optical and in NIR in retrieval analysis, the abundances of V and VO, the NIR-to-optical offset and the cloud deck pressure may be coupled with each other.

Y(HSeO3)(SeO3)(H2O)·(H2O) and Y2(SeO3)2(SeO4)(H2O)2·(H2O)0.75: Two yttrium selenites with a short UV cut-off edge explored from pure selenite compounds

Two UV inorganic yttrium selenite optical materials, namely, Y(HSeO3)(SeO3)(H2O)·(H2O) (1) and Y2(SeO3)2(SeO4)(H2O)2·(H2O)0.75 (2), have been synthesized successfully by hydrothermal reactions. Of these, Y2(SeO3)2(SeO4)(H2O)2·(H2O)0.75 (2) is a new mixed valent selenium oxide containing Se(IV) and Se(VI) cations simultaneously. Y(HSeO3)(SeO3)(H2O)·(H2O) (1) shows a 3D structure composed of 2D yttrium selenite layers linked by O−H···O bonds. Y2(SeO3)2(SeO4)(H2O)2·(H2O)0.75 (2) exhibits a 3D structure composed of 2D yttrium selenite layers connected by SeO4 tetrahedron. Y(HSeO3)(SeO3)(H2O)·(H2O) (1) crystallized in a chiral space group P212121 and could display a moderate SHG intensity about 0.7 times of the commercial KH2PO4 (KDP). Its powder laser damage threshold quantity was estimated to be 120.2 MW/cm2, which is about 46.2 times that of AGS (2.6 MW/cm2). UV–vis–NIR spectra reflectance spectra showed that these solids are wide band gap compounds with Eg of 5.5 and 5.1 eV respectively. Theoretical calculations disclosed that Y(HSeO3)(SeO3)(H2O)·(H2O) (1) is an indirect band gap structures and its bandgaps is determined by O and Se atoms.

Fractional-order Savitzky–Golay filter for pre-treatment of on-line vis–NIR spectra to predict phosphorus in soil

Visible and near infrared spectroscopy (vis-NIRS) has shown potential to predict soil phosphorous (P) with reasonable accuracy. However, spectra pre-treatment is essential in chemometric modelling. One of the most popular algorithms for spectral smoothing and differentiation is the integer-order Savitzky–Golay filter (IOSGF), which operates on a localised linear regression of several neighbouring points over a moving window. Herein, a modified Riemann–Liouville (RL) fractional-order Savitzky–Golay filter (MRLFOSGF) is presented based on the RL operators, as an extension of the conventional IOSGF. This filter was quantitatively analysed using power functions and Gaussian-type bands and were subsequently used to establish a partial least squares regression (PLSR) model for predicting P in soil. The results revealed that the MRLFOSGF offers significant flexibility with transition dynamics between integer-order SG derivatives and reduces baseline offsets and tilts. The PLSR model using the MRLFOSGF had a higher prediction accuracy than the corresponding PLSR model using the conventional IOSGF. This work demonstrates that the MRLFOSGF offers the advantage of wider applicability and better performance for predicting soil P than the conventional IOSGF.

Monitoring of salinity of water on the THA CHIN River basin using portable Vis-NIR spectrometer combined with machine learning algorithms

The goal of this work is to study the alternative practices for monitoring the salinity of water using a combination of portable vis-NIR spectrometer and machine learning approachs. Along 80 km of the Tha Chin River basin from the Gulf of Thailand, the data of salinity and NIR spectrum were collected during winter and summer seasons of Thailand. Salinity of water samples was measured by using a handheld electrical conductivity meter and NIR spectra was recorded with portable FQA-NIR GUN in the wavelength range of 600 to 1100 nm. The 10 machine learning models including partial least square regression (PLS), support vector machine (SVR), decision tree (DT), random forest (RF), adaptive boosting (AB), gradient boosting (GB), bagging meta-estimator (BME), extremely randomized trees (ERT), backpropagation neural networks (BPNN) and hybrid principal component analysis-neural network (PCNN) were applied to train the NIRs models for predicting salinity. All machine learning algorithms showed good prediction results which Rp2 values were higher than 0.84. The models built by tree-based algorithms (DT, RF, AB, GB, BME and ERT) displayed higher performances of calibration set and prediction set than those of PLS, SVM, BPNN and PCNN. Among these, the ERT algorithm showed the best performance Rp2 of 0.97, RMSEP of 0.41 g/L and RPD of 6.00. It was shown that NIR spectroscopy coupled with machine learning could be an alternative simpler way for predicting salinity of water.