Comparing the artificial neural network with parcial least squares for prediction of soil organic carbon and pH at different moisture content levels using visible and near-infrared spectroscopy

Tekin, Yücel; Tümsavaş, Zeynal; Mouazen, Abdul Mounem

doi:10.1590/s0100-06832014000600014

Cited by 13 publications

(8 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, Wijewardane et al [16] used an MIR spectral library to show superior model performance using ANN compared with PLSR for all soil properties (9 out of 12) except clay, silt and sand. In contrast, PLSR models were superior compared with ANN when predicting soil pH at soil moisture levels below 20% using a vis-NIR spectral library [43]. Similarly, a comparison between PLSR and Cubist using MIR data showed that Cubist outperformed the former in the prediction of total carbon and clay, but the prediction of cation exchange capacity was better using a PLSR model [38].…”

Section: Introductionmentioning

confidence: 96%

Accurate and Precise Prediction of Soil Properties from a Large Mid-Infrared Spectral Library

et al. 2019

View full text Add to dashboard Cite

Diffuse reflectance spectroscopy (DRS) is emerging as a rapid and cost-effective alternative to routine laboratory analysis for many soil properties. However, it has primarily been applied in project-specific contexts. Here, we provide an assessment of DRS spectroscopy at the scale of the continental United States by utilizing the large (n > 50,000) USDA National Soil Survey Center mid-infrared spectral library and associated soil characterization database. We tested and optimized several advanced statistical approaches for providing routine predictions of numerous soil properties relevant to studying carbon cycling. On independent validation sets, the machine learning algorithms Cubist and memory-based learner (MBL) both outperformed random forest (RF) and partial least squares regressions (PLSR) and produced excellent overall models with a mean R2 of 0.92 (mean ratio of performance to deviation = 6.5) across all 10 soil properties. We found that the use of root-mean-square error (RMSE) was misleading for understanding the actual uncertainty about any particular prediction; therefore, we developed routines to assess the prediction uncertainty for all models except Cubist. The MBL models produced much more precise predictions compared with global PLSR and RF. Finally, we present several techniques that can be used to flag predictions of new samples that may not be reliable because their spectra fall outside of the calibration set.

show abstract

Section: Introductionmentioning

confidence: 96%

Accurate and Precise Prediction of Soil Properties from a Large Mid-Infrared Spectral Library

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Our analysis pipeline incorporates 4 multivariate regression models and several pre-processing techniques. We tested two calibration models already applied to pears spectra for SSC quantification: Partial Least Squares (PLS) and Multiple Linear Regression (MLR) [11,12,13,14,15,24], another model that is gaining traction in Chemometrics and that has been applied in similar contexts, Support Vector Machines (SVM) [25,26] and a simple Multi Layer Perceptron (MLP) neural network for comparison purposes [27,28,29]. The numerical analysis of the data was done using Python 3.6.…”

Section: Methodsmentioning

confidence: 99%

Non-Destructive Soluble Solids Content Determination for ‘Rocha’ Pear Based on VIS-SWNIR Spectroscopy under ‘Real World’ Sorting Facility Conditions

Passos

Rodrigues

Cavaco

et al. 2019

Sensors

View full text Add to dashboard Cite

In this paper we report a method to determine the soluble solids content (SSC) of ‘Rocha’ pear (Pyrus communis L. cv. Rocha) based on their short-wave NIR reflectance spectra (500–1100 nm) measured in conditions similar to those found in packinghouse fruit sorting facilities. We obtained 3300 reflectance spectra from pears acquired from different lots, producers and with diverse storage times and ripening stages. The macroscopic properties of the pears, such as size, temperature and SSC were measured under controlled laboratory conditions. For the spectral analysis, we implemented a computational pipeline that incorporates multiple pre-processing techniques including a feature selection procedure, various multivariate regression models and three different validation strategies. This benchmark allowed us to find the best model/preproccesing procedure for SSC prediction from our data. From the several calibration models tested, we have found that Support Vector Machines provides the best predictions metrics with an RMSEP of around 0.82 ∘Brix and 1.09 ∘Brix for internal and external validation strategies respectively. The latter validation was implemented to assess the prediction accuracy of this calibration method under more ‘real world-like’ conditions. We also show that incorporating information about the fruit temperature and size to the calibration models improves SSC predictability. Our results indicate that the methodology presented here could be implemented in existing packinghouse facilities for single fruit SSC characterization.

show abstract

“…The high sensitivity of vis-NIR spectroscopy has been shown to be capable to detect trace minerals such as arsenic contamination in soil [ 24 ]. However, because of the strong moisture absorption peaks in the vicinity of the SOM features within the vis-NIR spectral range, the presence of moisture in the soil interferes the interpretation of SOM content significantly [ 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 ]. It has been reported that, the two strong moisture absorption peaks that are located at about 1470 nm and 1900 nm, and the relatively weaker water absorptions features at 600 nm, 738 nm and 836 nm [ 33 , 34 , 35 ], have induced spectral distortions of the SOM absorption features particularly in the vis-NIR spectral region at about 1400 nm and 1900 nm [ 25 , 36 , 37 , 38 , 39 ].…”

Section: Introductionmentioning

confidence: 99%

“…It has been reported that, the two strong moisture absorption peaks that are located at about 1470 nm and 1900 nm, and the relatively weaker water absorptions features at 600 nm, 738 nm and 836 nm [ 33 , 34 , 35 ], have induced spectral distortions of the SOM absorption features particularly in the vis-NIR spectral region at about 1400 nm and 1900 nm [ 25 , 36 , 37 , 38 , 39 ]. The distortion of the SOM spectral features induced significant errors in the assessment of the SOM content from the moist soil using spectroscopy methods [ 27 , 28 , 29 , 30 , 31 , 32 , 40 ].…”

Section: Introductionmentioning

confidence: 99%

“…All these methods, such as the EPO have been shown to improve the prediction of soil organic carbon (SOC) substantially with an increase of the coefficient of determination ( R 2 ) from almost zero when no moisture correction is made, to over 0.50 after the correction is applied [ 26 ]. Amongst these studies [ 27 , 28 , 29 , 30 , 31 , 32 , 40 ] the best coefficient of determination ( R 2 ) and the ratio of prediction to deviation (RPD) for assessing the SOM in moist soil (<35% w / w moisture) were about 0.86–0.88 [ 27 ] and 2.66 [ 29 ] respectively. Note that the assessments of SOM from relatively dry soil have achieved much better result with R 2 and RPD of 0.892 ± 0.004 and 3.053 ± 0.056 respectively [ 15 ].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Prediction of the Soil Organic Matter (SOM) Content from Moist Soil Using Synchronous Two-Dimensional Correlation Spectroscopy (2D-COS) Analysis

Wang

Zheng

et al. 2020

Sensors

View full text Add to dashboard Cite

This paper illustrates a simple yet effective spectroscopic technique for the prediction of soil organic matter (SOM) from moist soil through the synchronous 2D correlation spectroscopy (2D-COS) analysis. In the moist soil system, the strong overlap between the water absorption peaks and the SOM characteristic features in the visible-near infrared (Vis-NIR) spectral region have long been recognised as one of the main factors that causes significant errors in the prediction of the SOM content. The aim of the paper is to illustrate how the tangling effects due to the moisture and the SOM can be unveiled under 2D-COS through a sequential correlogram analysis of the two perturbation variables (i.e., the moisture and the SOM) independently. The main outcome from the 2D-COS analysis is the discovery of SOM-related bands at the 597 nm, 1646 nm and 2138 nm, together with the predominant water absorbance feature at the 1934 nm and the relatively less important ones at 1447 nm and 2210 nm. This information is then utilised to build partial least square regression (PLSR) models for the prediction of the SOM content. The experiment has shown that by discarding noisy bands adjacent to the SOM features, and the removal of the water absorption bands, the determination coefficient of prediction (Rp2) and the ratio of prediction to deviation (RPD) for the prediction of SOM from moist soil have achieved Rp2 = 0.92 and the RPD = 3.19, both of which are about 5% better than that of using all bands for building the PLSR model. The very high RPD (=3.19) obtained in this study may suggest that the 2D-COS technique is effective for the analysis of complex system like the prediction of SOM from moist soil.

show abstract

Comparing the artificial neural network with parcial least squares for prediction of soil organic carbon and pH at different moisture content levels using visible and near-infrared spectroscopy

Cited by 13 publications

References 32 publications

Accurate and Precise Prediction of Soil Properties from a Large Mid-Infrared Spectral Library

Accurate and Precise Prediction of Soil Properties from a Large Mid-Infrared Spectral Library

Non-Destructive Soluble Solids Content Determination for ‘Rocha’ Pear Based on VIS-SWNIR Spectroscopy under ‘Real World’ Sorting Facility Conditions

Prediction of the Soil Organic Matter (SOM) Content from Moist Soil Using Synchronous Two-Dimensional Correlation Spectroscopy (2D-COS) Analysis

Contact Info

Product

Resources

About