Comparison of statistical regression and data-mining techniques in estimating soil water retention of tropical delta soils

Nguyen, Phuong Minh; Haghverdi, Amir; Pue, Jan De; Botula, Yves-Dady; Le, Khoa; Waegeman, Willem; Cornelis, Wim

doi:10.1016/j.biosystemseng.2016.10.013

Cited by 44 publications

(37 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Therefore, the large negative correlation of sand with q pF2 , q pF3 , and q pF4.2 is in agreement with studies done by Salter and Williams (1965) and da Costa et al (2013). Silt and clay have, as expected, a strong positive correlation with the lower matric potentials (Nguyen et al, 2017).…”

Section: Discussionsupporting

confidence: 91%

Evaluation of Parametric and Nonparametric Machine‐Learning Techniques for Prediction of Saturated and Near‐Saturated Hydraulic Conductivity

Kotlar

Iversen²,

Lier

2019

Vadose Zone Journal

View full text Add to dashboard Cite

Core Ideas Accurate Ks and K10 were obtained using water content data at several matric potentials. Robust Ks and K10 prediction by machine‐learning methods confirmed by bootstrapping. Gaussian process regression predicted Ks and K10 with minimum number of predictors. Parametric and nonparametric supervised machine learning techniques were used to estimate saturated and near‐saturated hydraulic conductivities (Ks and K10, respectively) from easily measurable soil properties including the name of the pedological horizon (HOR), soil texture (sand, silt, and clay), organic matter (OM), bulk density (BD), and water contents (θpF1, θpF2, θpF3, and θpF4.2) measured at four different matric heads (−10, −100, −1000, and −15,848 cm, respectively). Using a stepwise linear model (SWLM) and the Lasso regression as parametric methods with 316 data in training and 135 data in the testing phase, four pedotransfer functions (PTFs) were obtained in which water contents for both methods play an important role compared with other variables. The SWLM showed better performance than Lasso in the testing phase for log(Ks) and log(K10) prediction, with RMSE values of 0.666 and 0.551 cm d−1 and R2 of 0.26 and 0.65. Nonparametric supervised machine learning methods trained and tested with a similar data set significantly improved the accuracy of Ks prediction, with R2 of 0.52, 0.36, and 0.53 for Gaussian process regression (GPR), support vector machine (SVM), and ensemble (ENS) methods in the testing stage. These methods also described 74.9, 66.7, and 72.5% of the variation of log(K10). Bootstrapping validated the strong performance of nonparametric techniques. The feature selection capability of GPR determined that instead of using a model with all predictors, HOR, silt, θpF1, and θpF3 are sufficient for the prediction of log(Ks), while HOR, silt, and OM can predict log(K10) as accurate as the comprehensive model with all variables.

show abstract

Section: Discussionsupporting

confidence: 91%

Evaluation of Parametric and Nonparametric Machine‐Learning Techniques for Prediction of Saturated and Near‐Saturated Hydraulic Conductivity

Kotlar

Iversen²,

Lier

2019

Vadose Zone Journal

View full text Add to dashboard Cite

show abstract

“…The analysis of the results also showed that the PTFs selected in the literature and not using French soils performed less well than the ANN‐PTFs, as already reported in the literature (Nguyen et al ., ) (Tables and ). The RMSE and R 2 were, respectively, greater and smaller than with the ANN‐PTFs, using the PTFs selected in the literature and not using French soils.…”

Section: Resultsmentioning

confidence: 99%

“…PTFs have been applied in a large number of studies in recent decades (e.g. Al Majou et al ., ; Haghverdi et al ., ; Khlosi et al ., ; Nguyen et al ., ).…”

Section: Introductionmentioning

confidence: 98%

See 1 more Smart Citation

Transferability of continuous‐ and class‐pedotransfer functions to predict water retention properties of semiarid Syrian soils

Majou

Hassani

Bruand

2018

Soil Use and Management

View full text Add to dashboard Cite

Hydraulic properties of soils, particularly water retention, are key for appropriate management of semiarid soils. Very few pedotransfer functions (PTFs) have been developed to predict these properties for soils of Mediterranean regions, where data are particularly scarce. We investigated the transferability of PTFs to semiarid soils. The quality of the prediction was compared to that for soils originating from temperate regions for which most PTFs were developed. We used two soil data sets: one from the Paris basin (French data set, n = 30) and a Syrian data set (n = 30). Soil samples were collected in winter when the water content was near field capacity. Composition and water content of the samples were determined at seven water potentials. Continuous‐ and class‐PTFs developed using different predictors were tested using the two data sets and their performance compared to those developed using artificial neural networks (ANN). The best performance and transferability of the PTFs for both data sets used soil water content at field capacity as predictor after stratification by texture. The quality of prediction was similar to that for ANN‐PTFs. Continuous‐ and class‐PTFs may be transferable to other countries with performances that vary according to their ability to account for variation in soil composition and structure. Taking into account predictors of composition (particle size distribution, texture, organic carbon content) and structure (bulk density, porosity, field capacity) did not lead to a better performance or the best transferability potential.

show abstract

“…Nguyen et al (2017) described 3 groups of PTFs for SWRC. The first group (point-PTFs) estimates soil moisture at specific matric potentials.…”

Section: Introductionmentioning

confidence: 99%

Pedotransfer functions of soil water properties to estimate the S-index

Caviglione

2018

Rev. bras. eng. agríc. ambient.

View full text Add to dashboard Cite

One big challenge for soil science is to translate existing data into data that is needed. Pedotransfer functions have been proposed for this purpose and they can be point or parametric when estimating the water retention characteristics. Many indicators of soil physical quality have been proposed, including the S-Index proposed by Dexter. The objective of this study was to assess the use of pedotransfer functions for soil water retention to estimate the S-index under field conditions in the diversity of soils of the Paraná state. Soil samples were collected from 36 sites with textures ranging from sandy to heavy clay in the layers of 0-0.10 and 0.10-0.20 m and under two conditions (native forest and cultivated soil). Water content at six matric potentials, bulk density and contents of clay, sand and silt were determined. Soil-water retention curve was fitted by the van Genuchten-Mualem model and the S-index was calculated. S-index was estimated from water retention curves obtained by the pedotransfer function of Tomasella (point and parametric). Although the coefficient of determination varied from 0.759 to 0.895, modeling efficiency was negative and the regression coefficient between observed and predicted data was different from 1 in all comparisons. Under field conditions in the soil diversity of the Paraná state, restrictions were found in S-index estimation using the evaluated pedotransfer functions.

show abstract

Comparison of statistical regression and data-mining techniques in estimating soil water retention of tropical delta soils

Cited by 44 publications

References 45 publications

Evaluation of Parametric and Nonparametric Machine‐Learning Techniques for Prediction of Saturated and Near‐Saturated Hydraulic Conductivity

Evaluation of Parametric and Nonparametric Machine‐Learning Techniques for Prediction of Saturated and Near‐Saturated Hydraulic Conductivity

Transferability of continuous‐ and class‐pedotransfer functions to predict water retention properties of semiarid Syrian soils

Pedotransfer functions of soil water properties to estimate the S-index

Contact Info

Product

Resources

About