Improved prediction of unsaturated hydraulic conductivity with the Mualem-van Genuchten model

Marcel Schaap, Feike J. Leij

Research output: Contribution to journalArticle

276 Citations (Scopus)

Abstract

In many vadose zone hydrological studies, it is imperative that the soil's unsaturated hydraulic conductivity is known. Frequently, the Mualem-van Genuchten model (MVG) is used for this purpose because it allows prediction of unsaturated hydraulic conductivity from water retention parameters. For this and similar equations, it is often assumed that a measured saturated hydraulic conductivity (K(s)) can be used as a matching point (K(o)) while a factor S(c)/(L) is used to account for pore connectivity and tortuosity (where S(e) is the relative saturation and L = 0.5). We used a data set of 235 soil samples with retention and unsaturated hydraulic conductivity data to test and improve predictions with the MVG equation. The standard practice of using K(o) = K(s) and L = 0.5 resulted in a root mean square error for log(K) (RMSE(K)) of 1.31. Optimization of the matching point (K(o)) and L to the hydraulic conductivity data yielded a RMSE(K) of 0.41. The fitted K(o) were, on average, about one order of magnitude smaller than measured K(s). Furthermore, L was predominantly negative, casting doubt that the MVG can be interpreted in a physical way. Spearman rank correlations showed that both K(o) and L were related to van Genuchten water retention parameters and neural network analyses confirmed that K(o) and L could indeed be predicted in this way. The corresponding RMSE(K) was 0.84, which was half an order of magnitude better than the traditional MVG model. Bulk density and textural parameters were poor predictors while addition of K(s) improved the RMSE(K) only marginally. Bootstrap analysis showed that the uncertainty in predicted unsaturated hydraulic conductivity was about one order of magnitude near saturation and larger at lower water contents.

Original languageEnglish (US)
Pages (from-to)843-851
Number of pages9
JournalSoil Science Society of America Journal
Volume64
Issue number3
StatePublished - May 2000
Externally publishedYes

Fingerprint

unsaturated hydraulic conductivity
hydraulic conductivity
prediction
water retention
uncertainty analysis
saturation
vadose zone
saturated hydraulic conductivity
tortuosity
neural networks
bulk density
soil sampling
water
connectivity
water content
soil
testing
parameter

ASJC Scopus subject areas

  • Soil Science
  • Earth-Surface Processes

Cite this

Improved prediction of unsaturated hydraulic conductivity with the Mualem-van Genuchten model. / Schaap, Marcel; Leij, Feike J.

In: Soil Science Society of America Journal, Vol. 64, No. 3, 05.2000, p. 843-851.

Research output: Contribution to journalArticle

@article{64f129bd891e47c7bf23c4c7d9ba1627,
title = "Improved prediction of unsaturated hydraulic conductivity with the Mualem-van Genuchten model",
abstract = "In many vadose zone hydrological studies, it is imperative that the soil's unsaturated hydraulic conductivity is known. Frequently, the Mualem-van Genuchten model (MVG) is used for this purpose because it allows prediction of unsaturated hydraulic conductivity from water retention parameters. For this and similar equations, it is often assumed that a measured saturated hydraulic conductivity (K(s)) can be used as a matching point (K(o)) while a factor S(c)/(L) is used to account for pore connectivity and tortuosity (where S(e) is the relative saturation and L = 0.5). We used a data set of 235 soil samples with retention and unsaturated hydraulic conductivity data to test and improve predictions with the MVG equation. The standard practice of using K(o) = K(s) and L = 0.5 resulted in a root mean square error for log(K) (RMSE(K)) of 1.31. Optimization of the matching point (K(o)) and L to the hydraulic conductivity data yielded a RMSE(K) of 0.41. The fitted K(o) were, on average, about one order of magnitude smaller than measured K(s). Furthermore, L was predominantly negative, casting doubt that the MVG can be interpreted in a physical way. Spearman rank correlations showed that both K(o) and L were related to van Genuchten water retention parameters and neural network analyses confirmed that K(o) and L could indeed be predicted in this way. The corresponding RMSE(K) was 0.84, which was half an order of magnitude better than the traditional MVG model. Bulk density and textural parameters were poor predictors while addition of K(s) improved the RMSE(K) only marginally. Bootstrap analysis showed that the uncertainty in predicted unsaturated hydraulic conductivity was about one order of magnitude near saturation and larger at lower water contents.",
author = "Marcel Schaap and Leij, {Feike J.}",
year = "2000",
month = "5",
language = "English (US)",
volume = "64",
pages = "843--851",
journal = "Soil Science Society of America Journal",
issn = "0361-5995",
publisher = "Soil Science Society of America",
number = "3",

}

TY - JOUR

T1 - Improved prediction of unsaturated hydraulic conductivity with the Mualem-van Genuchten model

AU - Schaap, Marcel

AU - Leij, Feike J.

PY - 2000/5

Y1 - 2000/5

N2 - In many vadose zone hydrological studies, it is imperative that the soil's unsaturated hydraulic conductivity is known. Frequently, the Mualem-van Genuchten model (MVG) is used for this purpose because it allows prediction of unsaturated hydraulic conductivity from water retention parameters. For this and similar equations, it is often assumed that a measured saturated hydraulic conductivity (K(s)) can be used as a matching point (K(o)) while a factor S(c)/(L) is used to account for pore connectivity and tortuosity (where S(e) is the relative saturation and L = 0.5). We used a data set of 235 soil samples with retention and unsaturated hydraulic conductivity data to test and improve predictions with the MVG equation. The standard practice of using K(o) = K(s) and L = 0.5 resulted in a root mean square error for log(K) (RMSE(K)) of 1.31. Optimization of the matching point (K(o)) and L to the hydraulic conductivity data yielded a RMSE(K) of 0.41. The fitted K(o) were, on average, about one order of magnitude smaller than measured K(s). Furthermore, L was predominantly negative, casting doubt that the MVG can be interpreted in a physical way. Spearman rank correlations showed that both K(o) and L were related to van Genuchten water retention parameters and neural network analyses confirmed that K(o) and L could indeed be predicted in this way. The corresponding RMSE(K) was 0.84, which was half an order of magnitude better than the traditional MVG model. Bulk density and textural parameters were poor predictors while addition of K(s) improved the RMSE(K) only marginally. Bootstrap analysis showed that the uncertainty in predicted unsaturated hydraulic conductivity was about one order of magnitude near saturation and larger at lower water contents.

AB - In many vadose zone hydrological studies, it is imperative that the soil's unsaturated hydraulic conductivity is known. Frequently, the Mualem-van Genuchten model (MVG) is used for this purpose because it allows prediction of unsaturated hydraulic conductivity from water retention parameters. For this and similar equations, it is often assumed that a measured saturated hydraulic conductivity (K(s)) can be used as a matching point (K(o)) while a factor S(c)/(L) is used to account for pore connectivity and tortuosity (where S(e) is the relative saturation and L = 0.5). We used a data set of 235 soil samples with retention and unsaturated hydraulic conductivity data to test and improve predictions with the MVG equation. The standard practice of using K(o) = K(s) and L = 0.5 resulted in a root mean square error for log(K) (RMSE(K)) of 1.31. Optimization of the matching point (K(o)) and L to the hydraulic conductivity data yielded a RMSE(K) of 0.41. The fitted K(o) were, on average, about one order of magnitude smaller than measured K(s). Furthermore, L was predominantly negative, casting doubt that the MVG can be interpreted in a physical way. Spearman rank correlations showed that both K(o) and L were related to van Genuchten water retention parameters and neural network analyses confirmed that K(o) and L could indeed be predicted in this way. The corresponding RMSE(K) was 0.84, which was half an order of magnitude better than the traditional MVG model. Bulk density and textural parameters were poor predictors while addition of K(s) improved the RMSE(K) only marginally. Bootstrap analysis showed that the uncertainty in predicted unsaturated hydraulic conductivity was about one order of magnitude near saturation and larger at lower water contents.

UR - http://www.scopus.com/inward/record.url?scp=0034193893&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034193893&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0034193893

VL - 64

SP - 843

EP - 851

JO - Soil Science Society of America Journal

JF - Soil Science Society of America Journal

SN - 0361-5995

IS - 3

ER -