Using neural networks to predict soil water retention and soil hydraulic conductivity

Marcel Schaap, Feike J. Leij

Research output: Contribution to journalArticle

136 Citations (Scopus)

Abstract

Direct measurement of hydraulic properties is time consuming, costly, and sometimes unreliable because of soil heterogeneity and experimental errors. Instead, hydraulic properties can be estimated from surrogate data such as soil texture and bulk density with pedotransfer functions (PTFs). This paper describes neural network PTFs to predict soil water retention, saturated and unsaturated hydraulic properties from limited or more extended sets of soil properties. Accuracy of prediction generally increased if more input data are used but there was always a considerable difference between predictions and measurements. The neural networks were combined with the bootstrap method to generate uncertainty estimates of the predicted hydraulic properties.

Original languageEnglish (US)
Pages (from-to)37-42
Number of pages6
JournalSoil and Tillage Research
Volume47
Issue number1-2
DOIs
StatePublished - Jun 2 1998
Externally publishedYes

Fingerprint

soil water retention
water retention
hydraulic property
hydraulic conductivity
neural networks
fluid mechanics
soil water
pedotransfer function
pedotransfer functions
soil
bootstrapping
prediction
soil heterogeneity
soil texture
bulk density
soil properties
soil property
uncertainty

Keywords

  • Hydraulic conductivity
  • Neural networks
  • Pedotransfer function
  • Water retention

ASJC Scopus subject areas

  • Agronomy and Crop Science
  • Soil Science

Cite this

Using neural networks to predict soil water retention and soil hydraulic conductivity. / Schaap, Marcel; Leij, Feike J.

In: Soil and Tillage Research, Vol. 47, No. 1-2, 02.06.1998, p. 37-42.

Research output: Contribution to journalArticle

@article{3581e61f0eff48f0bcd493518aa18315,
title = "Using neural networks to predict soil water retention and soil hydraulic conductivity",
abstract = "Direct measurement of hydraulic properties is time consuming, costly, and sometimes unreliable because of soil heterogeneity and experimental errors. Instead, hydraulic properties can be estimated from surrogate data such as soil texture and bulk density with pedotransfer functions (PTFs). This paper describes neural network PTFs to predict soil water retention, saturated and unsaturated hydraulic properties from limited or more extended sets of soil properties. Accuracy of prediction generally increased if more input data are used but there was always a considerable difference between predictions and measurements. The neural networks were combined with the bootstrap method to generate uncertainty estimates of the predicted hydraulic properties.",
keywords = "Hydraulic conductivity, Neural networks, Pedotransfer function, Water retention",
author = "Marcel Schaap and Leij, {Feike J.}",
year = "1998",
month = "6",
day = "2",
doi = "10.1016/S0167-1987(98)00070-1",
language = "English (US)",
volume = "47",
pages = "37--42",
journal = "Soil and Tillage Research",
issn = "0167-1987",
publisher = "Elsevier",
number = "1-2",

}

TY - JOUR

T1 - Using neural networks to predict soil water retention and soil hydraulic conductivity

AU - Schaap, Marcel

AU - Leij, Feike J.

PY - 1998/6/2

Y1 - 1998/6/2

N2 - Direct measurement of hydraulic properties is time consuming, costly, and sometimes unreliable because of soil heterogeneity and experimental errors. Instead, hydraulic properties can be estimated from surrogate data such as soil texture and bulk density with pedotransfer functions (PTFs). This paper describes neural network PTFs to predict soil water retention, saturated and unsaturated hydraulic properties from limited or more extended sets of soil properties. Accuracy of prediction generally increased if more input data are used but there was always a considerable difference between predictions and measurements. The neural networks were combined with the bootstrap method to generate uncertainty estimates of the predicted hydraulic properties.

AB - Direct measurement of hydraulic properties is time consuming, costly, and sometimes unreliable because of soil heterogeneity and experimental errors. Instead, hydraulic properties can be estimated from surrogate data such as soil texture and bulk density with pedotransfer functions (PTFs). This paper describes neural network PTFs to predict soil water retention, saturated and unsaturated hydraulic properties from limited or more extended sets of soil properties. Accuracy of prediction generally increased if more input data are used but there was always a considerable difference between predictions and measurements. The neural networks were combined with the bootstrap method to generate uncertainty estimates of the predicted hydraulic properties.

KW - Hydraulic conductivity

KW - Neural networks

KW - Pedotransfer function

KW - Water retention

UR - http://www.scopus.com/inward/record.url?scp=0032474481&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0032474481&partnerID=8YFLogxK

U2 - 10.1016/S0167-1987(98)00070-1

DO - 10.1016/S0167-1987(98)00070-1

M3 - Article

AN - SCOPUS:0032474481

VL - 47

SP - 37

EP - 42

JO - Soil and Tillage Research

JF - Soil and Tillage Research

SN - 0167-1987

IS - 1-2

ER -