Machine-learning-based quantitative estimation of soil organic carbon content by VIS/NIR spectroscopy

Jianli Ding, Aixia Yang, Jingzhe Wang, Vasit Sagan, Danlin Yu

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Soil organic carbon (SOC) is an important soil property that has profound impact on soil quality and plant growth. With 140 soil samples collected from Ebinur Lake Wetland National Nature Reserve, Xinjiang Uyghur Autonomous Region of China, this research evaluated the feasibility of visible/near infrared (VIS/NIR) spectroscopy data (350-2,500 nm) and simulated EO-1 Hyperion data to estimate SOC in arid wetland regions. Three machine learning algorithms including Ant Colony Optimization-interval Partial Least Squares (ACO-iPLS), Recursive Feature Elimination-Support Vector Machine (RF-SVM), and Random Forest (RF) were employed to select spectral features and further estimate SOC. Results indicated that the feature wavelengths pertaining to SOC were mainly within the ranges of 745-910 nm and 1,911-2,254 nm. The combination of RF-SVM and first derivative pre-processing produced the highest estimation accuracy with the optimal values of Rt (correlation coefficient of testing set), RMSEt and RPD of 0.91, 0.27% and 2.41, respectively. The simulated EO-1 Hyperion data combined with Support Vector Machine (SVM) based recursive feature elimination algorithm produced the most accurate estimate of SOC content. For the testing set, Rt was 0.79, RMSEt was 0.19%, and RPD was 1.61. This practice provides an efficient, low-cost approach with potentially high accuracy to estimate SOC contents and hence supports better management and protection strategies for desert wetland ecosystems.

Original languageEnglish
Article numbere5714
JournalPeerJ
Volume2018
Issue number10
DOIs
StatePublished - 1 Jan 2018

Fingerprint

ultraviolet-visible spectroscopy
Near infrared spectroscopy
Near-Infrared Spectroscopy
artificial intelligence
near-infrared spectroscopy
Organic carbon
soil organic carbon
Learning systems
Soil
Carbon
Soils
wetlands
Wetlands
Support vector machines
China
Machine Learning
wavelengths
least squares
deserts
soil quality

Keywords

  • Desert wetland soil
  • Ebinur lake wetland
  • Machine lea
  • Soil organic carbon

Cite this

Ding, Jianli ; Yang, Aixia ; Wang, Jingzhe ; Sagan, Vasit ; Yu, Danlin. / Machine-learning-based quantitative estimation of soil organic carbon content by VIS/NIR spectroscopy. In: PeerJ. 2018 ; Vol. 2018, No. 10.
@article{54bae12e9a00468da7bf412562714492,
title = "Machine-learning-based quantitative estimation of soil organic carbon content by VIS/NIR spectroscopy",
abstract = "Soil organic carbon (SOC) is an important soil property that has profound impact on soil quality and plant growth. With 140 soil samples collected from Ebinur Lake Wetland National Nature Reserve, Xinjiang Uyghur Autonomous Region of China, this research evaluated the feasibility of visible/near infrared (VIS/NIR) spectroscopy data (350-2,500 nm) and simulated EO-1 Hyperion data to estimate SOC in arid wetland regions. Three machine learning algorithms including Ant Colony Optimization-interval Partial Least Squares (ACO-iPLS), Recursive Feature Elimination-Support Vector Machine (RF-SVM), and Random Forest (RF) were employed to select spectral features and further estimate SOC. Results indicated that the feature wavelengths pertaining to SOC were mainly within the ranges of 745-910 nm and 1,911-2,254 nm. The combination of RF-SVM and first derivative pre-processing produced the highest estimation accuracy with the optimal values of Rt (correlation coefficient of testing set), RMSEt and RPD of 0.91, 0.27{\%} and 2.41, respectively. The simulated EO-1 Hyperion data combined with Support Vector Machine (SVM) based recursive feature elimination algorithm produced the most accurate estimate of SOC content. For the testing set, Rt was 0.79, RMSEt was 0.19{\%}, and RPD was 1.61. This practice provides an efficient, low-cost approach with potentially high accuracy to estimate SOC contents and hence supports better management and protection strategies for desert wetland ecosystems.",
keywords = "Desert wetland soil, Ebinur lake wetland, Machine lea, Soil organic carbon",
author = "Jianli Ding and Aixia Yang and Jingzhe Wang and Vasit Sagan and Danlin Yu",
year = "2018",
month = "1",
day = "1",
doi = "10.7717/peerj.5714",
language = "English",
volume = "2018",
journal = "PeerJ",
issn = "2167-8359",
publisher = "PeerJ Inc.",
number = "10",

}

Machine-learning-based quantitative estimation of soil organic carbon content by VIS/NIR spectroscopy. / Ding, Jianli; Yang, Aixia; Wang, Jingzhe; Sagan, Vasit; Yu, Danlin.

In: PeerJ, Vol. 2018, No. 10, e5714, 01.01.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Machine-learning-based quantitative estimation of soil organic carbon content by VIS/NIR spectroscopy

AU - Ding, Jianli

AU - Yang, Aixia

AU - Wang, Jingzhe

AU - Sagan, Vasit

AU - Yu, Danlin

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Soil organic carbon (SOC) is an important soil property that has profound impact on soil quality and plant growth. With 140 soil samples collected from Ebinur Lake Wetland National Nature Reserve, Xinjiang Uyghur Autonomous Region of China, this research evaluated the feasibility of visible/near infrared (VIS/NIR) spectroscopy data (350-2,500 nm) and simulated EO-1 Hyperion data to estimate SOC in arid wetland regions. Three machine learning algorithms including Ant Colony Optimization-interval Partial Least Squares (ACO-iPLS), Recursive Feature Elimination-Support Vector Machine (RF-SVM), and Random Forest (RF) were employed to select spectral features and further estimate SOC. Results indicated that the feature wavelengths pertaining to SOC were mainly within the ranges of 745-910 nm and 1,911-2,254 nm. The combination of RF-SVM and first derivative pre-processing produced the highest estimation accuracy with the optimal values of Rt (correlation coefficient of testing set), RMSEt and RPD of 0.91, 0.27% and 2.41, respectively. The simulated EO-1 Hyperion data combined with Support Vector Machine (SVM) based recursive feature elimination algorithm produced the most accurate estimate of SOC content. For the testing set, Rt was 0.79, RMSEt was 0.19%, and RPD was 1.61. This practice provides an efficient, low-cost approach with potentially high accuracy to estimate SOC contents and hence supports better management and protection strategies for desert wetland ecosystems.

AB - Soil organic carbon (SOC) is an important soil property that has profound impact on soil quality and plant growth. With 140 soil samples collected from Ebinur Lake Wetland National Nature Reserve, Xinjiang Uyghur Autonomous Region of China, this research evaluated the feasibility of visible/near infrared (VIS/NIR) spectroscopy data (350-2,500 nm) and simulated EO-1 Hyperion data to estimate SOC in arid wetland regions. Three machine learning algorithms including Ant Colony Optimization-interval Partial Least Squares (ACO-iPLS), Recursive Feature Elimination-Support Vector Machine (RF-SVM), and Random Forest (RF) were employed to select spectral features and further estimate SOC. Results indicated that the feature wavelengths pertaining to SOC were mainly within the ranges of 745-910 nm and 1,911-2,254 nm. The combination of RF-SVM and first derivative pre-processing produced the highest estimation accuracy with the optimal values of Rt (correlation coefficient of testing set), RMSEt and RPD of 0.91, 0.27% and 2.41, respectively. The simulated EO-1 Hyperion data combined with Support Vector Machine (SVM) based recursive feature elimination algorithm produced the most accurate estimate of SOC content. For the testing set, Rt was 0.79, RMSEt was 0.19%, and RPD was 1.61. This practice provides an efficient, low-cost approach with potentially high accuracy to estimate SOC contents and hence supports better management and protection strategies for desert wetland ecosystems.

KW - Desert wetland soil

KW - Ebinur lake wetland

KW - Machine lea

KW - Soil organic carbon

UR - http://www.scopus.com/inward/record.url?scp=85055112702&partnerID=8YFLogxK

U2 - 10.7717/peerj.5714

DO - 10.7717/peerj.5714

M3 - Article

AN - SCOPUS:85055112702

VL - 2018

JO - PeerJ

JF - PeerJ

SN - 2167-8359

IS - 10

M1 - e5714

ER -