Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets

Understanding spatial and temporal variability in soil organic carbon (SOC) content helps simultaneously assess soil fertility and several parameters that are strongly associated with it, such as structural stability, nutrient cycling, biological activity, and soil aeration. Therefore, it appears ne...

Full description

Bibliographic Details
Main Authors: Hayfa Zayani, Youssef Fouad, Didier Michot, Zeineb Kassouk, Nicolas Baghdadi, Emmanuelle Vaudour, Zohra Lili-Chabaane, Christian Walter
Format: Article
Language:English
Published: MDPI AG 2023-08-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/15/17/4264
_version_ 1827727819865587712
author Hayfa Zayani
Youssef Fouad
Didier Michot
Zeineb Kassouk
Nicolas Baghdadi
Emmanuelle Vaudour
Zohra Lili-Chabaane
Christian Walter
author_facet Hayfa Zayani
Youssef Fouad
Didier Michot
Zeineb Kassouk
Nicolas Baghdadi
Emmanuelle Vaudour
Zohra Lili-Chabaane
Christian Walter
author_sort Hayfa Zayani
collection DOAJ
description Understanding spatial and temporal variability in soil organic carbon (SOC) content helps simultaneously assess soil fertility and several parameters that are strongly associated with it, such as structural stability, nutrient cycling, biological activity, and soil aeration. Therefore, it appears necessary to monitor SOC regularly and investigate rapid, non-destructive, and cost-effective approaches for doing so, such as proximal and remote sensing. To increase the accuracy of predictions of SOC content, this study evaluated combining remote sensing time series with laboratory spectral measurements using machine and deep-learning algorithms. Partial least squares (PLS) regression, random forest (RF), and deep neural network (DNN) models were developed using Sentinel-2 (S2) time series of 58 sampling points of bare soil and according to three approaches. In the first approach, only S2 bands were used to calibrate and compare the performance of the models. In the second, S2 indices, Sentinel-1 (S1) indices, and S1 soil moisture were added separately during model calibration to evaluate their effects individually and then together. In the third, we added the laboratory indices incrementally and tested their influence on model accuracy. Using only S2 bands, the DNN model outperformed the PLS and RF models (ratio of performance to the interquartile distance RPIQ = 0.79, 1.36 and 1.67, respectively). Additional information improved performances only for model calibration, with S1 soil moisture yielding the most stable improvement among three iterations. Including equivalent indices of the S2 indices calculated using soil spectra obtained under laboratory conditions improved prediction of SOC, and the use of only two indices achieved good validation performances for the RF and DNN models (mean RPIQ = 2.01 and 1.77, respectively).
first_indexed 2024-03-10T23:14:23Z
format Article
id doaj.art-6f9f596457b6460c8236e4598e6249cb
institution Directory Open Access Journal
issn 2072-4292
language English
last_indexed 2024-03-10T23:14:23Z
publishDate 2023-08-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj.art-6f9f596457b6460c8236e4598e6249cb2023-11-19T08:46:44ZengMDPI AGRemote Sensing2072-42922023-08-011517426410.3390/rs15174264Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral DatasetsHayfa Zayani0Youssef Fouad1Didier Michot2Zeineb Kassouk3Nicolas Baghdadi4Emmanuelle Vaudour5Zohra Lili-Chabaane6Christian Walter7SAS, Institut Agro, INRAE, 65 Rue de St Brieuc, 35000 Rennes, FranceSAS, Institut Agro, INRAE, 65 Rue de St Brieuc, 35000 Rennes, FranceSAS, Institut Agro, INRAE, 65 Rue de St Brieuc, 35000 Rennes, FranceUniversité de Carthage, Institut National Agronomique de Tunisie, LR 17AGR01 (Lr GREEN-TEAM), 43 Avenue Charles Nicolle, Tunis 1082, TunisiaCIRAD, CNRS, INRAE, TETIS, Université de Montpellier, AgroParisTech, CEDEX 5, 34093 Montpellier, FranceINRAE, Université Paris-Saclay, AgroParisTech, UMR EcoSys, 91120 Palaiseau, FranceUniversité de Carthage, Institut National Agronomique de Tunisie, LR 17AGR01 (Lr GREEN-TEAM), 43 Avenue Charles Nicolle, Tunis 1082, TunisiaSAS, Institut Agro, INRAE, 65 Rue de St Brieuc, 35000 Rennes, FranceUnderstanding spatial and temporal variability in soil organic carbon (SOC) content helps simultaneously assess soil fertility and several parameters that are strongly associated with it, such as structural stability, nutrient cycling, biological activity, and soil aeration. Therefore, it appears necessary to monitor SOC regularly and investigate rapid, non-destructive, and cost-effective approaches for doing so, such as proximal and remote sensing. To increase the accuracy of predictions of SOC content, this study evaluated combining remote sensing time series with laboratory spectral measurements using machine and deep-learning algorithms. Partial least squares (PLS) regression, random forest (RF), and deep neural network (DNN) models were developed using Sentinel-2 (S2) time series of 58 sampling points of bare soil and according to three approaches. In the first approach, only S2 bands were used to calibrate and compare the performance of the models. In the second, S2 indices, Sentinel-1 (S1) indices, and S1 soil moisture were added separately during model calibration to evaluate their effects individually and then together. In the third, we added the laboratory indices incrementally and tested their influence on model accuracy. Using only S2 bands, the DNN model outperformed the PLS and RF models (ratio of performance to the interquartile distance RPIQ = 0.79, 1.36 and 1.67, respectively). Additional information improved performances only for model calibration, with S1 soil moisture yielding the most stable improvement among three iterations. Including equivalent indices of the S2 indices calculated using soil spectra obtained under laboratory conditions improved prediction of SOC, and the use of only two indices achieved good validation performances for the RF and DNN models (mean RPIQ = 2.01 and 1.77, respectively).https://www.mdpi.com/2072-4292/15/17/4264soil organic carbondeep learningSentinel-2spectral indicesSentinel-1soil moisture
spellingShingle Hayfa Zayani
Youssef Fouad
Didier Michot
Zeineb Kassouk
Nicolas Baghdadi
Emmanuelle Vaudour
Zohra Lili-Chabaane
Christian Walter
Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets
Remote Sensing
soil organic carbon
deep learning
Sentinel-2
spectral indices
Sentinel-1
soil moisture
title Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets
title_full Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets
title_fullStr Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets
title_full_unstemmed Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets
title_short Using Machine-Learning Algorithms to Predict Soil Organic Carbon Content from Combined Remote Sensing Imagery and Laboratory Vis-NIR Spectral Datasets
title_sort using machine learning algorithms to predict soil organic carbon content from combined remote sensing imagery and laboratory vis nir spectral datasets
topic soil organic carbon
deep learning
Sentinel-2
spectral indices
Sentinel-1
soil moisture
url https://www.mdpi.com/2072-4292/15/17/4264
work_keys_str_mv AT hayfazayani usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT yousseffouad usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT didiermichot usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT zeinebkassouk usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT nicolasbaghdadi usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT emmanuellevaudour usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT zohralilichabaane usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets
AT christianwalter usingmachinelearningalgorithmstopredictsoilorganiccarboncontentfromcombinedremotesensingimageryandlaboratoryvisnirspectraldatasets