Multivariate regional deep learning prediction of soil properties from near-infrared, mid-infrared and their combined spectra

Nyawasha Rumbidzai, Wadoux Alexandre M. J. C., Todoroff Pierre, Chikowo Régis, Falconnier Gatien, Lagorsse Maeva, Corbeels Marc, Cardinael Rémi. 2024. Multivariate regional deep learning prediction of soil properties from near-infrared, mid-infrared and their combined spectra. Geoderma Regional, 37:e00805, 11 p.

Résumé : Artificial neural network (ANN) models have been successfully used in infrared spectroscopy research for the prediction of soil properties. They often show better performance than conventional methods such as partial least squares regression (PLSR). In this paper we develop and evaluate a multivariate extension of ANN for predicting correlated soil properties: total carbon (C), total nitrogen (N), clay, silt, and sand contents, using visible near-infrared (vis-NIR), mid-infrared (MIR) or combined spectra (vis-NIR + MIR). We hypothesize that accounting for the correlation through joint modelling of soil properties with a single model can eliminate “pedological chimera”: unrealistic values that may arise when properties are predicted independently such as when calculating ratio or soil texture values. We tested two types of ANN models, a univariate (ANN-UV) and a multivariate model (ANN-MV), using a dataset of 228 soil samples collected from Murehwa district in Zimbabwe at two soil depth intervals (0–20 and 20–40 cm). The models were compared with results from a univariate PLSR (PLSR-UV) model. We found that the multivariate ANN model was better at conserving the observed correlations between properties and consequently gave realistic soil C:N and C:Clay ratios, but that there was no improvement in prediction accuracy over using a univariate model (ANN or PLSR). The use of combined spectra (vis-NIR + MIR) did not make any significant improvements in prediction accuracy of the multivariate ANN model compared to using the vis-NIR or MIR only. We conclude that the multivariate ANN model is better suited for the prediction of multiple correlated soil properties and that it is flexible and can account for compositional constrains. The multivariate ANN model helps to keep realistic ratio values – with strong implications for assessment studies that make use of such predicted soil values.

Mots-clés Agrovoc : réseau de neurones, spectroscopie, caractéristiques du sol, apprentissage machine, technique de prévision, conservation des sols, sol argileux, spectroscopie infrarouge

Mots-clés géographiques Agrovoc : Zimbabwe, La Réunion, France

Mots-clés libres : Spectroscopy, MIRS, Infrared spectroscopy, Artificial Neural Networks, Sub Saharan Africa, Soil organic carbon, Texture, Keras

Classification Agris : P30 - Sciences et aménagement du sol
P33 - Chimie et physique du sol
U10 - Informatique, mathématiques et statistiques

Champ stratégique Cirad : CTS 5 (2019-) - Territoires

Agences de financement hors UE : Agropolis Fondation, Agence Nationale de la Recherche, Total Foundation

Projets sur financement : (CHE) Agricultural Intensification and Dynamics of Soil Carbon Sequestration in Tropical and Temperate Farming Systems, (FRA) Agricultural Sciences for sustainable Development

Auteurs et affiliations

• Nyawasha Rumbidzai, University of Zimbabwe (ZWE) - auteur correspondant
• Wadoux Alexandre M. J. C., IRD (FRA)
• Todoroff Pierre, CIRAD-PERSYST-UPR AIDA (REU)
• Chikowo Régis, University of Zimbabwe (ZWE)
• Falconnier Gatien, CIRAD-PERSYST-UPR AIDA (ZWE) ORCID: 0000-0003-3291-650X
• Lagorsse Maeva, Université de Montpellier (FRA)
• Corbeels Marc, CIRAD-PERSYST-UPR AIDA (KEN) ORCID: 0000-0002-8084-9287
• Cardinael Rémi, CIRAD-PERSYST-UPR AIDA (ZWE) ORCID: 0000-0002-9924-3269

Source : Cirad-Agritrop (https://agritrop.cirad.fr/609432/)

Métriques

[ Page générée et mise en cache le 2024-12-18 ]