Valorisation d'une prospection électrique pour établir une carte de l'épaisseur du sol

Abstract : In the framework of CAREX project, this work aims at studying and comparing two approaches (spatial and non-spatial) to map the soil depth (the target variable) from the soil electrical resistivity (the auxiliary variable). Therefore, an electrical resistivity survey was performed by the ARP mobile device throughout the experimental site of INRA Epoisses. A principal component analysis of the three arrays of electrical resistivity measurements demonstrated that the first principal component (PC1) accounted for 85 % of the total variance. The non-spatial approach was based on two simple linear regression models considering the third electrical resistivity array ? that is the best correlated with the soil thickness - or PC1, as independent variables to model and map the target variable. The spatial approach, considers PC1 and its low frequencies (S2-PC1) respectively, as an external drift in kriging of the target variable. The results have shown that the spatial approach (especially kriging with S2-PC1 as an external drift) allowed reproducing the full spectrum of variation measured in the training data. In addition, the spatial approach provided more realistic maps compared to the statistical approach. Indeed, these maps are more contrasted with a smooth aspect and with larger homogeneous areas and would then be helpful for management at the plot scale.
Document type :
Master thesis
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/hal-01776951
Contributor : Memoires Ul <>
Submitted on : Monday, October 28, 2019 - 3:01:07 PM
Last modification on : Monday, October 28, 2019 - 3:02:38 PM

File

BUS_M_2013_NDJIBA_MITOMBO_DARI...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01776951, version 1

Collections

Citation

Daril Ndjiba Mitombo. Valorisation d'une prospection électrique pour établir une carte de l'épaisseur du sol. Sciences de la Terre. 2013. ⟨hal-01776951⟩

Share

Metrics

Record views

10