Journal cover Journal topic
SOIL An interactive open-access journal of the European Geosciences Union
Journal topic

Journal metrics

Journal metrics

  • CiteScore value: 7.57 CiteScore 7.57
  • SNIP value: 2.708 SNIP 2.708
  • SJR value: 2.150 SJR 2.150
  • IPP value: 7.02 IPP 7.02
  • Scimago H index value: 17 Scimago H index 17
Discussion papers
https://doi.org/10.5194/soil-2018-28
© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/soil-2018-28
© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Original research article 03 Sep 2018

Original research article | 03 Sep 2018

Review status
This discussion paper is a preprint. It is a manuscript under review for the journal SOIL (SOIL).

Using deep learning for Digital Soil Mapping

José Padarian, Budiman Minasny, and Alex B. McBratney José Padarian et al.
  • Sydney Institute of Agriculture & School of Life and Environmental Sciences, The University of Sydney, New South Wales, Australia

Abstract. Digital soil mapping has been widely used as a cost-effective method for generating soil maps. However, current DSM data representation rarely incorporates contextual information of the landscape. DSM models are usually calibrated using point observations intersected with spatially corresponding point covariates. Here, we demonstrate the use of the convolutional neural network model that incorporates contextual information surrounding an observation to significantly improve the prediction accuracy over conventional DSM models. We describe a convolutional neural network (CNN) model that takes inputs as images of covariates and explores spatial contextual information by finding non-linear local spatial relationships of neighbouring pixels. Unique features of the proposed model include: input represented as 3D stack of images, data augmentation to reduce overfitting, and simultaneously predicting multiple outputs. Using a soil mapping example in Chile, the CNN model was trained to simultaneously predict soil organic carbon at multiples depths across the country. The results showed the CNN model reduced the error by 30% compared with conventional techniques that only used point information of covariates. In the example of country-wide mapping at 100m resolution, the neighbourhood size from 3 to 9 pixels is more effective than at a point location and larger neighbourhood sizes. In addition, the CNN model produces less prediction uncertainty and it is able to predict soil carbon at deeper soil layers more accurately. Because the CNN model takes covariate represented as images, it offers a simple and effective framework for future DSM models.

José Padarian et al.
Interactive discussion
Status: open (until 15 Dec 2018)
Status: open (until 15 Dec 2018)
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
[Subscribe to comment alert] Printer-friendly Version - Printer-friendly version Supplement - Supplement
José Padarian et al.
José Padarian et al.
Viewed  
Total article views: 697 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
518 173 6 697 10 10
  • HTML: 518
  • PDF: 173
  • XML: 6
  • Total: 697
  • BibTeX: 10
  • EndNote: 10
Views and downloads (calculated since 03 Sep 2018)
Cumulative views and downloads (calculated since 03 Sep 2018)
Viewed (geographical distribution)  
Total article views: 695 (including HTML, PDF, and XML) Thereof 686 with geography defined and 9 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Cited  
Saved  
No saved metrics found.
Discussed  
No discussed metrics found.
Latest update: 10 Dec 2018
Publications Copernicus
Download
Short summary
Digital soil mapping has been widely used as a cost-effective method for generating soil maps. DSM models are usually calibrated using point observations and rarely incorporate contextual information of the landscape. Here, we use convolutional neural networks to incorporate spatial context. We used as input a 3D stack of covariate images to simultaneously predicting organic carbon content at multiple depths. The model reduced the error by 30 % compared with conventional techniques.
Digital soil mapping has been widely used as a cost-effective method for generating soil maps....
Citation
Share