Soil classification based on physical and chemical properties using random forests

Didier Dias*, Bruno Martins, João Pires, Luís M. de Sousa, Jacinto Estima, Carlos V. Damásio

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference paper

Abstract

Soil classification is a method of encoding the most relevant information about a given soil, namely its composition and characteristics, in a single class, to be used in areas like agriculture and forestry. In this paper, we evaluate how confidently we can predict soil classes, following the World Reference Base classification system, based on the physical and chemical characteristics of its layers. The Random Forests classifier was used with data consisting of 6 760 soil profiles composed by 19 464 horizons, collected in Mexico. Four methods of modelling the data were tested (i.e., standard depths, n first layers, thickness, and area weighted thickness). We also fine-tuned the best parameters for the classifier and for a k-NN imputation algorithm, used for addressing problems of missing data. Under-represented classes showed significantly worse results, by being repeatedly predicted as one of the majority classes. The best method to model the data was found to be the n first layers approach, with missing values being imputed with k-NN ($$k=1$$ ). The results present a Kappa value from 0.36 to 0.48 and were in line with the state of the art methods, which mostly use remote sensing data.

Original languageEnglish
Title of host publicationProgress in Artificial Intelligence - 19th EPIA Conference on Artificial Intelligence, EPIA 2019, Proceedings
EditorsPaulo Moura Oliveira, Paulo Novais, Luís Paulo Reis
PublisherSpringer Verlag
Pages212-223
ISBN (Electronic)9783030302412
ISBN (Print)9783030302405
DOIs
Publication statusPublished - 30 Aug 2019
Event19th EPIA Conference on Artificial Intelligence, EPIA 2019 - Vila Real, Portugal
Duration: 3 Sep 20196 Sep 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11804 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference19th EPIA Conference on Artificial Intelligence, EPIA 2019
CountryPortugal
CityVila Real
Period3/09/196/09/19

    Fingerprint

Keywords

  • Ensemble learning
  • Machine learning
  • Random Forests
  • Soil classification
  • Soil properties

Cite this

Dias, D., Martins, B., Pires, J., de Sousa, L. M., Estima, J., & Damásio, C. V. (2019). Soil classification based on physical and chemical properties using random forests. In P. Moura Oliveira, P. Novais, & L. P. Reis (Eds.), Progress in Artificial Intelligence - 19th EPIA Conference on Artificial Intelligence, EPIA 2019, Proceedings (pp. 212-223). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11804 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-30241-2_19