Explainable AI Analysis in Soil Gas Detection

• Felipe Valencia de Almeida University of São Paulo (USP) http://orcid.org/0000-0003-2031-6443
• Rosa Virginia Encinas Quille University of São Paulo (USP) https://orcid.org/0000-0001-9166-1741
• Danielle Monteiro University of São Paulo (USP) https://orcid.org/0009-0009-3954-2687
• Leandro Gomes de Freitas Institute for Technological Research (IPT) https://orcid.org/0000-0002-7860-5937
• Pedro Luiz Pizzigatti Corrêa University of São Paulo (USP)

DOI: https://doi.org/10.5753/bresci.2025.248241

Abstract

The study and monitoring of contaminated areas are essential for environmental risk assessment and decision-making. Among the pollutants of interest, subsurface contaminant gases, such as methane, are a concern due to their significant environmental impact. In this context, predictive analyses have been increasingly used to estimate the presence and concentration of these gases based on other variables. Machine learning models have shown good performance in such predictions; however, they often operate as ”black boxes,”making it difficult to understand how the results were obtained. Explainable AI aims to use techniques that enable the interpretation of model behavior, providing support for understanding how the presented results were generated. In this work, we employed SHAP and LIME techniques to analyze the explainability of Random Forest and XGBoost models in predicting contaminant gases in soil. A case study was conducted using data collected from gas monitoring wells at the USP-Leste campus, where the use of these techniques was found to improve model explainability.

Keywords: Machine Learning, xAI, Contaminated areas

References

BRZICA, D. (2022). Governing major cities in the context of new environmental challenges. Ecomomic Review/Ekonomické Rozhl’ady, 51(3).

CETESB (2022). Manual de gerenciamento de Áreas contaminadas. Technical report, CETESB.

Hartmann, A., Jasechko, S., Gleeson, T., Wada, Y., Andreo, B., Barberá, J. A., Brielmann, H., Bouchaou, L., Charlier, J.-B., Darling, W. G., Filippini, M., Garvelmann, J., Goldscheider, N., Kralik, M., Kunstmann, H., Ladouche, B., Lange, J., Lucianetti, G., Martín, J. F., Mudarra, M., Sánchez, D., Stumpp, C., Zagana, E., and Wagener, T. (2021). Risk of groundwater contamination widely underestimated because of fast flow into aquifers. Proceedings of the National Academy of Sciences, 118(20):e2024492118.

Hou, D., Al-Tabbaa, A., O’Connor, D., Hu, Q., Zhu, Y.-G., Wang, L., Kirkwood, N., Ok, Y. S., Tsang, D. C. W., Bolan, N. S., and Rinklebe, J. (2023). Sustainable remediation and redevelopment of brownfield sites. Nature Reviews Earth & Environment, 4(4):271–286.

Li, X., Yi, S., Cundy, A. B., and Chen, W. (2022). Sustainability transitions of contaminated sites: A global meta-analysis on economic effects of remediation behaviour. Land Degradation & Development, 33(11):1775–1786.

Lundberg, S. M. and Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in neural information processing systems, 30.

Moraes, S., Teixeira, C., and Maximiano, A. (2014). Guia de elaboração de planos de intervenção para o gerenciamento de áreas contaminadas. IPT - Instituto de Pesquisas Tecnológicas do Estado de São Paulo.

Mulenga, C. (2022). Soil governance and the control of mining pollution in zambia. Soil Security, 6:100039.

Quille, R., Vieira, G., Freitas, L., Corrêa, P., Souza, S., Barbosa, A., and Almeida, F. (2025). Dataset de poços de monitoramento de gases da usp leste. In Anais do XVI Workshop de Computação Aplicada à Gestão do Meio Ambiente e Recursos Naturais, pages 352–355, Porto Alegre, RS, Brasil. SBC.

Ribeiro, M. T., Singh, S., and Guestrin, C. (2016). ”why should i trust you?”explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pages 1135–1144.

SHAP (2025). Beeswarm plot. Disponível em: [link]. Acesso em 25 de junho de 2025.

Tarazona, J. V. (2024). Pollution, soil. In Wexler, P., editor, Encyclopedia of Toxicology (Fourth Edition), pages 801–807. Academic Press, Oxford, fourth edition edition.