Paper contribution

Application of Machine Learning techniques for estimating sinkhole parameters caused by underground mining

Alicja Szmigiel; Krzysztof Tajduś; Janusz Rusek

Proceedings
EUROCK 2026
Pages
1305-1310
Submitted
28 January 2026
Accepted
15 June 2026
Published
12 August 2026

Summary

Abstract

Sinkholes induced by underground mining activities create a significant risks to infrastructure, the environment, and human safety. Reliable prediction of sinkhole development and its size parameters is therefore crucial for effective risk assessment and land-use planning in mining regions. This study explores the applicability of machine learning techniques for predicting the geometric characteristics of sinkholes formed above abandoned and active mining excavations. A comprehensive database was compiled from several European countries, integrating geological and mining information, including the parameters of underground workings and recorded sinkhole dimensions. Multiple supervised learning algorithms were trained and evaluated to predict sinkhole size parameters. Model performance was assessed using standard statistical metrics to identify the most accurate predictive approach. Feature importance analysis was conducted to determine which parameters most strongly influence sinkhole size predictions. The findings contribute to advancing data-driven methods for hazard assessment in mining areas and provide a foundation for developing early warning and risk-mitigation tools supporting sustainable land management in regions affected by underground mining.

Contributors

Authors

  1. Alicja Szmigiel Strata Mechanics Research Institute Polish Academy of Sciences, Poland ORCID 0000-0001-6770-0421
  2. Krzysztof Tajduś Strata Mechanics Research Institute Polish Academy of Sciences, Poland ORCID 0000-0003-2014-0900
  3. Janusz Rusek AGH University of Krakow ORCID 0000-0003-0368-2580

Index terms

Keywords

Full text

Attachments

Works cited

References

  1. Bianchini S, Confuorto P, Intrieri E, et al (2022) Machine learning for sinkhole risk mapping in Guidonia-Bagni di Tivoli plain (Rome), Italy. Geocarto Int 37:16687–16715. https://doi.org/10.1080/10106049.2022.2113455
  2. Bilgilioğlu SS, Gezgin C, Iban MC, et al (2025) Explainable Sinkhole Susceptibility Mapping Using Machine-Learning-Based SHAP: Quantifying and Comparing the Effects of Contributing Factors in Konya, Türkiye. Applied Sciences 15:3139. https://doi.org/10.3390/app15063139
  3. Chen T, Guestrin C (2016) XGBoost. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. ACM, New York, NY, USA, pp 785–794
  4. Didier C (2009) Postmining Management in France: Situation and Perspectives. Risk Analysis 29:1347–1354. https://doi.org/10.1111/j.1539-6924.2009.01258.x
  5. GHARBI A, AYARI M, Touati Y El, et al (2025) Prediction of Mining-Induced Subsidence in Saudi Arabia Phosphate Mines Using ANN Method. International Journal of Advanced Computer Science and Applications 16:. https://doi.org/10.14569/IJACSA.2025.0160852
  6. Kim Y, Lee SS (2020) Application of Artificial Neural Networks in Assessing Mining Subsidence Risk. Applied Sciences 10:1302. https://doi.org/10.3390/app10041302
  7. Konieczny K, Słowik L (2019) An Example of Failure of an Office Building in Upper Silesia. MATEC Web Conf 284:3003. https://doi.org/10.1051/matecconf/201928403003
  8. Kretschmann J, Efremenkov AB, Khoreshok AA (2017) From Mining to Post-Mining: The Sustainable Development Strategy of the German Hard Coal Mining Industry. IOP Conf Ser Earth Environ Sci 50:012024. https://doi.org/10.1088/1755-1315/50/1/012024
  9. Kumar Singh S, Chakravarty D (2023) Efficient and Reliable Prediction of Dump Slope Stability in Mines using Machine Learning: An in-depth Feature Importance Analysis. Archives of Mining Sciences. https://doi.org/10.24425/ams.2023.148157
  10. LE TTT, NGUYEN TG, PHAM C Van, et al (2024) Prediction of Underground Mine’s Surface Subsidence using a Recursive Multi-Step Forecasting Model with an Artificial Neural Network. InżynieriaMineralna 2:. https://doi.org/10.29227/IM-2024-01-116
  11. Prokhorenkova L, Gusev G, Vorobev A, et al (2018) CatBoost: unbiased boosting with categorical features. In: Proceedings of the 32nd International Conference on Neural Information Processing Systems. Curran Associates Inc., Red Hook, NY, USA, pp 6639–6649
  12. Roman Ścigała, Katarzyna Szafulera, Marek Jendryś (2023) Assessment of sinkhole hazard in the post-mining area using the ERT method and numerical modeling. 75 Scientific Journals of the Maritime University of Szczecin 147:. https://doi.org/10.17402/570
  13. Szmigiel A, Apel DB (2022) Predicting the stability of open stopes using Machine Learning. Journal of Sustainable Mining 21:241–248. https://doi.org/10.46873/2300-3960.1369
  14. Szmigiel A, Tajduś K (2025) A Comprehensive Review of Sinkhole Modeling and Prediction Methods. Archives of Mining Sciences 293–312. https://doi.org/10.24425/ams.2025.154664
  15. Taheri K, Shahabi H, Chapi K, et al (2019) Sinkhole susceptibility mapping: A comparison between Bayes‐based machine learning algorithms. Land Degrad Dev 30:730–745. https://doi.org/10.1002/ldr.3255
  16. Zhu J, Pierskalla WP (2016) Applying a weighted random forests method to extract karst sinkholes from LiDAR data. J Hydrol (Amst) 533:343–352. https://doi.org/10.1016/j.jhydrol.2015.12.012