Journal of Advanced Informatics in Water, Soil, and Structure

Journal of Advanced Informatics in Water, Soil, and Structure

Analyzing the Uncertainty of Artificial Neural Network Models and Support Vector Machines in Estimating Saturated Hydraulic Conductivity

Document Type : Research Article

Authors
Amir Nabizadeh 1
Abbas Khashei 2
Peyman Afrasiab 1
Mohsen Pourreza Bilondi 2
1 Department of Water Engineering, University of Zabol, Zabol. Iran.
2 Department of Water Engineering, University of Birjand , Birjand . Iran.
10.22048/wss.2025.450399.1011
Abstract
Data-basis simulation models such as artificial neural network (ANN) and support vector machines (LS-SVM) are suitable substitute for expensive and laboratory methods of estimating soil hydraulic properties, including hydraulic saturation conductivity of soil. In this study, the objective is to predict the saturated hydraulic conductivity of soil in the study area, Sistan Dam, using ANN and LS-SVM. To this purpose, the saturated hydraulic conductivity of soil was measured in 112 points by the suction disc method. For selecting input variables among simple characteristics, the multistep regression method was used. In these models, the selection of different transfer and training functions represents a significant source of error. Consequently, a comprehensive analysis was conducted to identify uncertainty sources in simulating the saturated hydraulic conductivity of soil. To determine the best simulation model among neural network functions and support vector machines, a Monte Carlo method was employed to sample and evaluate their performance. This rigorous approach ensures a thorough examination of the strengths and limitations of these models, thereby enhancing the reliability and accuracy of soil conductivity predictions. Finally, the uncertainty in predicting the solution model can be analyzed to determine how much ANN models and support vector machines (LS-SVMs) are reliable. In general, when considering both linear and nonlinear scenarios in ANNs and LS-SVM, linear input methods may not be a suitable approach. This is primarily due to the complex and nonlinear nature of the soil properties and processes being modeled. Therefore, nonlinear input methods are typically preferred to accurately capture the intricate relationships and dynamics of soil systems. This highlights the importance of selecting appropriate input methods to ensure reliable and accurate modelling of soil properties and behaviour. In addition, among the ANN and LS-SVM models used in this study, ANN, with values of content ratio criteria (CR), (with the values of 90, 0.263, and 0.740) high degree of asymmetry index (S) and high and low asymmetry index (T), was of higher certainty and accuracy than the other models. Moreover, Logsig_trainlm and Tansig_trainbfg scenarios in predicting saturated hydraulic conductivity had a satisfactory process and less uncertainty.
Keywords
ANN
LS-SVM
Soil
Suction Disc
Monte Carlo
Evaluation Criteria
Subjects
  1. Akbarzadeh, A., Nouri, R., Farrokh Nia, A., Khakpur, A., Sabahi, M. S. (2000). Analysis of accuracy and uncertainty of intelligent models in predicting longitudinal propagation coefficient of rivers. Quarterly Journal of Water and Sewage, 21(3), 99–107.
  2. Anderson, M. P., Woessner, W. W., & Hunt, R. J. (2015). Applied Groundwater Modeling (2nd ed.). Academic Press.
  3. Ankeny, M. D., Ahmed, M., Kaspar, T. C., & Horton, R. (1991). Simple field method determining unsaturated hydraulic conductivity. Soil Science Society of America Journal, 55, 467–470.
  4. Arrington, K. E., Ventura, S. J., & Norman, J. M. (2013). Predicting saturated hydraulic conductivity for estimating maximum soil infiltration rates. Soil Science Society of America Journal, 77, 748–758.
  5. Bowden, G. J., Dandy, G. C., & Maier, H. R. (2005). Input determination for neural network models in water resources applications, Part 1: Background and methodology. Journal of Hydrology, 301(1), 75–92.
  6. Budiman, M., McBratney, A. B., & Bristow, K. L. (1999). Comparison of different ... [title truncated in original] [Please provide full title and source if available]
  7. Cornelis, W. M., Ronsyn, J., Van Meirvenne, M., & Hartmann, R. (2001). Evaluation of pedotransfer functions for predicting the soil moisture retention curve. Soil Science Society of America Journal, 65, 638–648.
  8. Dehghani, M., Saghafian, B., Nasiri Saleh, F., Farokhnia, A., & Noori, R. (2014). Uncertainty analysis of streamflow drought forecast using artificial neural networks and Monte–Carlo simulation. International Journal of Climatology, 34, 1169–1180.
  9. Eckhardt, K., Breuer, L., & Frede, H. G. (2003). Parameter uncertainty and the significance of simulated land use change effects. Journal of Hydrology, 273(1), 164–176.
  10. Ferrer, J. M., Estrela Monreal, T., Sánchez del Corral Jiménez, A., & García Meléndez, E. (2004). Constructing a saturated hydraulic conductivity map of Spain using pedotransfer functions and spatial prediction. Geoderma, 123, 257–277.
  11. Fuentes, J. P., Flury, M., & Bezdicek, D. F. (2004). Hydraulic properties in a silt loam soil under natural prairie, conventional tillage, and no-till. Soil Science Society of America Journal, 68, 1679–1688.
  12. Gorelick, S. M., & Zheng, C. (2015). Global change and the groundwater management challenge. Water Resources Research, 51(5), 3031–3051.
  13. Horne, A., Szemis, J. M., Kaur, S., Webb, J. A., Stewardson, M. J., Costa, A., & Boland, N. (2016). Optimization tools for environmental water decisions: A review of strengths, weaknesses, and opportunities to improve adoption. Environmental Modelling & Software, 84, 326–338.
  14. Marcé, R., Comerma, M., Garcia, J. C., & Armengol, J. (2004). A neuro-fuzzy modeling tool to estimate fluvial nutrient loads in watersheds under time-varying human impact. Limnology and Oceanography: Methods, 2(11), 342–355.
  15. Merdun, H., Çınar, Ö., Meral, R., & Apan, M. (2006). Comparison of artificial neural network and regression pedotransfer functions for prediction of soil water retention and saturated hydraulic conductivity. Soil & Tillage Research, 90, 108–116.
  16. Merdun, H., Ozer, C., Meral, R., & Apan, M. (2006). Comparison of artificial neural network and regression pedotransfer functions for prediction of soil water retention and saturated hydraulic conductivity. Soil Tillage Research, 90, 108–116.
  17. Alemi, M. H. (1981). Water and soil. Tehran: Tehran University Press. (In Persian)
  18. Moret, D., & Arrúe, J. L. (2007). Dynamics of soil hydraulic properties during fallow as affected by tillage. Soil and Tillage Research, 96, 103–113.
  19. Motaghian, H. R., & Mohammadi, J. (2011). Spatial estimation of saturated hydraulic conductivity from terrain attributes using regression, kriging, and artificial neural networks. Pedosphere, 21(2), 170–177.
  20. Noori, R., Abdoli, M. A., Farokhnia, A., & Abbasi, M. (2009). Results uncertainty of solid waste generation forecasting by hybrid of wavelet transform–ANFIS and wavelet transform–neural network. Expert Systems with Applications, 36(6), 9991–9999.
  21. Noori, R., Khakpour, A., Omidvar, B., & Farokhnia, A. (2010a). Comparison of ANN and principal component analysis–multivariate linear regression models for predicting the river flow based on developed discrepancy ratio statistic. Expert Systems with Applications, 37(8), 5856–5862.
  22. Noori, R., Hoshiyaripour, G. A., Ashrafi, K., & Araabi, B. N. (2010b). Uncertainty analysis of developed ANN and ANFIS models in prediction of carbon monoxide daily concentration. Atmospheric Environment, 44(4), 476–482.
  23. Pachepsky, Y. A., Tim Lin, D., & Varallyay, G. (1996). Artificial neural networks to estimate soil water retention from easily measurable data. Soil Science Society of America Journal, 60, 727–733.
  24. Pourreza Bilandi, Mohsen, Khasiei Sayuki, Abbas. (2015). Uncertainty analysis of the neural network output in simulating saturated hydraulic saturation of soil. Irrigation and Drainage Journal of Iran, 4, 9, 655–664.
  25. Rezaei, E., Khasiei Sayuki, A., Shahidi, A. (2014). Designing groundwater monitoring network using LS-SVM minimum squared model. Iran Water and Soil Research, 45(4), 389–396.
  26. Rogiers, B., Mallants, D., Batelaan, O., Gedeon, M., Huysmans, M., & Dassargues, A. (2011). Estimation of hydraulic conductivity and its uncertainty from grain-size data using GLUE and artificial neural networks. Mathematical Geosciences, 25.
  27. Sobieraj, J. A., Elsenbeer, H., & Vertessy, R. A. (2001). Pedotransfer functions for estimating saturated hydraulic conductivity. Journal of Hydrology and Hydromechanics, 251, 202–220.
  28. Stumpp, C., Engelhardt, S., Hofmann, M., & Huwe, B. (2009). Evaluation of pedotransfer functions for estimating soil hydraulic properties of prevalent soils in a catchment of the Bavarian Alps. European Journal of Forest Research, 128, 609–620.
  29. Wagner, B., Tarnawski, V. R., & Stöckl, M. (2004). Evaluation of pedotransfer functions predicting hydraulic properties of soils and deeper sediments. Journal of Plant Nutrition and Soil Science, 167, 236–245.
  30. White, J. T., Fienen, M. N., Barlow, P. M., & Welter, D. E. (2018). A tool for efficient, model-independent management optimization under uncertainty. Environmental Modelling & Software, 100, 213–221.
  31. Xiong, L., Wan, M., Wei, X., & O’Connor, K. (2009). Indices for assessing the prediction bounds of hydrological models and application by generalized likelihood uncertainty estimation. Hydrological Sciences Journal, 54(5).
  32. Xu, C., Xu, X., Liu, M., Liu, W., Yang, J., Luo, W., Zhang, R., & Kiely, G. (2017). Enhancing pedotransfer functions (PTFs) using soil spectral reflectance data for estimating saturated hydraulic conductivity in southwestern China. Catena, 158, 350–356.
  33. Yao, R. J., Yang, J. S., Wu, D. H., Li, F. R., Gao, P., & Wang, X. P. (2015). Evaluation of pedotransfer functions for estimating saturated hydraulic conductivity in coastal salt-affected mud farmland. Journal of Soils and Sediments, 15, 1–15.
  34. Zhao, C., Shao, M., Jia, X., Nasir, M., & Zhang, C. (2016). Using pedotransfer functions to estimate soil hydraulic conductivity in the Loess Plateau of China. Catena, 143, 1–6.
Journal of Advanced Informatics in Water, Soil, and Structure
Volume 1, Issue 2 - Serial Number 2
September 2025
Pages 187-202

  • Receive Date 10 July 2025
  • Revise Date 20 August 2025
  • Accept Date 10 September 2025
  • First Publish Date 10 September 2025
  • Publish Date 20 September 2025