Abstract
Effective characterization of lithology is vital for the conceptualization of complex aquifer systems, which is a prerequisite for the development of reliable groundwater-flow and contaminant-transport models. However, such information is often limited for most groundwater basins. This study explores the usefulness and potential of a hybrid soft-computing framework; a traditional artificial neural network with gradient descent-momentum training (ANN-GDM) and a traditional genetic algorithm (GA) based ANN (ANN-GA) approach were developed and compared with a novel hybrid self-organizing map (SOM) based ANN (SOM-ANN-GA) method for the prediction of lithology at a basin scale. This framework is demonstrated through a case study involving a complex multi-layered aquifer system in India, where well-log sites were clustered on the basis of sand-layer frequencies; within each cluster, subsurface layers were reclassified into four depth classes based on the maximum drilling depth. ANN models for each depth class were developed using each of the three approaches. Of the three, the hybrid SOM-ANN-GA models were able to recognize incomplete geologic pattern more reasonably, followed by ANN-GA and ANN-GDM models. It is concluded that the hybrid soft-computing framework can serve as a promising tool for characterizing lithology in groundwater basins with missing lithologic patterns.
Résumé
La caractérisation effective de la lithologie est indispensable pour la conceptualisation des systèmes aquifères complexes, un préalable au développement de modèles fiables d’écoulement d’eau souterraine et de transport de polluants. Toutefois, une telle information est souvent limitée pour la plupart des réservoirs souterrains. Cette étude explore l’utilité et le potentiel d’une structure de logiciels hybrides de calcul; un réseau neuronal artificiel classique avec l’apprentissage d’un gradient dynamique descendant (ANN-GDM) et un algorithme génétique classique (GA) basé sur approche ANN (ANN-GA) ont été développés et comparés à une carte originale hybride auto-configurée (SOM) basée sur la méthode ANN (SOM-ANN-GA) pour prédire la lithologie à l’échelle d’un bassin. Cette structure de calcul est utilisée dans une étude de cas mettant en jeu un système aquifère multicouches complexe en Inde, où des logs de forage ont été groupés sur la base de la fréquence des couches sableuses; dans chaque groupe, les couches de subsurface ont été reclassées en 4 classes sur la base de la profondeur maximale de forage. Des modèles ANN correspondants à chaque classe de profondeurs ont été développés en utilisant chacune des trois approches. Parmi les trois, les modèles hybrides SOM-ANN-GA ont été capables de reconnaître des structures géologiques incomplètes de façon assez satisfaisante, suivis par les modèles ANN-GA et ANN-GDM. En conclusion, la structure de logiciels hybrides de calcul peut servir comme un outil prometteur pour caractériser la lithologie de réservoirs souterrains avec des caractéristiques lithologiques manquantes.
Resumen
La caracterización efectiva de la litología es vital para la conceptualización de los sistemas acuíferos complejos, lo cual es un prerequisito para el desarrollo de modelos de flujo confiables del agua subterránea y de transporte de contaminantes. Sin embargo, dicha información está a menudo limitada para la mayoría de las cuencas de agua subterránea. Este estudio explora la utilidad y el potencial de un marco híbrido de un software computacional; se desarrolló un enfoque basado en una red neuronal artificial con ensayos del gradiente del momento descendente (ANN-GDM) y algoritmos genéticos tradicionales (GA) en base a ANN (ANN-GA) y se comparó con un método novedoso basado en mapa híbrido autoorganizado (SOM) a partir de ANN (SOM-ANN-GA), para la predicción de la litología en una escala de cuenca. Este marco se demostró a través de un caso de estudio que involucró a un sistema acuífero multicapa complejo en la India, donde los sitios de registros de pozos se agruparon sobre la base de las frecuencias de las capas de arena; dentro de cada grupo, las capas del subsuelo fueron reclasificadas en cuatro clases de profundidad basada en la máxima profundidad de las perforaciones. Se desarrollaron modelos ANN para cada clase de profundidad utilizando cada uno de las tres aproximaciones. De las tres, los modelos SOM-ANN-GA híbridos fueron capaces de reconocer patrones geológicos incompletos razonablemente, seguido por los modelos ANN-GDM ANN-GA. Se concluye que el marco del software computacional híbrido puede servir como una herramienta prometedora para caracterizar la litología en las cuencas de aguas subterráneas con patrones litológicos faltantes.
摘要
岩性的有效描述对于复杂含水层系统的概念化至关重要,而复杂含水层系统的概念化是建立可靠的地下水流和污染物运移模型的先决条件。然而,对于大多数地下水流域来说,这样的信息常常有限。本研究探索了一个混合柔性计算框架的有效性和潜力;建立了具有梯度下降—势头培训的传统人工神经网络(ANN-GDM)及基于人工神经网络(ANN-GA)方法的传统遗传算法(GA),并与根据流域尺度预测岩性的人工神经网络 (SOM-ANN-GA)方法得到的新的混合自组织图(SOM)进行了比较。通过印度一个复杂的多层含水层系统的研究实例展示了这个框架,在这个含水层系统中,测井地点根据砂层频率聚集一起;在每一个聚集群中,地表以下地层根据最大钻孔深度重新分为四个深度级别。采用三种方法的每种方法建立了每个深度级别的人工神经网络模型。三个模型中,混合的SOM-ANN-GA模型能够更加合理地识别不完全的地质模式,其次是ANN-GA模型和ANN-GDM模型。可以得出的结论就是,混合的柔性-计算框架可以作为一个有希望的工具,用来描述缺失岩性模式的地下水流域的岩性。
Resumo
A caracterização eficaz da litologia é vital para a conceituação de sistemas aquíferos complexos, sendo um pré-requisito para o desenvolvimento de modelos confiáveis de fluxo de águas subterrâneas e transporte de contaminantes. No entanto, tal informação é muitas vezes limitada para a maioria das bacias de águas subterrâneas. Este estudo explora a utilidade e o potencial de um quadro híbrido de computação flexível; uma rede neural artificial tradicional com treinamento do momento descendente do gradiente (ANN-GDM) e um algoritmo genético tradicional (ANN-GA) baseado na abordagem de redes neurais (ANN) foram desenvolvidos e comparados com um novo mapa híbrido original auto-organizavel (SOM) com base na abordagem de ANN (SOM-ANN-GA) para a predição da litologia em uma escala de bacia. Este arcabouço é demonstrado através de um estudo de caso envolvendo um sistema aquífero complexo multicamadas, na Índia, onde registros locais de poços foram agrupados com base em frequências de camada de areia; dentro de cada agrupamento, camadas do subsolo foram reclassificadas em quatro classes de profundidade com base na profundidade máxima de perfuração. Dos três, os modelos híbridos SOM-ANN-GA foram capazes de reconhecer um padrão geológico incompleto mais razoavelmente, seguido pelos modelos ANN-GA e ANN-GDM. Conclui-se que o quadro híbrido de computação flexível pôde servir como uma ferramenta promissora para caracterização da litologia em bacias de águas subterrâneas com padrões litológicos faltantes.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Aristodemou E, Pain C, Oliveira C, Goddard T (2005) Inversion of nuclear well-logging data using neural networks. Geophys Prospect 53:103–120
ASCE (ASCE Task Committee on Application of Artificial Neural Networks in Hydrology) (2000) Artificial neural networks in hydrology, part I: preliminary concepts. J Hydrol Eng ASCE 5(2):115–123
Aziz ARA, Wong KFV (1992) A neural‐network approach to the determination of aquifer parameters. Ground Water 30(2):164–166
Beale R, Jackson T (1990) Neural computing: an introduction. CRC, Boca Raton, FL
Benaouda D, Wadge G, Whitmarsh RB, Rothwell RG, MacLeod C (1999) Inferring the lithology of borehole rocks by applying neural network classifiers to downhole logs: an example from the ocean drilling program. Geophys J Int 136(2):477–491
Bhuiyan M (2001) An intelligent system’s approach to reservoir characterization in cotton valley. MSc Thesis, West Virginia University, Morgantown, WV, USA
Camp CV, Outlaw JE, Brown MC (1994) GIS-based groundwater modeling. Microcomput Civ Eng 9(4):281–293
Castillo PA, Merelo JJ, Prieto A, Rivas V, Romero G (2000) G-Prop: global optimization of multilayer perceptrons using GAs. Neurocomputing 35:149–163
Chang H, Chen H, Fang J (1997) Lithology determination from well logs with fuzzy associative memory neural network. IEEE Trans Geosci Remote Sens 35:773–780
Chikhi S, Batouche M, Shout H (2005) Hybrid neural network methods for lithology identification in the Algerian Sahara. Int J Comput Intell 1(1):25–33
Das S (2000) Groundwater resources and its management in Mahanadi delta. In: NK Mahalik (ed) Mahanadi Delta: geology, resources and biodiversity, chap 10. AIT, Ahmedabad, India, pp 95–116
Dayhoff JE (1990) Neural network architectures: an introduction. Reinhold, New York
Dubois MK, Bohling GC, Chakrabarti S (2007) Comparison of four approaches to a rock facies classification problem. Comput Geosci 33(5):599–617
Farsadnia E, Rostami Kamrood M, Moghaddam Nia A, Modarres R, Bray MT, Han D, Sadatinejad J (2014) Identification of homogeneous regions for regionalization of watersheds by two-level self-organizing feature maps. J Hydrol 509:387–397
Gangopadhyay S, Gautam TR, Gupta AD (1999) Subsurface characterization using artificial neural networks and GIS. J Comput Civ Eng 13(3):153–160
Garrett JH Jr, Ghaboussi J, Xu X, Ranjithan S (1992) Neural networks and their application to civil engineering. In: Allen RH (ed) Expert systems in civil engineering: knowledge representation. Expert Systems Committee of ASCE, New York
Gerken WC, Purvis LK, Butera RJ (2006) Genetic algorithm for optimization and specification of a neuron model. Neurocomputing 69:1039–1042
Goldberg DE (1989) Genetic algorithms in search, optimization and machine learning. Addison-Wesley, New York, pp 1–145
Haykin S (1994) Neural networks. Macmillan, New York
Hertz J, Krogh A, Palmer RG (1991) Introduction to the theory of neural computation. Addison-Wesley, Reading, UK
Hornik K, Stinchcombe M, White H (1989) Multilayer feedforward networks are universal approximators. Neural Netw 2:359–366
Hsieh B-Z, Lewis C, Lin Z-S (2005) Lithology identification of aquifers from geophysical well logs and fuzzy logic analysis: Shui-Lin Area, Taiwan. Comput Geosci 31(3):263–275
Jha MK, Sahoo S (2015) Efficacy of neural network and genetic algorithm techniques in simulating spatio-temporal fluctuations of groundwater. Hydrol Process 29(5):671–691
Jha MK, Kumar A, Nanda G, Bhatt G (2006) Evaluation of traditional and non-traditional optimization techniques for determining well parameters from step-drawdown test data by. J Hydrol Eng 11(6):617–630
Jones AJ (1993) Genetic algorithms and their applications to the design of neural networks. Neural Comput Applic 1:32–45
Kohonen T (1998) Self-organization and associative memory. Springer Series in Information Sciences. Springer, Berlin
Leloup JA, Lachkar Z, Boulanger JP, Thiria S (2007) Detecting decadal changes in ENSO using neural networks. Clim Dyn 28:147–162
Lobo VJAS (2009) Application of self-organizing maps to the maritime environment. In: Information fusion and geographic information systems. Springer, Berlin, pp 19–36
Mahalik NK (2000) Stratigraphy, palaeogeography and evolutionary history of Mahanadi delta. In: NK Mahalik (ed) Mahanadi Delta: geology, resources and biodiversity, chap 6. AIT, Ahmedabad, India, pp 53–70
Maier HR, Dandy GC (2000) Neural networks for the prediction and forecasting of water resources variables: a review of modeling issues and applications. Environ Model Softw 15:101–124
Maiti S, Tiwari RK, Kumpel HJ (2007) Neural network modeling and classification of lithofacies using well log data: a case study from KTB borehole site. Geophys J Int 169:733–746
NeuroSolutions (2010) Help version: 6. NeuroDimension, Gainesville, FL
Nguyen TT, Kawamura A, Tong TN, Nakagawa N, Amaguchi H, Gilbuena R (2015) Clustering spatio-seasonal hydrogeochemical data using self-organizing maps for groundwater quality assessment in the Red River Delta, Vietnam. J Hydrol 522:661–673
Nishiyama K, Endo S, Jinno K, Uvo CB, Olsson J, Berndtsson R (2007) Identification of typical synoptic patterns causing heavy rainfall in the rainy season in Japan by a self-organizing map. Atmos Res 83:185–200
Ojha M, Maiti S (2013) Sediment classification using neural networks: an example from the site-U1344A of IODP expedition 323 in the Bering Sea. Deep Sea Research Part II 125:202–213
Pezeshk S, Camp CV, Karprapu S (1996) Geophysical log interpretation using neural network. J Comput Civ Eng ASCE 10(2):136–142
Ranjithan S, Eheart JW, Garett JH Jr (1993) Neural network based screening for groundwater reclamation under uncertainty. Water Resour Res 29(3):563–574
Rizzo DM, Dougherty DE (1994) Characterization of aquifer properties using artificial neural networks: neural kriging. Water Resour Res 30(2):483–497
Rogers SJ, Fang JH, Karr CL, Stanley DA (1992) Determination of lithology from well logs using a neural network. Am Assoc Pet Geol 76(5):731–739
Rolon L, Mohaghegh SD, Ameri S, Gaskari R, McDaniel B (2009) Using artificial neural networks to generate synthetic well logs. J Nat Gas Sci Eng 1:118–133
Rumelhart DE, McClelland JL, the PDP Research Group (1986) Parallel distributed processing explorations in the microstructure of cognition. 1: foundations. The MIT Press, Cambridge, MA
Rushton KR (2003) Groundwater hydrology: conceptual and computational models. Wiley, London
Sahoo S, Jha MK (2013) Groundwater-level prediction using multiple linear regression and artificial neural network techniques: a comparative assessment. Hydrogeol J 21(8):1865–1887
Singh UK (2011) Fuzzy inference system for identification of geological stratigraphy of Prydz Bay, East Antarctica. J Appl Geophys 75:687–698
Srinivas VV, Tripathi S, Ramachandra Rao A, Govindaraju RS (2008) Regional flood frequency analysis by combining self-organizing feature map and fuzzy clustering. J Hydrol 348:148–166
Statistica (2001) STATISTICA: system reference. StatSoft, Tulsa, OK
Tonn T (2002) Neural network seismic reservoir characterization in a heavy oil reservoir. Lead Edge 21(3):309–312
Wallner M, Haberlandt U, Dietrich J (2013) A one-step similarity approach for the regionalization of hydrological model parameters based on self-organizing maps. J Hydrol 494:59–71
Wong PM, Jiang FX, Taggart IJ (1995) A critical comparison of neural networks and discriminant analysis in lithofacies, porosity and permeability predictions. J Petrol Geol 18:191–206
Acknowledgments
The authors gratefully acknowledge the staff of the Orissa Lift Irrigation Corporation (OLIC), Bhubaneswar, Odisha, for providing necessary lithologic data required for the present study. The help rendered by Madhumita Sahoo, during the initial stage of this study, is gratefully acknowledged. Authors are also thankful to the editor, associate editor and two reviewers for their constructive comments and suggestions.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Sahoo, S., Jha, M.K. Pattern recognition in lithology classification: modeling using neural networks, self-organizing maps and genetic algorithms. Hydrogeol J 25, 311–330 (2017). https://doi.org/10.1007/s10040-016-1478-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10040-016-1478-8