Classification of Groundwater Level Data Using SOM to Develop ANN-Based Forecasting Model

464 Abstract— Prediction of groundwater level in a watershed plays a crucial role in management of groundwater resources, especially in a semi-arid area where there is immense need to groundwater resources in order to prepare the requirement water for agriculture, municipal and industrial affairs. The aim of this study is to present a mathematical based model to estimate the groundwater level (GWL) in Ardabil located at northwest of Iran, with association of some hydrological data (e.g., rainfall, discharge, etc.). In this way identifying various zones with similar groundwater level can be a promising idea which leads to appropriate overview on water table of the study area as well as efficient modeling. For this purpose, the Self Organizing Map (SOM) was used to cluster the homogenous monitoring piezometers in the plain by utilizing GWL and Universal Transverse Mercator (UTM) data. The sensitivity analysis was performed over normalized and non-normalized data of GWL and UTM in order to investigate their effects on clustering. Conventional K-Means method was applied to verify the results of SOM method. The central piezometer of each cluster was selected as a representative by means of statistical technique. Afterwards the three layer feed forward Artificial Neural Network (ANN) model was utilized to calibrate a model via historical groundwater level records from the representative wells and relevant hydro-meteorological data. The last step was performed by simulating water table level of the representative piezometer from each zone of the plain via proposed model, to compare the computed and observed data. The results reveal the suitability of SOM clustering method with normalized data of GWL and also identify the specific piezometers that the GWL of them can represent the GWL in a particular region. Thus, adequate measures should be devoted on preserving such important monitoring piezometers and reliable data can be obtained from them in order to generalize the GWL data to that specific region. The modeling results can be utilized to frame the corresponding strategies to reduce the monitoring cost and to enhance the cost-effective benefits. The proposed methodology can be referred as a management plan for groundwater resources.

[1]  Vahid Nourani,et al.  Spatiotemporal Groundwater Level Forecasting in Coastal Aquifers by Hybrid Artificial Neural Network-Geostatistics Model: A Case Study , 2011 .

[2]  Teuvo Kohonen,et al.  The self-organizing map , 1990, Neurocomputing.

[3]  A. Dassargues,et al.  Exploratory data analysis and clustering of multivariate spatial hydrogeological data by means of GEO 3 DSOM , a variant of Kohonen ’ s Self-Organizing Map , 2006 .

[4]  S. Lallahem,et al.  On the use of neural networks to evaluate groundwater levels in fractured media , 2005 .

[5]  Sheng-Tun Li,et al.  Clustering spatial-temporal precipitation data using wavelet transform and self-organizing map neural network , 2010 .

[6]  Vahid Nourani,et al.  Semi‐distributed flood runoff model at the subcontinental scale for southwestern Iran , 2007 .

[7]  Juan B. Valdés,et al.  NONLINEAR MODEL FOR DROUGHT FORECASTING BASED ON A CONJUNCTION OF WAVELET TRANSFORMS AND NEURAL NETWORKS , 2003 .

[8]  Paulin Coulibaly,et al.  Groundwater level forecasting using artificial neural networks , 2005 .

[9]  Ji-hoon Kim,et al.  Application of cluster analysis for the hydrogeochemical factors of saline groundwater in Kimje, Korea , 2003 .

[10]  Ping Li,et al.  Application and comparison of two prediction models for groundwater levels: a case study in Western Jilin Province, China. , 2009 .

[11]  G. Visconti,et al.  A Neural Network Approach to Real-Time Rainfall Estimation for Africa Using Satellite Data , 2003 .

[12]  M. V. Velzen,et al.  Self-organizing maps , 2007 .

[13]  Lu-Hsien Chen,et al.  Application of Integrated Back-Propagation Network and Self-Organizing Map for Groundwater Level Forecasting , 2011 .

[14]  Amir Jalalkamali,et al.  Groundwater modeling using hybrid of artificial neural network with genetic algorithm , 2011 .

[15]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[16]  J. Adamowski,et al.  A wavelet neural network conjunction model for groundwater level forecasting , 2011 .

[17]  Robert H. Weisberg,et al.  A Review of Self-Organizing Map Applications in Meteorology and Oceanography , 2011 .

[18]  Vahid Nourani,et al.  Integrated artificial neural network for spatiotemporal modeling of rainfall-runoff-sediment processes. , 2010 .

[19]  Vahid Nourani,et al.  An ANN‐based model for spatiotemporal groundwater level forecasting , 2008 .