Evaluation of Bayesian Networks Model in Monthly Groundwater Level Prediction (Case study: Birjand Aquifer)

The planning of water resources is based on the volume of water extracted from the aquifer and accurate estimate of this volume considerably helps to development. In this study, the Bayesian networks model using continues and clustering structures was used to simulate the groundwater level of Birjand aquifer. Bayesian networks was calibrated with five input variables of aquifer recharge, water table, temperature, evaporation as well as groundwater withdrawals in the previous month and the groundwater level in the current month as output variable. In continues and clustering scenarios, analysis and calibration of input data is performed based on continuity and uncertainty of variables and some validation indexes respectively and then groundwater level was simulated. The final results showed that the Bayesian network is a powerful tool for simulation of groundwater level under uncertainty and average correlation coefficient in 13 piezometers is 0.83 and 0.56 for continues and clustering structures, respectively. Also it shows that continues structure can be applied to predict the groundwater level with higher correlation.