Cluster Modeling of Groundwater Level of Tabriz Plain Using ARIMA Model

The study of temporal and spatial variations and the estimation of groundwater fluctuations in researching and planning water resources management are important for continuation or development of exploitation.So far, various models have been used to estimate the groundwater level parameter. Stochastic models have been used more than the other models for hydrological studies. The most common stochastic model is autoregressive integrated moving average model (ARIMA). This model is based on the Markov chain theory.In previous studies, one or a few wells have been studied using stochastic models but in this research, 46 selected stations with the monthly data of the Tabriz catchment were used to analyze temporal and spatial variations of groundwater level.In this case study, there was a good correlation between the values of the groundwater level of wells near each other than the other wells, thus to avoid the effects of closed wells fluctuations, at first the whole stations are divided into seven sections using the clustering process, thenspatial studies done just for selected station in every section. The results indicate that the first, third and seventh clusters didn't have acceptable accuracy in simulation due to the throw data and four other clusters had good accuracy. Also, the second and first clusters had the best and worst modeling with maximum R-Square and minimum RMSE, respectively.The use of the Arima model on various stations in the Tabriz catchment area revealed that the accuracy of the model is reduced when the time series have more fluctuations with trends in the mean and variance that lead to unsteady series.