The Evaluation of the Performance of SARIMA Time Series Model in the Simulation of the Average Monthly Discharge of the Kashkan Afrineh and Kakareza Rivers (Lorestan province)

The importance of planning and managing water resources, as well as an increasing population growth and the limitation of surface water resources in the country, has made the accurate prediction of rivers' flow by using modern tools and methods of modeling, as an inevitable necessity. In addition, proper river flow prediction in river management, flood warning systems, and especially planning for optimal operation is required. In order to predict the flow of a river, several methods have been developed over the past years. In general, these methods can be classified into two categories of conceptual models and models based on statistics or data. The basis of the most of the predictive methods is the simulation type of the current status of the system which is called "modeling". Considering that in most cases the conceptual models require accurate data and knowledge of processes affecting the phenomenon, and this has so far been accompanied by many problems, researchers have turned to the statistical models. Over the past four decades, time-series models have been widely used in predictive river flow as a statistical model. In each science, the collected statistics corresponding to the variable which is going to be predicted and which is available in the past periods are called time series. Indeed, a time series is a set of statistical data collected at equal and regular intervals. Statistical methods that use such statistical data are called analytical methods of time series. The basis of many decisions in hydrological processes and decisions on exploitation of water resources is according to the prediction and analysis of time series. Assessment of the temporal changes of base flow in watersheds, particularly in low flow seasons is very important. Time series models are represented in three main forms: self-correlated models (AR), moving average (MA) models, and self-correlated and moving average (ARMA) models. The condition of using these models is the static nature of the used data. If the data is not static, the data series must be static with the existing methods. The existence of "I" in ARIMA indicates the non-static nature of the original data and the change in the data for modeling. If the data series has a cyclic and rotational state, the type of model is seasonal or SARIMA. Time series models have two components of (p, d, q) and s (P, D, Q). S (P, D, Q) is a seasonal component. P and q are respectively autoregressive parameters and non-seasonal moving average. P and Q are autoregressive parameters and seasonal moving average. The other parameters, D and d, are differential parameters for making the time series static. The statistical and probabilistic models have been presented and developed. This study aimed to analyze and compare the performance of series 30 and 56 years and monthly average discharge related to the Kakareza River in the Selsele city and the Kashkan Afrineh River in the Poldokhtar city in Lorestan province. To this end, the first climate in this region was determined. Next, the autocorrelation function and partial autocorrelation real data draws in XLSTAT software was done. Subsequently, the data was normalized using the Box-Cox and logarithmic. Then, the data trend that indicated non-stationary was determined. After that by using the different operator in MINITAB software, the data trend was removed and the suitable model with the lowest Akaike was selected. Then both periods 12 and 24 months for the two regions were simulated. Results showed that the selected models in 12 and 24 months periods had respectively a correlation coefficient of .92 and .86 for the kakareza river and .94, .88 for the Kashkan Afrineh river. Discussion and conclusion: The most significant difference between the observed and the simulated values is in two months of Esfand and Farvardin. In addition, due to high precipitation, there was a significant increase in the amount of discharge in Farvardin. According to the climatic conditions in the study areas, the model showed a better performance in semi-arid areas.