Introduction and application of transfer function model to predict the process of rainfall-groundwater level in Golestan province

Groundwater is one of the most valuable water resource in Iran, which is mainly influenced by rainfall patterns in a region. In the present study, 30-years rainfall and groundwater level data were used in the three watersheds of Galikesh, Ramian and Mohammadabad to model the rainfall and groundwater level. The prediction of the groundwater level using the rainfall data for the next 12 months was done using the Transfer Function (TF) and using SAS and MINITAB software. Validation of predicted values was done using MAD, RMSE and MAPE indices. The results showed that according to the autocorrelation diagrams, all time series have seasonal trend with a period of 12 months. The results of the autocorrelation diagrams showed that in Galikesh and Mohammadabad watersheds, rainfall has a direct effect on the groundwater level with a delay of three months, but in the Ramian watershed, this delay was one month. It was also found that the transfer function model had a good performance in fitting the monthly groundwater level in all 3 studied watersheds.