In this research, a simulation–optimization (S/O) model is used in order to estimate aquifer parameters on two aquifers. In this model, meshless local Petrov–Galerkin (MLPG) is used for simulation purpose and modified teaching–learning-based optimization (MTLBO) algorithm is engaged as optimization model. Linking these two powerful models generates a S/O model named MLPG-MTLBO. The proposed model is applied on two aquifers: a standard and a real field aquifer. In standard aquifer, parameters are only transmissivity coefficients in x and y direction for three zones. The acquired results by MLPG-MTLBO are really close to true values. This fact presents the power of MLPG-MTLBO inverse model. Therefore, it is applied on field aquifer. Unconfined aquifer of Birjand recognized as real case study. Parameters which are needed to be estimated are specific yields and hydraulic conductivity coefficients. These parameters are computed by MLPG-MTLBO and entered to the groundwater flow model. The achieved groundwater table compared with observation data and RMSE is calculated. RMSE value is 0.356 m; however, this error criterion for MLPG and FDM is 0.757 m and 1.197 m, respectively. This means that estimation is precise and makes the RMSE to reduce from 0.757 to 0.356 m, and also, MLPG-MTLBO is an accurate model for this aim.

This article investigates the effect of different lengths on prediction accuracy. For this purpose, 732 monthly data of streamflow of the Kortian gauging station at the Kortian Stream Watershed were used. To study the impact of the type of data in terms of monthly or seasonal observation data on the accuracy of modeling results, monthly data were converted into seasonal data and the results of monthly and seasonal modeling were compared. Therefore, multiplicative ARIMA models were performed for the monthly and seasonal modeling. Compared with the seasonal modeling, the monthly modeling presented more precise results that the sum of the squares errors of monthly and seasonal modeling were 0.9408 and 2.5 respectively. For the monthly modeling, 5 different lengths of data were used. C1 model used the last 60 data, C2 used the last 120 recorded observations, C3 used the last 240 data, C4 used the last 480 observations and C5 used the last 708 data. To test the precision of models, 24 observations were held out. Among C1 to C5 models, the C4 model presented the best results in predicting 2 years ahead and C1 had the worst results.