فهرست مطالب محبوبه یونسی

Time series analysis is divided into two general categories, parametric and nonparametric methods. In this study, the monthly rainfall data in a 35-year statistical period of kakarreza hydrometric station (located in Khorramabad city) was tested by the parametric and nonparametric method for periodic Autoregressive (AR) with constant and variable coefficients individually. Also, winter method and decomposition were tested and evaluated. Regarding the fact that the mean and standard deviations have not the significant difference in parametric and non-parametric models. The results of modeling are almost identical in two situations and in a constant coefficient situation, the autoregressive model with order 2 gives a slightly better result (R2=0.534, ACI=-446 and RMSE=6.07). Generally, periodic autoregressive model with variable coefficients has better performance than autoregressive model with constant coefficients. Analysis of the seasonal decomposition model shows that the residual series are non-random. Also in the winter’s model, the graph of residuals versus fitted values exhibits a specific structure and the graph has a trend, so it cannot be assumed that the residual variance is a constant. As a result, these models do not have the capability to estimate river flow.

Drought prediction plays an important role in the planning and management of natural resources, water resources and plant water requirements. Wavelet transform is one of the new and highly effective methods for analysing signals and time series. Using the mother wavelet, the standard precipitation index (SPI) signal was analyzed and the results were considered as inputs of the artificial neural network models and the gene expression programming (GEP). Multi-layer perceptron (MLP), radial basis function (RBF), (GEP), as well as the wavelet-artificial neural networks integrated model and multi-layer perceptron (WA-MLP), radial basis function (WA- RBF) and wavelet- gene expression programming (WA- GEP) were used for drought forecasting. The rainfall data collected at the Bidestan Station for a period of 44 years were used on the Shoor Watershed in the Province of Qazvin. Moisture condition was calculated using the SPI in the short-term period of 3 months. To estimate the SPI in each period, the respective amounts were considered from the previous cycles. The results showed that among the six applied models, the WA-GEP predicted the SPI values and the short-term drought condition with a higher accuracy. The WA-GEP model proved to be the best scenario in the validation stage of R2, RMSE, MAE and NS of 0.911, 0.037, 0.022 and 0.845, respectively.

Prediction of groundwater level fluctuations is an essential for planning in arid and semi-arid regions. In this study, artificial neural network models, ARIMA time series and multivariate linear regression models were used to predict groundwater fluctuations of two piezometers located in Kerman plain. In order to achieve this goal, the depth of groundwater of the monthly piezometers was used during the years 2002-2013. The results of studying different models of ARIMA model showed that the ARIMA (0,1,1) and (2,0,2) for south Baghin piezometer and ARIMA (1,1,1) and (2,0,0) for Airport areas piezometer are the best-fit time series model with the data. In the model of MLP and RBF artificial neural network, MLP with 2 and 4 layers of hidden and RBF with 8 and 10 hidden layers for southern Baghin piezometers and the airport areas have the best fit with the data. In multivariate linear regression modeling, for each of the two piezometers, the best correlations of the multivariable linear regression model show that the multivariate linear regression relationship of groundwater depth of the current month is a function of groundwater depth of one month prior; in other words, the depth of water the ground water level has the highest dependence on groundwater depth of its prior month. The results showed that prediction of groundwater depth by multivariate linear regression model is better than Neural Network and ARIMA model.

Drought prediction in water resources systems plays an important role in reducing drought damage. In recent decades, Traditional methods including: fitting and mathematical models have been widely used to predict droughts. The combination of wavelet theory and neural networks has led to the expansion of the wavelet-neural networks. The application of the wavelet as training function in the neural network has recently been identified as a substitute method in neural networks. In these models, the position and scale coefficients of the wavelets are optimized in addition to the weights (Thuillard, 2000). Considering the importance of short-term drought prediction in water resources engineering and the nonlinear characteristics of the SPI series of three months, the purpose of this study is to present an Artificial Wavelet Neural Networks integrated model for predicting short-term drought at Bidestan station in Qazvin plain.
In this research, Multi-Layer Perceptron (MLP), Radial Base Function (RBF), ARIMA time series, as well as Artificial Wavelet Neural Networks integrated model and Multi-layer Perceptron (WA-MLP) and Radial Bonding Function (WA- RBF) were used, which is done by analyzing the time series investigated by the wavelet transformation and the entry of these sub-series into an artificial neural network.
According to previous researches on drought prediction, short-term drought prediction (with the definition of a three-month standard rainfall index) using the combined model of Wavelet-Neural Network and comparing its results with artificial neural network and ARIMA time series models has not been compared. In this paper, five short-term drought prediction models have been compared and a better performance model has been introduced.