Prediction Of Monthly Rainfall In Kirkuk Using Artificial Neural Network And Time Series Models

Abstract

One of the major problems in water resources management is the rainfall forecasting. With the effect of rainfall on water resources as a foregone conclusion, more accurate prediction of rainfall would enable more efficient utilization of water resources and power generation. On the other hand, climate and rainfall are highly non-linear and complicated phenomena, which require non-linear mathematical modeling and simulation for accurate prediction. One of the non-linear techniques being recently used for rainfall forecasting is the Artificial Neural Networks (ANN) approach which has the ability of mapping between input and output patterns without a prior knowledge of the system being modeled. In this study, three rainfall prediction models were developed and implemented based on past observations such as time series models based on autoregressive integrated moving average (ARIMA),Artificial Neural Network ANN model and Multi Linear Regression MLR model. A Feed Forward Neural Network FFNN model was applied to predict the rainfall on monthly basis. In order to evaluate the performance of three models, statistical parameters were used to make the comparison between these models. These parameters include the correlation coefficient (R) and Root Mean Square Errors(RMSE). The data set that has been used in this study includes monthly measurements for the rainfall, mean temperature, wind speed and relative humidity from year 1970 to 2008 for Kirkuk station. The models were trained with (25 years) of monthly rainfall data. The ANN, ARIMA and MLR approaches are applied to the data to derive the weights and the regression coefficients respectively. The performances of the models were evaluated by using remaining (13 years) of data. By comparing R2 values (0.91, 0.85, and 0.823) of the models, the study reveals that ANN model can be used as an appropriate forecasting tool to predict the monthly rainfall, which is preferable over the ARIMA model and MLR model.