Forecasting of Maximum and Minimum Monthly Averages of Temperature in Baghdad City Using Multi-Layer Neural Network

Abstract

The research aims to forecast the averages of maximum and minimum temperature averages for Baghdad synoptic station by using the multi-layered neural network. The research deals with two major problems , First , determining the number of input nodes , second , determining the hidden nodes in the layer . The results showed the importance and qualification of the network through the proper choosing for number of input nodes and number of hidden nodes in the layer, by using the statistical procedures AIC , BIC , and ALCc . Then the maximum and minimum temperature averages for Baghdad station has predicted for 2013 by using the previous statistical procedures and compared its results with the real measured data for the station , by using the MSE , MDE , and PMC statistical procedures, and then select the best model to forecast of temperatures averages for 2014 .