Modeling Temperature prediction Using the Wavelet and Neural Networks Approach: A Case Study of Chongqing Municipality in Southwestern China

The temperature coefficient plays important role in many fields. This study takes into account Chongqing Municipality of China as a typical case. In order to predict the future temperature more exactly, the wavelet and neural networks method are used to obtain more valuable information based on the limited historical temperature data in Chongqing. Main steps are as follows: The first step, decomposed signals under different scales are obtained by using wavelet multi-resolution analysis. The temperature data of Chongqing from 1951 years to 2009 years are used as the training set. They are first normalized and then utilized as input data for the artificial neural network The second step, we carry out temperature signal de-noising by using wavelet transform. The third step, neural networks are used to predict the temperature signal. The method is tested by the temperature data of 2010 whole year The results show that the predicted results are in good agreement with the actual data. So this method is reliable and useful in the temperature's prediction.