Forecasting Maximum Temperatures via Fuzzy Nearest Neighbour Model over Delhi

This paper introduces nearest neighbour based fuzzy model (NNFM) based on membership values for forecasting the daily maximum temperature at Delhi . Fuzzy membership values has been used to make single point forecasts into the future on the basis of past nearest neighbours. Compared with other statistical method and artificial neural network (ANN) technique, this approach has the advantages of faster and highly automated model synthesis as well as improved prediction and forecasting accuracies. The NNFM model developed using daily temperature data of 15 years (1991 to 2005) that were used to predict the maximum temperature for the next days. The model forecast has been tested with the actual observation of the independent data sets of 2006 and also the results are compared with the method of persistence and ANN technique. The overall performance statistics shows that the proposed model is capable to produce good results with independent cases, providing about 91% correct forecast within ±2◦Cof the observed values for the next day along with a positive value of skill score over persistence method as well as ANN technique. The method is found to be promising for operational application.