Estimation of average monthly rainfall with neighbourhood values: comparative study between soft computing and statistical approach

Abstract

In this study, we demonstrate how connectionist models, in particular, multilayer perceptron network can be used for prediction of rainfall. Here we give a comparative study between conventional approach (using multivariate linear regression) and soft computing approach using artificial neural network (ANN). The basic idea is to identify a computational model to characterise the relation between the average monthly rainfalls of a region with that of different neighbouring regions. The model exploits both spatial as well as temporal information to achieve better prediction. Once the computational model is obtained, it is used to predict the average monthly rainfall. Early prediction of rainfall is expected to play a key role in economic planning.