A Modular Neural Network Architecture for Rainfall Estimation

J. McCullagh (Australia)


Modular Neural Networks, Ensembles, RainfallEstimation


Assessment of global climate change is a critical research area for the future of man and his environment. Rainfall estimation is one of the key parameters in this research. Artificial Intelligence (AI) techniques including Artificial Neural Networks (ANNs) have been used by researchers to attempt to improve the estimation of rainfall. However the success of these methods has been limited. A single multi-layered back propagation neural network used on complex problems involving different sub-tasks will often show strong inter sub-task interference effects that lead to slow learning and poor generalisation. Modular Neural Networks (MNNs) provide an extension to the concept of ANNs in that they involve breaking down a complex task into a number of simpler tasks, and solving each of these individually. This research explores the application of MNNs to the rainfall estimation problem. The results demonstrate that an improvement in performance can be achieved by the application of MNNs to this problem domain.

Important Links:

Go Back