Paper Title
Forecasting A South-West Monsoon Onset using Neural Networks

This paper presents the performance of an artificial neural network to forecast the arrival of South West monsoon to Sri Lanka. Ground level precipitation data of the wet zone and cloud shape stability time extracted from satellite images were the inputs of the network. A feed-forward back propagation network was used to predict the onset within the potential period of April, May and June. Output of the network is either 1 or 0 which represents whether a particular day satisfies the monsoon condition or not. A predetermined consecutive number of such occurrences appear is taken as onset day of the monsoon. The network was trained with the onset determined by the Meteorological Department of India which is available in a public web site. Data for the period from 2012 to 2016 were used to train the network and the data in 2017 and 2018 and also in previous years were used to test the network. The proposed system is able to predict the monsoon situation with an accuracy of 90.26± 10.49%. Keywords- Neural Networks, Cloud Shape Stability, Precipitation, Monsoon Onset