Prediction of Monsoon Rainfall Using Large Scale ClimateSignals: A Case Study

摘要

In this paper, a comprehensive investigation on application Artificial NeuralNetworks in long-term rainfall forecasting is presented. A variety of ANN-modelshave been developed to process and train a system with large scale climate signals forthe summer precipitation spells. The summer monsoon is one of the most dynamicclimate systems that controls rainfall variation in some Asian countries such as Indiaand Pakistan and delivers a component of annual rainfall in these regions. The sum ofrainfall during July, August and September is called summer monsoon rainfall.In order to quantify the effects of large scale climate signals on the precipitation in thestudy area, long-term records of SST (Sea Surface Temperature) and SLP (Sea LevelPressure) over Oman Sea, Arabian Sea and Indian Ocean have been examined. TheSSTs over west coast of India as well as the SSTs over the Oman Sea shows a highcorrelation with monsoon rainfall in Iran. Also SLP over the Northern part of Indiashows a significant correlation with recorded rainfall at different points of the region.These signals can be used as useful predictors for monsoon rainfall at the southeasternpart of Iran. The results show that considering a set of predictors developed inthis study could significantly increase the accuracy of long-lead precipitationforecasting in the study area using ANN models.