An Improved Rainfall Prediction Model for Minimizing the Negetive Impact of Increased Rainfall Days in Minna, Nigeria

Abstract

This research aim at predicting the rainfall of Minna metropolis of Niger State Nigeria using using classification method of ANN. In this approach, four atmospheric parameters comprising thaose of rainfall, relative humidity, minimum temperature and maximum temperature spanning from January 2010 to December, 2019 were acquired from the Geography department of the Federal University of Technology, Minna. the default threshold classification method of ANN was investigated. the result reveal that for a default threshold of 0.5, a prediction accuracy of 69%, sensitivity of 63%, specificity of 84%, an error value of 1.3% and a total of 66 rainfall days were predicted at against 32 rainfall days in the data set. the implication of this result is that more rainfall days were anticipated in the metropolis which could lead to flooding in a long run. it was recommended that for more acurate rainfall prediction, more robust data be used for network training