Analysis of rainfall depth based on climatology conditions using artificial neural networks

Abstract

The quantity of rain that falls on the earth cannot be known with certainty. Floods and droughts due to a small quantity of rainfall are frequent events in some areas of Indonesia. The depth of rainfall at a certain time can be anticipated with accurate information. Along with rapid advances in technology, the forecasting of patterns of rainfall can be performed by artificial intelligence models, using historical data for the climatological parameters. The aim of this study is to predict rainfall depth based on climatology data. There are three categories of data that were obtained using NeuroSolutions for Excel: monthly, daily and hourly data. The input data are temperature, pressure, duration of sunshine, and humidity. The output data is rainfall depth. Based on the results of running calculations on monthly, daily, and hourly data, it was indicated that monthly, daily, and hourly data have relative errors of 11.49%, 8.49%, and 19.32% respectively.