Artificial Neural Network Backpropagation for Predicting Rainfall (Case Study in Sultan Muhammad Kaharuddin Meteorological Station)

Abstract

The climate of a region is strongly influenced and determined by the amount of rainfall in the region. Therefore, knowing the results of rainfall predictions in that area will provide an overview of the climatic conditions that will occur so that anticipatory steps can be taken against unwanted possibilities. This study aims to make predictions of rainfall in Sumbawa Regency for 2023 using a Backpropagation Neural Network. The data used is rainfall data from 2008 to 2022 recorded at the Sultan Muhammad Kaharuddin Meteorological Station. Data Processing is carried out in three stages, namely the data normalization stage, the data training stage, and the data testing stage. The prediction results obtained show that the highest rainfall occurs at the beginning and end of the year, namely January, February, March, and April with rainfall of 343,4 mm/month, 421,3 mm/month, 295,2 mm/month, and 134,1 mm/month. Whereas at the end of the year it takes place in November (178.8 mm/month) and December (223.7 mm/month). The precipitation decline graph (start of dry season) starts from May (24.2mm/month) to September (22.1mm/month) with a dry peak in July (2mm/month). Whereas for the end of the year it takes place in November (302,1 mm/month) and December (308,4 mm/month). The decrease in rainfall (early dry season) begins in May (51,7 mm/month) to October (66 mm/month) with the peak of the dry season in June to August (0 mm/month).

Keywords

• Rainfall Prediction
• Artifial Neural Network
• Backpropagation