Download PDFOpen PDF in browser

Rain Prediction Using Convolutional Neural Network (CNN) Method Based on Digital Image

EasyChair Preprint 6669, version 2

Versions: 12history
15 pagesDate: September 26, 2021

Abstract

Hydrometeorological disasters such as floods and landslides are natural disasters caused by heavy rain. These natural disasters often occur in Indonesia, not only causing material losses, but natural disasters also often take lives. To reduce the impact of natural disasters, it is necessary to predict rain which is one of the factors in natural disasters and any other needs. Rain prediction was developed using numerical models cannot predict rain accurately. Because of this reason, the rain prediction system was developed using the Convolution Neural Network (CNN) method in this research. One thousand of cloud images from Garut sky were used for the training process. It consists of two categories, cloudy images and rain images, to build predictive models. The simulation process is carried out by inputting a cloud image through several processes such as preprocessing, feature extraction, and learning process, so this system can predict the rain in the next hour. The accuracy of this system can reach up to 98% obtained from the results of tests carried out such as 80:20 data partition, 0.001 learning rate, and 50 epoch. The model that has been built can strengthen the existing rain models and provide more accurate information about the occurrence of hydrometeorological disasters.

Keyphrases: Cloud Image, Convolutional Neural Network (CNN), image processing, rain prediction.

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@booklet{EasyChair:6669,
  author    = {Alya Syifa Ihsani and Anggunmeka Luhur Prasasti and Wendi Harjupa and Umar Ali Ahmad and Reza Rendian Septiawan},
  title     = {Rain Prediction Using Convolutional Neural Network (CNN) Method Based on Digital Image},
  howpublished = {EasyChair Preprint 6669},
  year      = {EasyChair, 2021}}
Download PDFOpen PDF in browser