PREDICTION MODEL DEVELOPMENT AND ACCURACY IMPROVEMENT OF MONTHLY RAINFALL IN INDONESIAN REGION BASED ON TROPICAL RAINFALL MEASURING MISSION (TRMM), COSMIC RAYS AND SUNSPOT NUMBER DATA USING ARTIFICIAL NEURAL NETWORK

• Main Author: TEJO NUGROHO ( NIM: 32408301 ); Pembimbing : Prof. Safwan Hadi, Ph. D, JALU
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/18747
• Institution: Institut Teknologi Bandung

Improvement of monthly rainfall prediction accuracy in Indonesian region based on Tropical Rainfall Measuring Mission (TRMM) data using artificial neural network (ANN) model has been investigated. The external factors, solar activity and cosmic rays were involved as the network input.Energetic particles from solar flare and cosmic rays affect global electrical properties in the atmosphere. In troposphere, these particles play a role in the formation of aerosol, the important factor in cloud condensation nuclei formation which controls rainfall.The purpose of this study is to improve the accuracy of rainfall prediction in Indonesian region based on TRMM data by involving solar activity and cosmic rays as the driven factors. Time series data that used as network input were TRMM, sunspot numbers (SSN) and cosmic rays (CR).The fuzzy c-means clustering and wavelet methods were implemented to identify solar activity factor on rainfall in Indonesian region based on Outgoing Longwave Radiation (OLR) data. Solar activity signal is characterized by the appearance of a period of about 9 to 13 years at the OLR time series data that analyzed.Improved accuracy of rainfall prediction is represented by the increase of statistical correlation coefficient value (R) as an output network with two inputs (TRMM+SSN or TRMM+CR) compared with satellite TRMM data. As reference it was used the observed rainfall data from rain gauge stations (OBS).The monthly rainfall prediction accuracy increase up to 7,7% for Jakarta region,25,7% for Pontianak region and 31,4% for Jayapura region. For Kupang,Padang,Manado, Makassar and Lampung region the enhancement of it accuracy are 6,8%,34,8%, 36,3%, 5,4% and 35,2% respectively.Monthly rainfall prediction accuracy during April-October from 1998 to 2010 in the Jakarta region reaches 97%. Meanwhile for Pontianak and Jayapura region are 98%. During October-April in the same years, the monthly rainfall prediction accuracy in Jakarta, Pontianak and Jayapura region are 96%, 97% dan 97% respectively.This research indicates that solar activity and cosmic rays are important factors that should be taken into considerationin an attempt to increase the accuracy of rainfall prediction. The finding of this research can be applied by institutions or government agencies, such as the LAPAN and BMKG for rainfall prediction.