ANALISIS KLIMATOLOGI INDEKS OSILASI SELATAN (SOI) UNTUK PENDUGAAN MUSIM TIGA-BULAN KE DEPAN MENGGUNAKAN REGRESI LINIER: PENDUGAAN SOI MUSIM JFM TAHUN 2002
Main Article Content
Abstract
Telah dilakukan analisis terhadap data klimatologi SOI untuk periode tahun 1900 - 2000 guna mengetahui tingkat peramalannya, serta peristiwa besar yang diakibatkannya . 1200 data bulanan SOI ini di stratifikasi menjadi data musim (season) dua-bulanan dan tiga-bulanan. Hasil analisis auotokorelasi menunjukkan bahwa dua deret musiman ini berkaitan erat, dengan korelasi 0,71. Dari hasil ini, dilakukan analisis regresi linier untuk memformulasikan model training, dan kemudian dilakukan validasi. Validasi silang tiga lipatan (three fold cross validation) menunjukkan bahwa model M3 = 1.027 M2 + - 1.12 merupakan model training yang menunjukkan kinerja paling baik, berdasarkan dengan tingkat korelasi antara SOI-musim hasil dugaan (predicted) dengan SOI -musim hasil pengamatan yang memiliki tingkat korelasi sebesar 0,9. Dengan hasil ini maka model training terpilih digunakan untuk melakukan prediksi SOI-musim tiga bulanan ke depan. Dengan data musim ND tahun 2001, hasil model menunjukkan bahwa musim JFM tahun 2002 adalah normal.
Climatology of Southern Oscillation Index (SOI) within period of 1900 - 2000 was out to analyzed to find out its predictability, and possibility impacts might follow. Total of 1200 monthly SOI data was stratified into new series , namely 2-month SOI -season and 3- month SOI-seson. The result of auto -correlation analysis indicates that this two series have strong lag-correlation ie 0,71. Base on this result , linear regression analysis was applied to formulate a model training, and then to validate its model. Three-fold Cross Validation of model training indicates that model training M3 = 1.027 M2 - 1.12 has better performance, because correlation between SOI -season prediction and SOI real data, is 0.9. Therefore the model may be used in forecasting activity to predict three SOI- season ahead. Based on SOI season ND 2001 the model says that JFM SOI season is normal.
Climatology of Southern Oscillation Index (SOI) within period of 1900 - 2000 was out to analyzed to find out its predictability, and possibility impacts might follow. Total of 1200 monthly SOI data was stratified into new series , namely 2-month SOI -season and 3- month SOI-seson. The result of auto -correlation analysis indicates that this two series have strong lag-correlation ie 0,71. Base on this result , linear regression analysis was applied to formulate a model training, and then to validate its model. Three-fold Cross Validation of model training indicates that model training M3 = 1.027 M2 - 1.12 has better performance, because correlation between SOI -season prediction and SOI real data, is 0.9. Therefore the model may be used in forecasting activity to predict three SOI- season ahead. Based on SOI season ND 2001 the model says that JFM SOI season is normal.
Article Details
How to Cite
Untung Haryanto. (2002). ANALISIS KLIMATOLOGI INDEKS OSILASI SELATAN (SOI) UNTUK PENDUGAAN MUSIM TIGA-BULAN KE DEPAN MENGGUNAKAN REGRESI LINIER: PENDUGAAN SOI MUSIM JFM TAHUN 2002. Jurnal Sains & Teknologi Modifikasi Cuaca, 3(1), 17–21. https://doi.org/10.29122/jstmc.v3i1.2155
Section
Articles