EMPIRICAL ORTHOGONAL FUNCTION (EOF) ANALYSIS BASED ON GOOGLE COLAB ON SEA SURFACE TEMPERATURE (SST) DATASET IN INDONESIAN WATERS

Abstract

Global Sea Surface Temperature (SST) data observed from yearly to yearly is limited in its use to determine spatial and temporal variations. The analysis was carried out on SST data in Indonesian waters for 252 months or for 21 years, starting from January 2000 to December 2020. The method used for analysis was Empirical Orthogonal Function (EOF) with the help of a statistical engine, Google Colab. The EOF method aims to reduce large data into several modes without eliminating the main information from the observed data. Analysis with this method resulted in the three largest principal components initialized with EOF1EOF2 and EOF3 modes. The EOF1 mode explains 56.8% of the total variation and is the dominant pattern representing almost all SST data in Indonesian waters. The EOF2 mode represents 24.5% of the total variation. The EOF3 modes each account for 13.4% of the total variation. Each EOF mode contains coefficients containing variables in the form of grid data and eigenvectors. Grid data describe geographic locations and eigenvectors describe spatial dimensions. The effectiveness of the three resulting EOF modes is kept close to the original data. Mapping of SST in the Indonesian Territory for 20 years has been carried out in this research, this study describes the seasonal visualization of SST data in Indonesian waters using Google Colab. This visualization shows the comparison of the distribution of sea surface temperature in the Indonesian waters throughout the year with seasonal patterns.