Implication of empirical orthogonal function analysis to objectively analyzed sea surface temperature data of Bay of Bengal

摘要

In this study, the Empirical Orthogonal Function (EOF) analysis was accomplished on a long term monthly Sea Surface Temperature (SST) data over the Bay of Bengal (BOB) through the Singular Value Decomposition (SVD) method. Monthly temperature data was obtained from the assimilation of recent ARGO observations into the Levitus Climatology using objective analysis technique. The aim of this study is divided into two categories : the intrinsic valuable information obtained through EOF and reconstruction of the whole data from the dominant modes. Computational accuracy was calculated through skill analysis. Comparison between the original and reconstructed dataset revealed that EOF analysis extracts the small intrinsic signals and acts like a smoothing technique filtering out the unwanted signals. Reconstruction with five dominant modes is sufficient to express the basin dynamics for the BOB. Skill analysis revealed accuracy of reconstruction with respect to original SST in the range of 98-99％ at different spatial locations.