Reconstruction of missing data in orbital images of sea surface temperature in the southwest region of the Atlantic Ocean

Abstract

An empirical orthogonal function-based technique was used to reconstruct missing data of Sea Surface Temperature (SST) images. Sea surface temperature data was used to assess the benefits of a univariate reconstruction, and the combination of both SST and Maps of Sea Level Anomalies (MSLA) data was used in a multivariate approach. In the multivariate approach the effect of addition of MSLA data in the reconstruction of Brazil Current (BC) features was assessed. Results were satisfactory in both approaches. The combination of SST plus MSLA significantly improves results obtained by reconstruction; statistical analyses found an improvement of ~ 5%. All experiments correctly represent the mesoscale features off the southwest South Atlantic.

Keywords

• Reconstrução de Imagens. Imagens de Satélite. Corrente do Brasil. Função Ortogonal Empírica.