Reconstruction of Satellite-Derived Sea Surface Temperature Data Based on an Improved DINEOF Algorithm

An improved data interpolating empirical orthogonal function (I-DINEOF) algorithm was proposed in this study. Compared with the ordinary DINEOF algorithm, in the I-DINEOF algorithm, the existing data are not necessary to be selected for cross-validation and the initial matrix is directly used for reconstruction. Instead of using single EOF to reconstruct the whole spatio-temporal matrix, the initial matrix is divided into several subareas and each subarea is reconstructed by the most suitable EOF. To validate the accuracy of the I-DINEOF algorithm, a real sea surface temperature (SST) data set and three synthetic data sets with different missing data percentage are reconstructed by using the DINEOF and I-DINEOF algorithms. Four parameters (Pearson correlation coefficient, signal-to-noise ratio, root-mean-square error, and mean absolute difference) are used as a measure of reconstructed accuracy. Compared with the DINEOF algorithm, the I-DINEOF algorithm is less affected by the missing data and can significantly enhance the accuracy of reconstruction.